at Namespace Reference

Distributions kernel adapted from THRandom.cpp The kernels try to follow std::random distributions signature For instance: in ATen auto gen = at::detail::createCPUGenerator(); at::uniform_real_distribution<double> uniform(0, 1); auto sample = uniform(gen.get());. More...

Namespaces
namespace	$
namespace	autocast
namespace	cpp_custom_type_hack
namespace	cpu
namespace	cuda
namespace	detail
namespace	impl
namespace	indexing
namespace	internal
namespace	meta
namespace	metal
namespace	namedinference
namespace	native
namespace	Reduction
namespace	sequence_number
namespace	sparse
namespace	tracer
namespace	transformation
namespace	vec256
namespace	vitals
namespace	vml
namespace	vulkan

Classes
struct	AccumulateType
struct	AccumulateType< BFloat16, false >
struct	AccumulateType< bool, true >
struct	AccumulateType< c10::complex< double >, false >
struct	AccumulateType< c10::complex< double >, true >
struct	AccumulateType< c10::complex< float >, false >
struct	AccumulateType< c10::complex< float >, true >
struct	AccumulateType< char, false >
struct	AccumulateType< char, true >
struct	AccumulateType< double, false >
struct	AccumulateType< double, true >
struct	AccumulateType< float, false >
struct	AccumulateType< float, true >
struct	AccumulateType< Half, true >
struct	AccumulateType< int16_t, false >
struct	AccumulateType< int16_t, true >
struct	AccumulateType< int32_t, false >
struct	AccumulateType< int32_t, true >
struct	AccumulateType< int64_t, false >
struct	AccumulateType< int64_t, true >
struct	AccumulateType< int8_t, false >
struct	AccumulateType< int8_t, true >
struct	AccumulateType< uint8_t, false >
struct	AccumulateType< uint8_t, true >
struct	AffineQuantizer
	AffineQuantizer uses affine transformation to do quantization. More...
struct	ATenDLMTensor
struct	AutoGradMode
struct	AutoNonVariableTypeMode
class	BaseContext
	Virtual interface for the Context class in Caffe2. More...
struct	BatchDim
struct	BatchedTensorImpl
struct	BroadcastingVmapTransform
class	CallbackManager
class	Context
struct	CPUGeneratorImpl
struct	CUDAHooksArgs
struct	CUDAHooksInterface
struct	DefaultPtrTraits
class	DeprecatedTypeProperties
struct	DeprecatedTypePropertiesDeleter
class	DeprecatedTypePropertiesRegistry
struct	DimCounter
struct	Dimname
class	DisableRecordFunctionGuard
struct	DistAccumType
struct	DistAccumType< BFloat16 >
struct	DistAccumType< double >
struct	DistAccumType< float >
struct	DistAccumType< Half >
struct	DynamicLibrary
struct	FormatGuard
class	GenericPackedTensorAccessor
class	GenericPackedTensorAccessor< T, 1, PtrTraits, index_t >
class	GenericPackedTensorAccessorBase
struct	GradMode
struct	HIPHooksArgs
struct	HIPHooksInterface
struct	IterArgs
class	MatrixRef
	MatrixRef - Like an ArrayRef, but with an extra recorded strides so that we can easily view it as a multidimensional array. More...
struct	MetalImpl
struct	MetalTensorImpl
struct	MultiBatchVmapTransform
struct	NamedTensorMeta
struct	NamesMode
struct	NoGradGuard
struct	NoNamesGuard
struct	NonUniformQuantizer
	NonUniformQuantizer is the parent class for all non-uniform quantizers. More...
struct	NoTF32Guard
struct	ObserverContext
struct	OpaqueTensorImpl
struct	OperandInfo
struct	PerChannelAffineFloatQParamsQuantizer
	PerChannelAffineFloatQParamsQuantizer is the same as PerChannelAffineQuantizer except that it expects both scale and zero point to be floating point values. More...
struct	PerChannelAffineQuantizer
	PerChannelAffineQuantizer is the same as PerTensorAffineQuantizer except that we have an independent scale and zero_point parameter for each channel. More...
struct	PerTensorAffineQuantizer
	PerTensorAffineQuantizer stores a scale and a zero_point, which is used for all the values in the Tensor. More...
class	PTThreadPool
struct	QTensorImpl
	QTensorImpl is a TensorImpl for Quantized Tensors, it stores Quantizer which specifies the quantization scheme and parameters, for more information please see ATen/quantized/Quantizer.h. More...
struct	Quantizer
	Quantizer is the class for storing all the information that's necessary to perform quantize and dequantize operation. More...
struct	Range
struct	RecordFunction
class	RecordFunctionCallback
	RecordFunctionCallback represents a pair of callbacks to be used with RecordFunction, members: start, end - the callbacks to run when entering and exiting the scope; optionally, the start callback may return an ObserverContext which will be passed to the end callback, use appropriate constructor accordingly. More...
class	RecordFunctionGuard
struct	RecordFunctionTLS
struct	SparseTensorImpl
struct	SplitUntil32Bit
	A container-like struct that acts as if it contains splits of a TensorIterator that can use 32-bit indexing. More...
struct	strided_tensor_iter
struct	strided_tensor_iter_fixed
struct	StringView
class	Tensor
class	TensorAccessor
class	TensorAccessor< T, 1, PtrTraits, index_t >
class	TensorAccessorBase
struct	TensorArg
struct	TensorGeometry
struct	TensorGeometryArg
struct	TensorIterator
struct	TensorIteratorBase
class	TensorIteratorConfig
class	ThreadLocalState
class	ThreadLocalStateGuard
struct	UniformQuantizer
	UniformQuantizer is the parent class for all uniform quantizers. More...
struct	VmapPhysicalToLogicalMap
struct	VmapPhysicalView
struct	VulkanOpaqueTensorImpl

Typedefs
template<typename T , bool is_cuda>
using	acc_type = typename AccumulateType< T, is_cuda >::type
using	BatchDims = SmallVector< BatchDim, kBatchDimsStackSize >
using	BatchDimsRef = ArrayRef< BatchDim >
using	DimnameList = c10::ArrayRef< Dimname >
using	DimVector = SmallVector< int64_t, kDimVectorStaticSize >
	A container for sizes or strides. More...
using	ConstQuantizerPtr = const c10::intrusive_ptr< Quantizer > &
using	QuantizerPtr = c10::intrusive_ptr< Quantizer >
template<typename T , size_t N, template< typename U > class PtrTraits = DefaultPtrTraits>
using	PackedTensorAccessor32 = GenericPackedTensorAccessor< T, N, PtrTraits, int32_t >
template<typename T , size_t N, template< typename U > class PtrTraits = DefaultPtrTraits>
using	PackedTensorAccessor64 = GenericPackedTensorAccessor< T, N, PtrTraits, int64_t >
template<typename T >
using	dist_acctype = typename DistAccumType< T >::type
using	NameVector = SmallVector< Dimname, kDimVectorStaticSize >
using	loop2d_t = TensorIteratorBase::loop2d_t
typedef c10::SmallVector< uint64_t, kSoftLimitCallbacks >	CallbackHandles
typedef std::vector< std::unique_ptr< ObserverContext > >	ObserverContextList
typedef uint64_t	RecordFunctionHandle
typedef uint64_t	CallbackHandle
typedef std::vector< std::pair< RecordFunctionCallback, CallbackHandle > >	RecordFunctionCallbacks
using	TensorList = ArrayRef< Tensor >
using	Stream = c10::Stream
using	DimMask = TensorIteratorBase::DimMask
using	PtrVector = TensorIteratorBase::PtrVector
using	loop_t = TensorIteratorBase::loop_t
using	StrideVector = TensorIteratorBase::StrideVector
using	CheckedFrom = const char *
using	VmapPhysicalViewVec = SmallVector< VmapPhysicalView, kVmapTransformStaticInputSize >
using	VmapDimVector = SmallVector< int64_t, kVmapStaticDimVecSize >
using	DataType = caffe2::TypeIdentifier

Enumerations
enum class	NameType : uint8_t { BASIC , WILDCARD }
enum class	MemOverlap { NO , YES , TOO_HARD }
enum class	MemOverlapStatus { FULL , PARTIAL , NO , TOO_HARD }
enum class	FastSetupType : uint8_t { NONE , CONTIGUOUS , CHANNELS_LAST , NON_OVERLAPPING_DENSE }

Functions
static SmallVector< indexing::TensorIndex, kVmapStaticDimVecSize >	computeIndex (int64_t linear_idx, IntArrayRef sizes)
static bool	areAllReturnsTensors (const FunctionSchema &schema)
static bool	areAnyArgumentsTensorList (const FunctionSchema &schema)
static bool	isInplaceOp (const c10::FunctionSchema &schema)
static void	warnFallback (const c10::FunctionSchema &schema, bool is_inplace)
void	batchedTensorInplaceForLoopFallback (const c10::OperatorHandle &op, torch::jit::Stack *stack)
static Tensor	safeStack (TensorList tensors)
void	batchedTensorForLoopFallback (const c10::OperatorHandle &op, torch::jit::Stack *stack)
Tensor	makeBatched (const Tensor &tensor, BatchDims bdims)
Tensor	addBatchDim (const Tensor &tensor, int64_t level, int64_t dim)
bool	inplaceIsVmapCompatible (const Tensor &self, const Tensor &other)
bool	isBatchedTensor (const Tensor &tensor)
BatchedTensorImpl *	unsafeGetBatchedImpl (Tensor tensor)
BatchedTensorImpl *	maybeGetBatchedImpl (Tensor tensor)
std::bitset< kVmapMaxTensorDims >	createBatchDimBitset (BatchDimsRef bdims)
std::bitset< kVmapNumLevels >	createVmapLevelsBitset (BatchDimsRef bdims)
std::ostream &	operator<< (std::ostream &out, const BatchDim &bdim)
static bool	is_allowed_dim_on_scalar_tensor (int64_t dim)
Tensor	sum_batching_rule (const Tensor &self, IntArrayRef dims, bool keepdim, optional< ScalarType > dtype)
bool	isPhysicalScalarTensor (const Tensor &logical_tensor)
template<typename F , F Func, typename... ExtraArgs>
Tensor	binary_pointwise_batching_rule (const Tensor &self, const Tensor &other, ExtraArgs... args)
Tensor	expand_batching_rule (const Tensor &self, IntArrayRef size, bool implicit)
std::vector< Tensor >	chunk_batching_rule (const Tensor &self, int64_t chunks, int64_t dim)
Tensor	clamp_batching_rule (const Tensor &self, optional< Scalar > min, optional< Scalar > max)
Tensor	clamp_min_batching_rule (const Tensor &self, Scalar min)
Tensor	clamp_max_batching_rule (const Tensor &self, Scalar max)
std::vector< Tensor >	tensor_split_sections_batching_rule (const Tensor &self, int64_t sections, int64_t dim)
std::vector< Tensor >	tensor_split_indices_batching_rule (const Tensor &self, IntArrayRef indices, int64_t dim)
Tensor	unsqueeze_batching_rule (const Tensor &self, int64_t dim)
Tensor &	fill_inplace_scalar_batching_rule (Tensor &self, Scalar value)
Tensor &	fill_inplace_tensor_batching_rule (Tensor &self, const Tensor &value)
Tensor &	zero_inplace_batching_rule (Tensor &self)
Tensor	squeeze_batching_rule (const Tensor &self)
Tensor	squeeze_dim_batching_rule (const Tensor &self, int64_t dim)
Tensor	trace_batching_rule (const Tensor &self)
Tensor	trace_backward_batching_rule (const Tensor &grad, IntArrayRef input_sizes)
Tensor	transpose_int_batching_rule (const Tensor &self, int64_t dim0, int64_t dim1)
Tensor	permute_batching_rule (const Tensor &self, IntArrayRef dims)
Tensor	select_batching_rule (const Tensor &self, int64_t dim, int64_t index)
static int64_t	getGradInputPhysicalDim (int64_t dim, IntArrayRef input_sizes, int64_t num_batch_dims)
Tensor	select_backward_batching_rule (const Tensor &grad, IntArrayRef input_sizes, int64_t dim, int64_t index)
Tensor	slice_batching_rule (const Tensor &self, int64_t dim, c10::optional< int64_t > start, c10::optional< int64_t > end, int64_t step)
Tensor	slice_backward_batching_rule (const Tensor &grad, IntArrayRef input_sizes, int64_t dim, int64_t start, int64_t end, int64_t step)
Tensor	diagonal_batching_rule (const Tensor &self, int64_t offset, int64_t dim1, int64_t dim2)
Tensor	diagonal_backward_batching_rule (const Tensor &grad, IntArrayRef input_sizes, int64_t offset, int64_t dim1, int64_t dim2)
Tensor	movedim_batching_rule (const Tensor &self, IntArrayRef source, IntArrayRef destination)
Tensor	reshape_batching_rule (const Tensor &self, IntArrayRef shape)
std::vector< Tensor >	split_batching_rule (const Tensor &self, int64_t split_size, int64_t dim)
std::vector< Tensor >	split_with_sizes_batching_rule (const Tensor &self, IntArrayRef split_sizes, int64_t dim)
std::vector< Tensor >	unbind_batching_rule (const Tensor &self, int64_t dim)
Tensor	unfold_batching_rule (const Tensor &self, int64_t dim, int64_t size, int64_t step)
Tensor	contiguous_batching_rule (const Tensor &self, MemoryFormat memory_format)
Tensor	view_batching_rule (const Tensor &self, IntArrayRef size)
Tensor	view_as_complex_batching_rule (const Tensor &self)
static void	checkBatchDimsAtFrontInLayout (IntArrayRef physical_strides, int64_t num_batch_dims)
static optional< int64_t >	maximum_indexable_location (IntArrayRef sizes, IntArrayRef strides, int64_t storage_offset)
static void	checkBasicAsStridedValidForSlice (const Tensor &physical_tensor, int64_t num_batch_dims, IntArrayRef sizes, IntArrayRef strides, optional< int64_t > maybe_storage_offset)
Tensor	as_strided_batching_rule (const Tensor &tensor, IntArrayRef sizes, IntArrayRef strides, optional< int64_t > storage_offset)
template<typename F , F Func, typename... ExtraArgs>
Tensor	unwrap_and_call (const Tensor &input, ExtraArgs... args)
template<typename F , F Func, typename... ExtraArgs>
Tensor	unwrap_and_call_method (const Tensor &input, ExtraArgs... extra_args)
Tensor	pow_scalar_Tensor_batching_rule (Scalar other, const Tensor &self)
Tensor	clone_batching_rule (const Tensor &self, optional< MemoryFormat > memory_format)
Tensor	mv_batching_rule (const Tensor &self, const Tensor &other)
Tensor	dot_batching_rule (const Tensor &self, const Tensor &other)
Tensor	bmm_batching_rule (const Tensor &self, const Tensor &other)
Tensor	mm_batching_rule (const Tensor &self, const Tensor &other)
Tensor	cat_batching_rule (TensorList tensors, int64_t dim)
Tensor	stack_batching_rule (TensorList tensors, int64_t dim)
Tensor	to_dtype_layout_batching_rule (const Tensor &self, optional< ScalarType > dtype, optional< Layout > layout, optional< Device > device, optional< bool > pin_memory, bool non_blocking, bool copy, optional< MemoryFormat > memory_format)
Tensor	new_zeros_batching_rule (const Tensor &self, IntArrayRef size, optional< ScalarType > dtype, optional< Layout > layout, optional< Device > device, optional< bool > pin_memory)
Tensor	new_empty_batching_rule (const Tensor &self, IntArrayRef size, c10::optional< ScalarType > dtype, c10::optional< Layout > layout, c10::optional< Device > device, c10::optional< bool > pin_memory)
Tensor	new_empty_strided_batching_rule (const Tensor &self, IntArrayRef size, IntArrayRef stride, optional< ScalarType > dtype, optional< Layout > layout, optional< Device > device, optional< bool > pin_memory)
template<typename F , F Func>
Tensor	comparison_pointwise_batching_rule (const Tensor &self, const Tensor &other)
	TORCH_LIBRARY_IMPL (_, Batched, m)
	TORCH_LIBRARY_IMPL (aten, Batched, m)
Context &	globalContext ()
Allocator *	getCPUAllocator ()
static void	init ()
static DeprecatedTypeProperties &	getDeprecatedTypeProperties (Backend p, ScalarType s)
static DeprecatedTypeProperties &	CPU (ScalarType s)
static DeprecatedTypeProperties &	CUDA (ScalarType s)
static DeprecatedTypeProperties &	HIP (ScalarType s)
static bool	hasCUDA ()
static bool	hasHIP ()
static bool	hasXLA ()
static size_t	getNumGPUs ()
static bool	hasOpenMP ()
static bool	hasMKL ()
static bool	hasLAPACK ()
static bool	hasMAGMA ()
static bool	hasMKLDNN ()
static void	manual_seed (uint64_t seed)
TORCH_API bool	is_custom_op (const c10::OperatorName &opName)
DeprecatedTypePropertiesRegistry &	globalDeprecatedTypePropertiesRegistry ()
std::ostream &	operator<< (std::ostream &out, const Dimname &dimname)
static void	check_valid_identifier (const std::string &name)
bool	operator== (const Dimname &lhs, const Dimname &rhs)
bool	operator!= (const Dimname &lhs, const Dimname &rhs)
std::ios_base &	defaultfloat (std::ios_base &__base)
std::ostream &	operator<< (std::ostream &out, const DeprecatedTypeProperties &t)
static std::tuple< double, int64_t >	__printFormat (std::ostream &stream, const Tensor &self)
static void	__printIndent (std::ostream &stream, int64_t indent)
static void	printScale (std::ostream &stream, double scale)
static void	__printMatrix (std::ostream &stream, const Tensor &self, int64_t linesize, int64_t indent)
void	__printTensor (std::ostream &stream, Tensor &self, int64_t linesize)
std::ostream &	print (std::ostream &stream, const Tensor &tensor_, int64_t linesize)
static std::ostream &	operator<< (std::ostream &out, const Tensor &t)
static void	print (const Tensor &t, int64_t linesize=80)
static std::ostream &	operator<< (std::ostream &out, Scalar s)
TORCH_API void	launch (std::function< void()> func)
Tensor &	internal_set_names_inplace (Tensor &tensor, optional< DimnameList > names)
Tensor &	internal_set_names_inplace (Tensor &tensor, std::vector< Dimname > &&names, bool validate_names)
DimnameList	default_names (size_t len)
static void	check_unique_names (DimnameList names)
void	check_names_valid_for (const Tensor &tensor, DimnameList names)
void	check_names_valid_for (size_t tensor_dim, DimnameList names)
std::ostream &	operator<< (std::ostream &out, const Range &range)
Tensor	unsafeTensorFromTH (void *th_pointer, bool retain)
Storage	unsafeStorageFromTH (void *th_pointer, bool retain)
template<typename T >
std::pair< int64_t, int64_t >	collapse_dims (T *sizes, T *strides, int64_t dims, const int excludeDim=-1)
Tensor	sort_strides (Tensor &tensor_)
bool	_all_equal_numel (at::ArrayRef< Tensor > tensors)
std::string	_all_equal_numel_error (at::ArrayRef< Tensor > tensors)
bool	_apply_preamble (ArrayRef< Tensor > tensors)
int64_t	_max_dim_tensors (ArrayRef< Tensor > tensors)
void	iterate (int64_t size)
template<typename Arg , typename... Args>
void	iterate (int64_t size, Arg &iter, Args &... iter_tail)
bool	iterate_continue ()
template<typename Arg , typename... Args>
bool	iterate_continue (Arg &iter, Args &... iter_tail)
int64_t	max_iterate_size ()
template<typename Arg , typename... Args>
int64_t	max_iterate_size (Arg &iter, Args &... iter_tail)
void	iterate_overflow ()
template<typename Arg , typename... Args>
void	iterate_overflow (Arg &iter, Args &... iter_tail)
void	forward (int64_t offset)
template<typename Arg , typename... Args>
void	forward (int64_t offset, Arg &iter, Args &... iter_tail)
int64_t	max_dim ()
template<typename Arg , typename... Args>
int64_t	max_dim (Arg &iter, Args &... iter_tail)
void	apply_op ()
template<typename Op , typename... Args>
void	apply_op (int64_t numel, int64_t offset, const Op &op, Args... iters)
template<typename scalar1 , typename scalar2 , typename Op >
void	CPU_tensor_apply2 (Tensor tensor1, Tensor tensor2, const Op op)
template<typename scalar1 , typename scalar2 , typename scalar3 , typename Op >
void	CPU_tensor_apply3 (Tensor tensor1, Tensor tensor2, Tensor tensor3, const Op op)
template<typename scalar1 , typename scalar2 , typename scalar3 , typename scalar4 , typename Op >
void	CPU_tensor_apply4 (Tensor tensor1, Tensor tensor2, Tensor tensor3, Tensor tensor4, const Op op)
static	cpu_fixed_malloc (void *, ptrdiff_t)
static	cpu_fixed_realloc (void *, void *, ptrdiff_t)
static	cpu_fixed_free (void *state, void *allocation)
	C10_DECLARE_REGISTRY (CUDAHooksRegistry, CUDAHooksInterface, CUDAHooksArgs)
	C10_DECLARE_REGISTRY (HIPHooksRegistry, HIPHooksInterface, HIPHooksArgs)
optional< Device >	device_of (const Tensor &t)
	Return the Device of a Tensor, if the Tensor is defined. More...
optional< Device >	device_of (TensorList t)
	Return the Device of a TensorList, if the list is non-empty and the first Tensor is defined. More...
constexpr bool	should_include_kernel_dtype (const char *kernel_tag_str, at::ScalarType scalar_type)
	The method should_include_kernel_dtype() returns true/false based on whether the switching code for a specific dtype should be included based on build time constants generated from tracing model execution. More...
DLDataType	getDLDataType (const Tensor &t)
DLContext	getDLContext (const Tensor &tensor, const int64_t &device_id)
static Device	getATenDevice (const DLContext &ctx)
ScalarType	toScalarType (const DLDataType &dtype)
void	deleter (DLManagedTensor *arg)
DLManagedTensor *	toDLPack (const Tensor &src)
Tensor	fromDLPack (const DLManagedTensor *src)
static void *	checkDL (void *x)
std::vector< int64_t >	infer_size (IntArrayRef a, IntArrayRef b)
std::tuple< std::vector< int64_t >, std::vector< int64_t > >	inferExpandGeometry (IntArrayRef tensor_sizes, IntArrayRef tensor_strides, IntArrayRef sizes)
std::vector< int64_t >	infer_dense_strides (IntArrayRef tensor_sizes, IntArrayRef tensor_strides)
bool	are_expandable (IntArrayRef shape1, IntArrayRef shape2)
void	check_defined (std::initializer_list< std::reference_wrapper< const Tensor > > tensors, const char *api_name)
std::tuple< Tensor >	expand_inplace (const Tensor &tensor, const Tensor &to_expand)
std::tuple< Tensor >	expand_inplace (const Tensor &tensor, const Tensor &to_expand, const char *api_name)
std::tuple< Tensor, Tensor >	expand_inplace (const Tensor &tensor, const Tensor &to_expand1, const Tensor &to_expand2)
std::tuple< Tensor, Tensor >	expand_inplace (const Tensor &tensor, const Tensor &to_expand1, const Tensor &to_expand2, const char *api_name)
std::tuple< Tensor, Tensor >	expand_outplace (const Tensor &to_expand1, const Tensor &to_expand2)
std::tuple< Tensor, Tensor >	expand_outplace (const Tensor &to_expand1, const Tensor &to_expand2, const char *api_name)
std::tuple< Tensor, Tensor, Tensor >	expand_outplace (const Tensor &to_expand1, const Tensor &to_expand2, const Tensor &to_expand3)
std::tuple< Tensor, Tensor, Tensor >	expand_outplace (const Tensor &to_expand1, const Tensor &to_expand2, const Tensor &to_expand3, const char *api_name)
std::tuple< Tensor >	expand_size (const Tensor &to_expand, IntArrayRef sizes)
std::tuple< Tensor >	expand_size (const Tensor &to_expand, IntArrayRef sizes, const char *api_name)
std::vector< Tensor >	expand_outplace (TensorList to_expand)
static Tensor	sum_to (Tensor tensor, const IntArrayRef shape)
static bool	is_expandable_to (IntArrayRef shape, IntArrayRef desired)
std::vector< int64_t >	infer_size (IntArrayRef shape, int64_t numel)
TensorOptions	initialTensorOptions ()
MemOverlap	has_internal_overlap (const Tensor &tensor)
MemOverlap	has_internal_overlap (TensorImpl *t)
void	assert_no_internal_overlap (const Tensor &t)
void	assert_no_internal_overlap (TensorImpl *t)
MemOverlapStatus	get_overlap_status (const Tensor &a, const Tensor &b)
MemOverlapStatus	get_overlap_status (TensorImpl *a, TensorImpl *b)
void	assert_no_partial_overlap (const Tensor &a, const Tensor &b)
void	assert_no_partial_overlap (TensorImpl *a, TensorImpl *b)
void	assert_no_overlap (const Tensor &a, const Tensor &b)
void	assert_no_overlap (TensorImpl *a, TensorImpl *b)
static std::string	toDimnameRepr (const Tensor &tensor)
int64_t	dimname_to_position (const Tensor &tensor, Dimname dim)
std::vector< int64_t >	dimnames_to_positions (const Tensor &tensor, DimnameList dims)
static void	report_positional_error (const Dimname &name, const Dimname &other_name, DimnameList names, DimnameList other_names, const char *action)
static void	check_for_misalignment (const Dimname &name, DimnameList names, DimnameList other_names, const char *action)
std::vector< Dimname >	unify_from_right (DimnameList names, DimnameList other_names, const char *action)
bool	has_names (TensorList tensors)
void	reportNYIDimnameOverload (const char *op_name)
static bool	use_two_pass_reduction (TensorIteratorBase &iter)
static void	two_pass_reduction (TensorIteratorBase &iter, loop2d_t loop)
static void	parallel_dim_reduction (TensorIteratorBase &iter, loop2d_t loop)
static int	find_split_dim (TensorIteratorBase &iter)
	Chooses a dimension over which to parallelize. More...
static std::tuple< int64_t, int64_t >	round_columns (TensorIteratorBase &iter, int dim, int multiple, int64_t begin, int64_t end)
template<typename T , typename std::enable_if< std::is_integral< T >::value, int >::type = 0>
C10_HOST_DEVICE bool	_isnan (T val)
template<typename T , typename std::enable_if< c10::is_complex< T >::value, int >::type = 0>
bool	_isnan (T val)
template<typename T , typename std::enable_if< std::is_same< T, at::BFloat16 >::value, int >::type = 0>
C10_HOST_DEVICE bool	_isnan (at::BFloat16 val)
template<typename T >
C10_HOST_DEVICE T	exp (T x)
template<>
C10_HOST_DEVICE double	exp< double > (double x)
template<typename T >
C10_HOST_DEVICE T	log (T x)
template<>
C10_HOST_DEVICE double	log< double > (double x)
template<typename T >
C10_HOST_DEVICE T	tan (T x)
template<>
C10_HOST_DEVICE double	tan< double > (double x)
int64_t	divup (int64_t x, int64_t y)
TORCH_API void	init_num_threads ()
TORCH_API void	set_num_threads (int)
TORCH_API int	get_num_threads ()
TORCH_API int	get_thread_num ()
TORCH_API bool	in_parallel_region ()
template<class F >
void	parallel_for (const int64_t begin, const int64_t end, const int64_t grain_size, const F &f)
template<class scalar_t , class F , class SF >
scalar_t	parallel_reduce (const int64_t begin, const int64_t end, const int64_t grain_size, const scalar_t ident, const F &f, const SF &sf)
TORCH_API std::string	get_parallel_info ()
TORCH_API void	set_num_interop_threads (int)
TORCH_API int	get_num_interop_threads ()
TORCH_API void	intraop_launch (std::function< void()> func)
TORCH_API std::shared_ptr< c10::ivalue::Future >	intraop_launch_future (std::function< void()> func)
TORCH_API int	intraop_default_num_threads ()
template<typename F0 , typename F1 >
void	intraop_invoke (const F0 &f0, const F1 &f1)
QuantizerPtr	make_per_tensor_affine_quantizer (double scale, int64_t zero_point, ScalarType scalar_type)
QuantizerPtr	make_per_channel_affine_quantizer (const Tensor &scales, const Tensor &zero_points, int64_t axis, ScalarType scalar_type)
QTensorImpl *	get_qtensorimpl (const Tensor &self)
int64_t	get_sub_byte_tensor_size (int64_t size_bytes, at::ScalarType t)
Tensor	new_qtensor (IntArrayRef sizes, const TensorOptions &options, QuantizerPtr quantizer)
C10_EXPORT void	set_quantizer_ (const Tensor &self, ConstQuantizerPtr quantizer)
const RecordFunctionTLS &	get_record_function_tls_ ()
void	set_record_function_tls_ (const RecordFunctionTLS &tls)
RecordFunctionCallbacks	_getTLSCallbacks ()
void	_setTLSCallbacks (const RecordFunctionCallbacks &callbacks)
bool	hasCallbacks ()
bool	hasGlobalCallbacks ()
	hasGlobalCallbacks returns whether there're global callbacks registered with pushGlobalCallback More...
bool	hasThreadLocalCallbacks ()
	hasThreadLocalCallbacks returns whether there're callbacks registered with addThreadLocalCallback More...
CallbackHandle	addThreadLocalCallback (RecordFunctionCallback cb)
	addThreadLocalCallback adds a thread local callback to run with RecordFunction, returns handle to use with removeThreadLocalCallback More...
CallbackHandle	addGlobalCallback (RecordFunctionCallback cb)
	addGlobalCallback adds a global callback to run with RecordFunction: More...
void	removeCallback (CallbackHandle handle)
	removeCallback removes a callback given the handle returned by addThreadLocalCallback or addGlobalCallback; More...
void	clearGlobalCallbacks ()
	clearGlobalCallbacks removes all global callbacks WARNING: not thread safe More...
void	clearThreadLocalCallbacks ()
	clearThreadLocalCallbacks removes all thread local callbacks More...
void	clearCallbacks ()
bool	isRecordFunctionEnabled ()
	isRecordFunctionEnabled returns whether RecordFunction is enabled thread locally More...
void	enableRecordFunction (bool enable=true)
	enableRecordFunction enables RecordFunction thread locally More...
void	bumpRecordAllFunctions ()
void	releaseRecordAllFunctions ()
bool	checkRecordAllFunctions ()
bool	shouldRunRecordFunction (bool *pre_sampled)
Tensor	var (const Tensor &self, int dim)
std::tuple< Tensor, Tensor >	var_mean (const Tensor &self, int dim)
Tensor	std (const Tensor &self, int dim)
std::tuple< Tensor, Tensor >	std_mean (const Tensor &self, int dim)
Tensor	from_blob (void *data, IntArrayRef sizes, IntArrayRef strides, const std::function< void(void *)> &deleter, const TensorOptions &options={}, const c10::optional< Device > target_device=c10::nullopt)
Tensor	from_blob (void *data, IntArrayRef sizes, const std::function< void(void *)> &deleter, const TensorOptions &options={})
Tensor	from_blob (void *data, IntArrayRef sizes, IntArrayRef strides, const TensorOptions &options={})
Tensor	from_blob (void *data, IntArrayRef sizes, const TensorOptions &options={})
int64_t	numel (const Tensor &tensor)
int64_t	size (const Tensor &tensor, int64_t dim)
int64_t	stride (const Tensor &tensor, int64_t dim)
	TORCH_LIBRARY (aten, m)
int64_t	get_device (Tensor self)
static DispatchKey	legacyExtractDispatchKey (const Tensor &t)
static Tensor	operator+ (const Tensor &x, const Tensor &y)
static Tensor	operator+ (const Tensor &x, Scalar y)
static Tensor	operator+ (Scalar x, const Tensor &y)
static Tensor	operator* (const Tensor &x, const Tensor &y)
static Tensor	operator* (const Tensor &x, Scalar y)
static Tensor	operator* (Scalar x, const Tensor &y)
static Tensor	operator- (const Tensor &x, const Tensor &y)
static Tensor	operator- (const Tensor &x, Scalar y)
static Tensor	operator- (Scalar x, const Tensor &y)
static Tensor	operator/ (const Tensor &x, const Tensor &y)
static Tensor	operator/ (const Tensor &x, Scalar y)
static Tensor	operator/ (Scalar x, const Tensor &y)
static Tensor	operator% (const Tensor &x, const Tensor &y)
static Tensor	operator% (const Tensor &x, Scalar y)
static Tensor	operator% (Scalar x, const Tensor &y)
static Tensor	operator& (const Tensor &x, const Tensor &y)
static Tensor	operator& (const Tensor &x, Scalar y)
static Tensor	operator& (Scalar x, const Tensor &y)
static Tensor	operator| (const Tensor &x, const Tensor &y)
static Tensor	operator| (const Tensor &x, Scalar y)
static Tensor	operator| (Scalar x, const Tensor &y)
static Tensor	operator^ (const Tensor &x, const Tensor &y)
static Tensor	operator^ (const Tensor &x, Scalar y)
static Tensor	operator^ (Scalar x, const Tensor &y)
static Tensor	operator< (const Tensor &x, const Tensor &y)
static Tensor	operator< (const Tensor &x, Scalar y)
static Tensor	operator< (Scalar x, const Tensor &y)
static Tensor	operator<= (const Tensor &x, const Tensor &y)
static Tensor	operator<= (const Tensor &x, Scalar y)
static Tensor	operator<= (Scalar x, const Tensor &y)
static Tensor	operator> (const Tensor &x, const Tensor &y)
static Tensor	operator> (const Tensor &x, Scalar y)
static Tensor	operator> (Scalar x, const Tensor &y)
static Tensor	operator>= (const Tensor &x, const Tensor &y)
static Tensor	operator>= (const Tensor &x, Scalar y)
static Tensor	operator>= (Scalar x, const Tensor &y)
static Tensor	operator== (const Tensor &x, const Tensor &y)
static Tensor	operator== (const Tensor &x, Scalar y)
static Tensor	operator== (Scalar x, const Tensor &y)
static Tensor	operator!= (const Tensor &x, const Tensor &y)
static Tensor	operator!= (const Tensor &x, Scalar y)
static Tensor	operator!= (Scalar x, const Tensor &y)
std::ostream &	operator<< (std::ostream &out, TensorGeometryArg t)
void	checkDim (CheckedFrom c, const Tensor &tensor, const char *name, int pos, int64_t dim)
void	checkDim (CheckedFrom c, const TensorGeometryArg &t, int64_t dim)
void	checkDimRange (CheckedFrom c, const TensorGeometryArg &t, int64_t dim_start, int64_t dim_end)
void	checkContiguous (CheckedFrom c, const TensorGeometryArg &t)
void	checkAllContiguous (CheckedFrom c, at::ArrayRef< TensorArg > ts)
void	checkSize (CheckedFrom c, const TensorGeometryArg &t, IntArrayRef sizes)
void	checkSize (CheckedFrom c, const TensorGeometryArg &t, int64_t dim, int64_t size)
void	checkAllSame (CheckedFrom c, ArrayRef< TensorArg > tensors, void(*fn)(CheckedFrom, const TensorArg &, const TensorArg &))
void	checkSameSize (CheckedFrom c, const TensorArg &t1, const TensorArg &t2)
void	checkAllSameSize (CheckedFrom c, ArrayRef< TensorArg > tensors)
void	checkNumel (CheckedFrom c, const TensorGeometryArg &t, int64_t numel)
void	checkSameNumel (CheckedFrom c, const TensorArg &t1, const TensorArg &t2)
void	checkAllSameNumel (CheckedFrom c, ArrayRef< TensorArg > tensors)
void	checkSameGPU (CheckedFrom c, const TensorArg &t1, const TensorArg &t2)
void	checkAllSameGPU (CheckedFrom c, ArrayRef< TensorArg > tensors)
void	checkSameType (CheckedFrom c, const TensorArg &t1, const TensorArg &t2)
void	checkScalarType (CheckedFrom c, const TensorArg &t, ScalarType ty)
void	checkScalarTypes (CheckedFrom c, const TensorArg &t, at::ArrayRef< ScalarType > l)
void	checkAllSameType (CheckedFrom c, ArrayRef< TensorArg > tensors)
void	checkSameDim (CheckedFrom c, const TensorGeometryArg &t1, const TensorGeometryArg &t2)
void	checkDefined (CheckedFrom c, const TensorArg &t)
void	checkAllDefined (CheckedFrom c, ArrayRef< TensorArg > ts)
void	checkBackend (CheckedFrom c, const Tensor &t, Backend backend)
void	checkBackend (CheckedFrom c, at::ArrayRef< Tensor > tensors, at::Backend backend)
void	checkDeviceType (CheckedFrom c, const Tensor &t, DeviceType device_type)
void	checkDeviceType (CheckedFrom c, at::ArrayRef< Tensor > tensors, at::DeviceType device_type)
void	checkLayout (CheckedFrom c, const Tensor &t, Layout layout)
void	checkLayout (CheckedFrom c, at::ArrayRef< Tensor > tensors, at::Layout layout)
void *	maybe_data_ptr (const Tensor &tensor)
void *	maybe_data_ptr (const TensorArg &tensor)
bool	geometry_is_contiguous (IntArrayRef sizes, IntArrayRef strides)
void	check_dim_size (const Tensor &tensor, int64_t dim, int64_t dim_size, int64_t size)
TORCH_API void	checkSameNumel (CheckedFrom c, const TensorGeometryArg &t1, const TensorGeometryArg &t2)
template<typename T >
std::function< T(void)>	wrapPropagateTLSState (std::function< T(void)> callback)
int	_crash_if_asan (int arg)
static TensorImpl *	checked_dense_tensor_unwrap (const Tensor &expr, const char *name, int pos, const char *api, bool allowNull, DeviceType device_type, ScalarType scalar_type)
static std::vector< TensorImpl * >	checked_dense_tensor_list_unwrap (ArrayRef< Tensor > tensors, const char *name, int pos, DeviceType device_type, ScalarType scalar_type)
template<size_t N>
std::array< int64_t, N >	check_intlist (ArrayRef< int64_t > list, const char *name, int pos)
int64_t	sum_intlist (ArrayRef< int64_t > list)
template<typename C , typename std::enable_if< std::is_integral< typename C::value_type >::value, int >::type = 0>
int64_t	prod_intlist (const C &container)
template<typename Iter , typename std::enable_if< std::is_integral< typename std::iterator_traits< Iter >::value_type >::value, int >::type = 0>
int64_t	prod_intlist (Iter begin, Iter end)
template<typename T >
static T *	check_generator (c10::optional< Generator > gen)
	Utility function to static cast input Generator* to the backend generator type (CPU/CUDAGeneratorImpl etc.) More...
template<typename T >
static T *	get_generator_or_default (const c10::optional< Generator > &gen, const Generator &default_gen)
	Utility function used in tensor implementations, which supplies the default generator to tensors, if an input generator is not supplied. More...
void	check_size_nonnegative (IntArrayRef size)
std::string	get_mkl_version ()
std::string	get_mkldnn_version ()
std::string	get_openmp_version ()
std::string	used_cpu_capability ()
std::string	show_config ()
	Returns a detailed string describing the configuration PyTorch. More...
std::string	get_cxx_flags ()
template<typename... Args>
Tensor	unsupportedRandomOp (Args... args)
template<typename... Args>
Tensor &	unsupportedRandomOp_ (Args... args)
	TORCH_LIBRARY_IMPL (_, VmapMode, m)
	TORCH_LIBRARY_IMPL (aten, VmapMode, m)
static bool	areBdimsAtFrontInOrder (BatchDimsRef bdims)
static Tensor	permuteBatchDimsToFront (BatchedTensorImpl *batched)
static BatchDims	computeFrontBatchDimsFromLevels (std::bitset< kVmapNumLevels > levels_bitset)
static std::pair< Tensor, std::bitset< kVmapNumLevels > >	getPhysicalTensorAndLevels (const Tensor &self)
static Tensor	alignBatchDimsAtFront (const Tensor &self, std::bitset< kVmapNumLevels > requested_levels, int64_t requested_example_dim)
static std::pair< std::bitset< kVmapNumLevels >, int64_t >	getLevelsAndLargestLogicalDim (TensorList logical_tensors)
static int64_t	maybe_wrap_dim (int64_t dim, int64_t dim_post_expr, bool wrap_scalar=true)
static int64_t	maybe_wrap_dim (int64_t dim, TensorImpl *tensor)
static int64_t	maybe_wrap_dim (int64_t dim, TensorList tensors)
static int64_t	maybe_wrap_dim (int64_t dim, const std::vector< std::vector< int64_t > > &tensor_sizes)
static void	maybe_wrap_dims_n (int64_t *dims, int64_t ndims, int64_t dim_post_expr)
template<typename Container >
void	maybe_wrap_dims (Container &dims, int64_t dim_post_expr)
static int64_t	legacy_cat_wrap_dim (int64_t dim, const std::vector< std::vector< int64_t > > &tensor_sizes)
static int64_t	legacy_cat_wrap_dim (int64_t dim, TensorList tensors)
static void	wrap_all_dims (std::vector< int64_t > &dims_to_wrap, int64_t tensor_total_dims)
static std::bitset< dim_bitset_size >	dim_list_to_bitset (IntArrayRef dims, int64_t ndims)
	REGISTER_CONTEXT (DeviceType::CPU, caffe2::CPUContext)
	REGISTER_COPY_BYTES_FUNCTION (DeviceType::CPU, DeviceType::CPU, caffe2::CopyBytesWrapper)
	C10_DEFINE_TYPED_REGISTRY (ContextRegistry, at::DeviceType, at::BaseContext, std::unique_ptr, at::Device)
	C10_DECLARE_TYPED_REGISTRY (ContextRegistry, at::DeviceType, at::BaseContext, std::unique_ptr, at::Device)
std::unique_ptr< at::BaseContext >	CreateContext (const at::Device &device)
	REGISTER_CONTEXT (DeviceType::IDEEP, caffe2::IDEEPContext)
	REGISTER_COPY_BYTES_FUNCTION (DeviceType::IDEEP, DeviceType::CPU, CopyBytesWrapper)
	REGISTER_COPY_BYTES_FUNCTION (DeviceType::CPU, DeviceType::IDEEP, CopyBytesWrapper)
	REGISTER_COPY_BYTES_FUNCTION (DeviceType::IDEEP, DeviceType::IDEEP, CopyBytesWrapper)

Variables
constexpr int64_t	kVmapMaxTensorDims = 64
constexpr int64_t	kVmapNumLevels = 64
constexpr int64_t	kBatchDimsStackSize = 5
static const char	cublas_config_var_name [] = "CUBLAS_WORKSPACE_CONFIG"
static const char *const	cublas_deterministic_configs [] = { ":4096:8", ":16:8" }
thread_local bool	override_allow_tf32_flag = false
static Symbol	kWildcard = Symbol::dimname("*")
constexpr size_t	kDimVectorStaticSize = 5
thread_local bool	GradMode_enabled = true
	thread_local is a feature that is not enabled by Caffe2 mobile build (e.g. More...
constexpr int	MERSENNE_STATE_N = 624
constexpr int	MERSENNE_STATE_M = 397
constexpr uint32_t	MATRIX_A = 0x9908b0df
constexpr uint32_t	UMASK = 0x80000000
constexpr uint32_t	LMASK = 0x7fffffff
thread_local bool	NamesMode_enabled = true
constexpr size_t	kMaxNamedTensorDim = 64
static Allocator	CPU_fixed_allocator
constexpr const char *	CUDA_HELP
enum C10_API_ENUM	RecordScope
enum C10_API_ENUM	BACKWARD_FUNCTION
enum C10_API_ENUM	TORCHSCRIPT_FUNCTION
enum C10_API_ENUM	KERNEL_FUNCTION_DTYPE
enum C10_API_ENUM	USER_SCOPE
enum C10_API_ENUM	NUM_SCOPES
constexpr std::size_t	kSoftLimitCallbacks = 4
	$
constexpr int64_t	kVmapTransformStaticInputSize = 4
constexpr int64_t	kVmapStaticDimVecSize = 8
constexpr size_t	dim_bitset_size = 64

Detailed Description

Distributions kernel adapted from THRandom.cpp The kernels try to follow std::random distributions signature For instance: in ATen auto gen = at::detail::createCPUGenerator(); at::uniform_real_distribution<double> uniform(0, 1); auto sample = uniform(gen.get());.

This file provides distributions compatible with ATen/core/DistributionsHelper.h but backed with the std RNG implementation instead of the ATen one.

Contains the implementation of parallel reductions in TensorIterator.

This file contains some tensor-agnostic operations to be used in the core functions of the SobolEngine

This file contains tensor-agnostic SoboleEngine constants.

Flush-To-Zero and Denormals-Are-Zero mode.

vs std::random

std::mt19937 gen; std::uniform_real_distribution uniform(0, 1); auto sample = uniform(gen);

Flush-To-Zero (FTZ) and Denormals-Are-Zero (DAZ) are modes that bypass IEEE 754 methods of dealing with denormal floating-point numbers on x86-64 and some x86 CPUs. They result in reduced precision for values near zero, but increased performance.

See https://software.intel.com/en-us/articles/x87-and-sse-floating-point-assists-in-ia-32-flush-to-zero-ftz-and-denormals-are-zero-daz

Caffe2 mobile builds currently do not depend on all of ATen so this is required to allow using the faster ATen RNG for normal builds but keep the build size small on mobile. RNG performance typically doesn't matter on mobile builds since the models are small and rarely using random initialization.

Typedef Documentation

acc_type

template<typename T , bool is_cuda>

using at::acc_type = typedef typename AccumulateType<T, is_cuda>::type

Definition at line 52 of file AccumulateType.h.

BatchDims

using at::BatchDims = typedef SmallVector<BatchDim, kBatchDimsStackSize>

Definition at line 42 of file BatchedTensorImpl.h.

BatchDimsRef

using at::BatchDimsRef = typedef ArrayRef<BatchDim>

Definition at line 43 of file BatchedTensorImpl.h.

CallbackHandle

typedef uint64_t at::CallbackHandle

Definition at line 446 of file record_function.h.

CallbackHandles

typedef c10::SmallVector<uint64_t, kSoftLimitCallbacks> at::CallbackHandles

Definition at line 88 of file record_function.h.

CheckedFrom

using at::CheckedFrom = typedef const char*

Definition at line 40 of file TensorUtils.h.

ConstQuantizerPtr

typedef const c10::intrusive_ptr< Quantizer > & at::ConstQuantizerPtr

Definition at line 8 of file QuantizerBase.h.

DataType

using at::DataType = typedef caffe2::TypeIdentifier

Definition at line 105 of file typeid.h.

DimMask

using at::DimMask = typedef TensorIteratorBase::DimMask

Definition at line 13 of file TensorIterator.cpp.

DimnameList

using at::DimnameList = typedef c10::ArrayRef<Dimname>

Definition at line 36 of file Dimname.h.

DimVector

using at::DimVector = typedef SmallVector<int64_t, kDimVectorStaticSize>

A container for sizes or strides.

Definition at line 11 of file DimVector.h.

dist_acctype

template<typename T >

using at::dist_acctype = typedef typename DistAccumType<T>::type

Definition at line 29 of file TransformationHelper.h.

loop2d_t

typedef TensorIteratorBase::loop2d_t at::loop2d_t

Definition at line 12 of file TensorIteratorReduce.cpp.

loop_t

using at::loop_t = typedef TensorIteratorBase::loop_t

Definition at line 15 of file TensorIterator.cpp.

NameVector

using at::NameVector = typedef SmallVector<Dimname, kDimVectorStaticSize>

Definition at line 11 of file NamedTensorUtils.h.

ObserverContextList

typedef std::vector<std::unique_ptr<ObserverContext> > at::ObserverContextList

Definition at line 89 of file record_function.h.

PackedTensorAccessor32

template<typename T , size_t N, template< typename U > class PtrTraits = DefaultPtrTraits>

using at::PackedTensorAccessor32 = typedef GenericPackedTensorAccessor<T, N, PtrTraits, int32_t>

Definition at line 228 of file TensorAccessor.h.

PackedTensorAccessor64

template<typename T , size_t N, template< typename U > class PtrTraits = DefaultPtrTraits>

using at::PackedTensorAccessor64 = typedef GenericPackedTensorAccessor<T, N, PtrTraits, int64_t>

Definition at line 231 of file TensorAccessor.h.

PtrVector

using at::PtrVector = typedef TensorIteratorBase::PtrVector

Definition at line 14 of file TensorIterator.cpp.

QuantizerPtr

using at::QuantizerPtr = typedef c10::intrusive_ptr<Quantizer>

Definition at line 9 of file QuantizerBase.h.

RecordFunctionCallbacks

typedef std::vector<std::pair<RecordFunctionCallback, CallbackHandle> > at::RecordFunctionCallbacks

Definition at line 450 of file record_function.h.

RecordFunctionHandle

typedef uint64_t at::RecordFunctionHandle

Definition at line 90 of file record_function.h.

Stream

typedef c10::Stream at::Stream

Definition at line 56 of file TensorBody.h.

StrideVector

using at::StrideVector = typedef TensorIteratorBase::StrideVector

Definition at line 17 of file TensorIterator.cpp.

TensorList

typedef ArrayRef< Tensor > at::TensorList

Definition at line 54 of file TensorBody.h.

VmapDimVector

using at::VmapDimVector = typedef SmallVector<int64_t, kVmapStaticDimVecSize>

Definition at line 39 of file VmapTransforms.h.

VmapPhysicalViewVec

using at::VmapPhysicalViewVec = typedef SmallVector<VmapPhysicalView, kVmapTransformStaticInputSize>

Definition at line 33 of file VmapTransforms.h.

Enumeration Type Documentation

FastSetupType

enum class at::FastSetupType : uint8_t

strong

Enumerator
NONE
CONTIGUOUS
CHANNELS_LAST
NON_OVERLAPPING_DENSE

Definition at line 134 of file TensorIterator.h.

MemOverlap

enum class at::MemOverlap

strong

Enumerator
NO
YES
TOO_HARD

Definition at line 14 of file MemoryOverlap.h.

MemOverlapStatus

enum class at::MemOverlapStatus

strong

Enumerator
FULL
PARTIAL
NO
TOO_HARD

Definition at line 16 of file MemoryOverlap.h.

NameType

enum class at::NameType : uint8_t

strong

Enumerator
BASIC
WILDCARD

Definition at line 10 of file Dimname.h.

Function Documentation

__printFormat()

static std::tuple<double, int64_t> at::__printFormat ( std::ostream &  stream, const Tensor &  self  )

static

Definition at line 40 of file Formatting.cpp.

References std::ceil(), defaultfloat(), std::floor(), caffe2::int64_t, at::native::isfinite(), std::log10(), torch.jit._trace::make_tuple(), caffe2.perfkernels.hp_emblookup_codegen::offset, std::pow(), scale, size(), stream, sz, and at::native::metal::mpscnn::z.

Referenced by __printMatrix(), and print().

__printIndent()

static void at::__printIndent ( std::ostream &  stream, int64_t  indent  )

static

Definition at line 128 of file Formatting.cpp.

References torch::jit::indent(), caffe2::int64_t, and stream.

Referenced by __printMatrix().

__printMatrix()

static void at::__printMatrix ( std::ostream &  stream, const Tensor &  self, int64_t  linesize, int64_t  indent  )

static

Definition at line 139 of file Formatting.cpp.

References __printFormat(), __printIndent(), c, torch::jit::indent(), caffe2::int64_t, printScale(), caffe2::row, scale, size(), stream, and sz.

Referenced by __printTensor(), and print().

__printTensor()

void at::__printTensor	(	std::ostream &	stream,
		Tensor &	self,
		int64_t	linesize
	)

Definition at line 190 of file Formatting.cpp.

References __printMatrix(), caffe2::counter, caffe2::int64_t, size(), torch.cuda.profiler::start(), stream, and tensor.

Referenced by print().

_all_equal_numel()

bool at::_all_equal_numel ( at::ArrayRef< Tensor >  tensors )

inline

Definition at line 195 of file CPUApplyUtils.h.

References caffe2::int64_t, numel(), and caffe2::tensors.

Referenced by _apply_preamble().

_all_equal_numel_error()

std::string at::_all_equal_numel_error ( at::ArrayRef< Tensor >  tensors )

inline

Definition at line 206 of file CPUApplyUtils.h.

References caffe2::tensors.

Referenced by _apply_preamble().

_apply_preamble()

bool at::_apply_preamble ( ArrayRef< Tensor >  tensors )

inline

Definition at line 222 of file CPUApplyUtils.h.

References _all_equal_numel(), _all_equal_numel_error(), AT_ERROR, checkDeviceType(), checkLayout(), c10::kCPU, c10::kStrided, at::Tensor::numel(), and caffe2::tensors.

Referenced by CPU_tensor_apply2(), CPU_tensor_apply3(), and CPU_tensor_apply4().

_crash_if_asan()

TORCH_API int at::_crash_if_asan

(

int

arg

)

Definition at line 13 of file Utils.cpp.

References setup::arg, and bench_ops::x.

Referenced by THPModule_crashIfATenASAN().

_getTLSCallbacks()

TORCH_API RecordFunctionCallbacks at::_getTLSCallbacks

(

)

Definition at line 338 of file record_function.cpp.

_isnan() [1/3]

template<typename T , typename std::enable_if< std::is_same< T, at::BFloat16 >::value, int >::type = 0>

C10_HOST_DEVICE bool at::_isnan ( at::BFloat16  val )

inline

Definition at line 51 of file NumericUtils.h.

References _isnan(), and val.

_isnan() [2/3]

template<typename T , typename std::enable_if< std::is_integral< T >::value, int >::type = 0>

C10_HOST_DEVICE bool at::_isnan ( T  val )

inline

Definition at line 22 of file NumericUtils.h.

Referenced by _isnan(), at::native::MinMaxOps< scalar_t, acc_scalar_t, index_t >::combine(), at::native::detail::LessOrNan< scalar_t >::operator()(), at::native::detail::GreaterOrNan< scalar_t >::operator()(), and at::native::NanSumOps< acc_t, data_t >::reduce().

_isnan() [3/3]

template<typename T , typename std::enable_if< c10::is_complex< T >::value, int >::type = 0>

bool at::_isnan ( T  val )

inline

Definition at line 38 of file NumericUtils.h.

References c10::aten::isnan(), and val.

_max_dim_tensors()

int64_t at::_max_dim_tensors ( ArrayRef< Tensor >  tensors )

inline

Definition at line 234 of file CPUApplyUtils.h.

References dim, caffe2::int64_t, std::max(), at::Tensor::ndimension(), and caffe2::tensors.

Referenced by CPU_tensor_apply2(), CPU_tensor_apply3(), and CPU_tensor_apply4().

_setTLSCallbacks()

TORCH_API void at::_setTLSCallbacks

(

const RecordFunctionCallbacks &

callbacks

)

Definition at line 342 of file record_function.cpp.

References r, and caffe2.experiments.python.net_construct_bench::sort.

addBatchDim()

TORCH_API Tensor at::addBatchDim	(	const Tensor &	tensor,
		int64_t	level,
		int64_t	dim
	)

Definition at line 119 of file BatchedTensorImpl.cpp.

References dim, c10::SmallVectorImpl< T >::emplace_back(), caffe2.python.onnx.frontend::level, makeBatched(), maybeGetBatchedImpl(), and tensor.

Referenced by at::native::_add_batch_dim().

addGlobalCallback()

TORCH_API CallbackHandle at::addGlobalCallback

(

RecordFunctionCallback

cb

)

addGlobalCallback adds a global callback to run with RecordFunction:

WARNING: not thread safe, typically addGlobalCallback can be called only during the program initialization

Definition at line 372 of file record_function.cpp.

Referenced by pytorch_jni::PytorchJni::preModuleLoadSetupOnce(), and THPAutograd_initExtension().

addThreadLocalCallback()

TORCH_API CallbackHandle at::addThreadLocalCallback

(

RecordFunctionCallback

cb

)

addThreadLocalCallback adds a thread local callback to run with RecordFunction, returns handle to use with removeThreadLocalCallback

Definition at line 367 of file record_function.cpp.

Referenced by THPAutograd_initExtension().

alignBatchDimsAtFront()

static Tensor at::alignBatchDimsAtFront ( const Tensor &  self, std::bitset< kVmapNumLevels >  requested_levels, int64_t  requested_example_dim  )

static

Definition at line 125 of file VmapTransforms.cpp.

References c10::aten::copy(), getPhysicalTensorAndLevels(), caffe2::int64_t, caffe2.python.onnx.frontend::level, c10::SmallVectorTemplateCommon< T, typename >::rbegin(), c10::ArrayRef< T >::size(), at::Tensor::sizes(), and TORCH_INTERNAL_ASSERT.

Referenced by at::MultiBatchVmapTransform::logicalToPhysical(), and at::BroadcastingVmapTransform::logicalToPhysical().

apply_op() [1/2]

void at::apply_op ( )

inline

Definition at line 311 of file CPUApplyUtils.h.

Referenced by CPU_tensor_apply2(), CPU_tensor_apply3(), and CPU_tensor_apply4().

apply_op() [2/2]

template<typename Op , typename... Args>

void at::apply_op ( int64_t  numel, int64_t  offset, const Op &  op, Args...  iters  )

inline

Definition at line 315 of file CPUApplyUtils.h.

References forward(), caffe2::int64_t, iterate(), iterate_continue(), iterate_overflow(), max_dim(), numel(), and caffe2.perfkernels.hp_emblookup_codegen::offset.

are_expandable()

bool at::are_expandable ( IntArrayRef  shape1, IntArrayRef  shape2  )

inline

Definition at line 25 of file ExpandUtils.h.

References caffe2::int64_t, and c10::ArrayRef< T >::size().

areAllReturnsTensors()

static bool at::areAllReturnsTensors ( const FunctionSchema &  schema )

static

Definition at line 26 of file BatchedFallback.cpp.

References setup::arg.

Referenced by batchedTensorForLoopFallback().

areAnyArgumentsTensorList()

static bool at::areAnyArgumentsTensorList ( const FunctionSchema &  schema )

static

Definition at line 33 of file BatchedFallback.cpp.

References setup::arg.

Referenced by batchedTensorForLoopFallback().

areBdimsAtFrontInOrder()

static bool at::areBdimsAtFrontInOrder ( BatchDimsRef  bdims )

static

Definition at line 7 of file VmapTransforms.cpp.

References dim, c10::attr::idx(), caffe2::int64_t, and c10::ArrayRef< T >::size().

Referenced by permuteBatchDimsToFront().

as_strided_batching_rule()

Tensor at::as_strided_batching_rule	(	const Tensor &	tensor,
		IntArrayRef	sizes,
		IntArrayRef	strides,
		optional< int64_t >	storage_offset
	)

Definition at line 565 of file BatchingRegistrations.cpp.

References checkBasicAsStridedValidForSlice(), checkBatchDimsAtFrontInLayout(), c10::SmallVectorTemplateCommon< T, typename >::end(), c10::SmallVectorImpl< T >::insert(), at::MultiBatchVmapTransform::logicalToPhysical(), at::namedinference::num_batch_dims(), c10::SmallVectorImpl< T >::reserve(), sizes, strides, tensor, and TORCH_CHECK.

Referenced by TORCH_LIBRARY_IMPL().

assert_no_internal_overlap() [1/2]

TORCH_API void at::assert_no_internal_overlap

(

const Tensor &

t

)

Definition at line 28 of file MemoryOverlap.cpp.

References assert_no_internal_overlap(), and at::Tensor::unsafeGetTensorImpl().

Referenced by at::native::_cat_out_cpu(), assert_no_internal_overlap(), at::native::templates::bernoulli_impl_(), at::TensorIteratorBase::compute_mem_overlaps(), at::native::gather_out_cpu_cuda(), at::native::index_add_cpu_(), at::native::index_copy_(), at::native::index_out(), at::native::index_select_out_cpu_(), at::native::masked_scatter__cpu(), at::native::masked_select_out_impl_cpu(), at::native::scatter_(), at::native::scatter_add_(), at::native::scatter_fill_(), at::native::scatter_reduce_(), at::native::scatter_scalar_reduce_(), and at::native::take_out_cpu_template().

assert_no_internal_overlap() [2/2]

TORCH_API void at::assert_no_internal_overlap

(

TensorImpl *

t

)

Definition at line 32 of file MemoryOverlap.cpp.

References has_internal_overlap(), TORCH_CHECK, and YES.

assert_no_overlap() [1/2]

TORCH_API void at::assert_no_overlap	(	const Tensor &	a,
		const Tensor &	b
	)

Definition at line 81 of file MemoryOverlap.cpp.

References caffe2.contrib.aten.docs.sample::a, assert_no_overlap(), and at::native::metal::mpscnn::b.

Referenced by at::native::_index_put_impl_(), assert_no_overlap(), at::native::gather_out_cpu_cuda(), at::native::index_add_cpu_(), at::native::index_copy_(), at::native::index_fill_(), at::native::index_out(), at::native::index_select_out_cpu_(), at::native::masked_select_out_impl_cpu(), at::native::scatter_(), at::native::scatter_add_(), at::native::scatter_fill_(), at::native::scatter_reduce_(), at::native::scatter_scalar_reduce_(), and at::native::take_out_cpu_template().

assert_no_overlap() [2/2]

TORCH_API void at::assert_no_overlap	(	TensorImpl *	a,
		TensorImpl *	b
	)

Definition at line 85 of file MemoryOverlap.cpp.

References caffe2.contrib.aten.docs.sample::a, at::native::metal::mpscnn::b, FULL, get_overlap_status(), PARTIAL, and TORCH_CHECK.

assert_no_partial_overlap() [1/2]

TORCH_API void at::assert_no_partial_overlap	(	const Tensor &	a,
		const Tensor &	b
	)

Definition at line 70 of file MemoryOverlap.cpp.

References caffe2.contrib.aten.docs.sample::a, assert_no_partial_overlap(), and at::native::metal::mpscnn::b.

Referenced by assert_no_partial_overlap(), at::TensorIteratorBase::compute_mem_overlaps(), at::native::gather_out_cpu_cuda(), at::native::masked_fill__cuda(), at::native::masked_fill_impl_cpu(), and at::native::take_out_cpu_template().

assert_no_partial_overlap() [2/2]

void at::assert_no_partial_overlap	(	TensorImpl *	a,
		TensorImpl *	b
	)

Definition at line 74 of file MemoryOverlap.cpp.

References caffe2.contrib.aten.docs.sample::a, at::native::metal::mpscnn::b, get_overlap_status(), PARTIAL, and TORCH_CHECK.

batchedTensorForLoopFallback()

void at::batchedTensorForLoopFallback	(	const c10::OperatorHandle &	op,
		torch::jit::Stack *	stack
	)

Definition at line 245 of file BatchedFallback.cpp.

References areAllReturnsTensors(), areAnyArgumentsTensorList(), caffe2::argument, arguments, batch_sizes, batchedTensorInplaceForLoopFallback(), c10::SmallVectorTemplateCommon< T, typename >::begin(), computeIndex(), at::Tensor::defined(), torch::jit::drop(), c10::SmallVectorTemplateCommon< T, typename >::end(), c10::attr::idx(), index, c10::SmallVectorImpl< T >::insert(), caffe2::int64_t, isInplaceOp(), torch::jit::last(), at::MultiBatchVmapTransform::logicalToPhysical(), maybeGetBatchedImpl(), at::namedinference::num_batch_dims(), prod_intlist(), torch::jit::push(), c10::SmallVectorTemplateBase< T, isPodLike >::push_back(), caffe2::returns(), safeStack(), c10::SmallVectorTemplateCommon< T, typename >::size(), torch.distributions.constraints::stack, tensor, TORCH_CHECK, TORCH_INTERNAL_ASSERT, and warnFallback().

batchedTensorInplaceForLoopFallback()

void at::batchedTensorInplaceForLoopFallback	(	const c10::OperatorHandle &	op,
		torch::jit::Stack *	stack
	)

Definition at line 87 of file BatchedFallback.cpp.

References caffe2::argument, arguments, batch_sizes, c10::SmallVectorTemplateCommon< T, typename >::begin(), computeIndex(), createVmapLevelsBitset(), at::Tensor::defined(), torch::jit::drop(), c10::SmallVectorTemplateCommon< T, typename >::end(), llvm::findLastSet(), c10::attr::idx(), index, caffe2::int64_t, torch::jit::last(), at::MultiBatchVmapTransform::logicalToPhysical(), maybeGetBatchedImpl(), at::namedinference::num_batch_dims(), prod_intlist(), torch::jit::push(), c10::SmallVectorTemplateBase< T, isPodLike >::push_back(), c10::SmallVectorTemplateCommon< T, typename >::size(), torch.distributions.constraints::stack, tensor, TORCH_CHECK, TORCH_INTERNAL_ASSERT, and warnFallback().

Referenced by batchedTensorForLoopFallback().

binary_pointwise_batching_rule()

template<typename F , F Func, typename... ExtraArgs>

Tensor at::binary_pointwise_batching_rule	(	const Tensor &	self,
		const Tensor &	other,
		ExtraArgs...	args
	)

Definition at line 85 of file BatchingRegistrations.cpp.

References compare-fastrnn-results::args, dim, at::Tensor::dim(), isPhysicalScalarTensor(), at::MultiBatchVmapTransform::logicalToPhysical(), at::BroadcastingVmapTransform::logicalToPhysical(), at::meta::other, at::native::result_type(), and tensor.

Referenced by TORCH_LIBRARY_IMPL().

bmm_batching_rule()

Tensor at::bmm_batching_rule	(	const Tensor &	self,
		const Tensor &	other
	)

Definition at line 833 of file BatchingRegistrations.cpp.

References dim, at::Tensor::dim(), at::BroadcastingVmapTransform::logicalToPhysical(), torch.backends.cuda::matmul, at::meta::other, at::Tensor::sizes(), tensor, and TORCH_CHECK.

Referenced by TORCH_LIBRARY_IMPL().

bumpRecordAllFunctions()

TORCH_API void at::bumpRecordAllFunctions

(

)

Definition at line 487 of file record_function.cpp.

Referenced by at::CallbackManager::addGlobalCallback(), at::CallbackManager::addThreadLocalCallback(), and at::ThreadLocalStateGuard::ThreadLocalStateGuard().

C10_DECLARE_REGISTRY() [1/2]

at::C10_DECLARE_REGISTRY	(	CUDAHooksRegistry	,
		CUDAHooksInterface	,
		CUDAHooksArgs
	)

C10_DECLARE_REGISTRY() [2/2]

at::C10_DECLARE_REGISTRY	(	HIPHooksRegistry	,
		HIPHooksInterface	,
		HIPHooksArgs
	)

C10_DECLARE_TYPED_REGISTRY()

at::C10_DECLARE_TYPED_REGISTRY	(	ContextRegistry	,
		at::DeviceType	,
		at::BaseContext	,
		std::unique_ptr	,
		at::Device
	)

C10_DEFINE_TYPED_REGISTRY()

at::C10_DEFINE_TYPED_REGISTRY	(	ContextRegistry	,
		at::DeviceType	,
		at::BaseContext	,
		std::unique_ptr	,
		at::Device
	)

cat_batching_rule()

Tensor at::cat_batching_rule	(	TensorList	tensors,
		int64_t	dim
	)

Definition at line 872 of file BatchingRegistrations.cpp.

References torch.distributions.constraints::cat, dim, c10::fmap(), at::MultiBatchVmapTransform::logicalToPhysical(), caffe2::tensors, TORCH_INTERNAL_ASSERT, and at::native::metal::view().

Referenced by TORCH_LIBRARY_IMPL().

check_defined()

void at::check_defined ( std::initializer_list< std::reference_wrapper< const Tensor > >  tensors, const char *  api_name  )

inline

Definition at line 40 of file ExpandUtils.h.

References AT_ERROR, at::Tensor::defined(), and caffe2::tensors.

Referenced by expand_inplace(), expand_outplace(), and expand_size().

check_dim_size()

TORCH_API void at::check_dim_size	(	const Tensor &	tensor,
		int64_t	dim,
		int64_t	dim_size,
		int64_t	size
	)

Definition at line 286 of file TensorUtils.cpp.

References dim, at::Tensor::dim(), size(), at::Tensor::size(), tensor, and TORCH_CHECK.

Referenced by at::meta::upsample_nearest1d_backward().

check_for_misalignment()

static void at::check_for_misalignment ( const Dimname &  name, DimnameList  names, DimnameList  other_names, const char *  action  )

static

Definition at line 52 of file NamedTensorUtils.cpp.

References fastrnns.bench::action, c10::ArrayRef< T >::begin(), c10::ArrayRef< T >::end(), caffe2::it, name, microbenchmarks::names, and TORCH_CHECK.

Referenced by unify_from_right().

check_generator()

template<typename T >

static T* at::check_generator ( c10::optional< Generator >  gen )

inlinestatic

Utility function to static cast input Generator* to the backend generator type (CPU/CUDAGeneratorImpl etc.)

Definition at line 113 of file Utils.h.

References caffe2.python.gradient_check_test::device_type, tools.autograd.gen_python_functions::gen(), and TORCH_CHECK.

check_intlist()

template<size_t N>

std::array<int64_t, N> at::check_intlist	(	ArrayRef< int64_t >	list,
		const char *	name,
		int	pos
	)

Definition at line 74 of file Utils.h.

References AT_ERROR, c10::prim::list(), N, name, and caffe2::python::res.

check_names_valid_for() [1/2]

void at::check_names_valid_for	(	const Tensor &	tensor,
		DimnameList	names
	)

Definition at line 51 of file NamedTensor.cpp.

References at::impl::check_names_valid_for(), microbenchmarks::names, tensor, and at::Tensor::unsafeGetTensorImpl().

Referenced by at::native::align_to(), and at::native::refine_names().

check_names_valid_for() [2/2]

void at::check_names_valid_for	(	size_t	tensor_dim,
		DimnameList	names
	)

Definition at line 55 of file NamedTensor.cpp.

References check_unique_names(), kMaxNamedTensorDim, microbenchmarks::names, and TORCH_CHECK.

check_size_nonnegative()

void at::check_size_nonnegative ( IntArrayRef  size )

inline

Definition at line 131 of file Utils.h.

References size(), TORCH_CHECK, and bench_ops::x.

Referenced by at::detail::empty_cpu(), at::native::empty_out(), at::native::empty_strided_cpu(), and new_qtensor().

check_unique_names()

static void at::check_unique_names ( DimnameList  names )

static

Definition at line 37 of file NamedTensor.cpp.

References caffe2::it, microbenchmarks::names, and TORCH_CHECK.

Referenced by check_names_valid_for().

check_valid_identifier()

static void at::check_valid_identifier ( const std::string &  name )

static

Definition at line 37 of file Dimname.cpp.

References at::Dimname::isValidName(), name, and TORCH_CHECK.

Referenced by at::Dimname::fromSymbol().

checkAllContiguous()

TORCH_API void at::checkAllContiguous	(	CheckedFrom	c,
		at::ArrayRef< TensorArg >	ts
	)

Definition at line 62 of file TensorUtils.cpp.

References c, checkContiguous(), and at::Tensor::defined().

checkAllDefined()

TORCH_API void at::checkAllDefined	(	CheckedFrom	c,
		ArrayRef< TensorArg >	ts
	)

Definition at line 209 of file TensorUtils.cpp.

References c, and checkDefined().

checkAllSame()

void at::checkAllSame	(	CheckedFrom	c,
		ArrayRef< TensorArg >	tensors,
		void(*)(CheckedFrom, const TensorArg &, const TensorArg &)	fn
	)

Definition at line 85 of file TensorUtils.cpp.

References c, at::Tensor::defined(), caffe2.perfkernels.hp_emblookup_codegen::fn, and caffe2::tensors.

Referenced by checkAllSameGPU(), checkAllSameNumel(), checkAllSameSize(), and checkAllSameType().

checkAllSameGPU()

TORCH_API void at::checkAllSameGPU	(	CheckedFrom	c,
		ArrayRef< TensorArg >	tensors
	)

Definition at line 150 of file TensorUtils.cpp.

References c, checkAllSame(), checkSameGPU(), and caffe2::tensors.

checkAllSameNumel()

TORCH_API void at::checkAllSameNumel	(	CheckedFrom	c,
		ArrayRef< TensorArg >	tensors
	)

Definition at line 126 of file TensorUtils.cpp.

References c, checkAllSame(), checkSameNumel(), and caffe2::tensors.

checkAllSameSize()

void at::checkAllSameSize	(	CheckedFrom	c,
		ArrayRef< TensorArg >	tensors
	)

Definition at line 105 of file TensorUtils.cpp.

References c, checkAllSame(), checkSameSize(), and caffe2::tensors.

checkAllSameType()

TORCH_API void at::checkAllSameType	(	CheckedFrom	c,
		ArrayRef< TensorArg >	tensors
	)

Definition at line 190 of file TensorUtils.cpp.

References c, checkAllSame(), checkSameType(), and caffe2::tensors.

checkBackend() [1/2]

TORCH_API void at::checkBackend	(	CheckedFrom	c,
		at::ArrayRef< Tensor >	tensors,
		at::Backend	backend
	)

Definition at line 224 of file TensorUtils.cpp.

References c, checkBackend(), and caffe2::tensors.

checkBackend() [2/2]

void at::checkBackend	(	CheckedFrom	c,
		const Tensor &	t,
		Backend	backend
	)

Definition at line 216 of file TensorUtils.cpp.

References c10::TensorOptions::backend(), c, at::Tensor::defined(), at::Tensor::options(), TORCH_CHECK, and c10::toString().

Referenced by at::native::addcdiv_out(), at::native::addcmul_out(), at::native::batch_norm_cpu(), and checkBackend().

checkBasicAsStridedValidForSlice()

static void at::checkBasicAsStridedValidForSlice ( const Tensor &  physical_tensor, int64_t  num_batch_dims, IntArrayRef  sizes, IntArrayRef  strides, optional< int64_t >  maybe_storage_offset  )

static

Definition at line 506 of file BatchingRegistrations.cpp.

References maximum_indexable_location(), at::namedinference::num_batch_dims(), sizes, at::Tensor::sizes(), c10::ArrayRef< T >::slice(), at::Tensor::storage_offset(), strides, at::Tensor::strides(), TORCH_CHECK, and c10::optional< T >::value_or().

Referenced by as_strided_batching_rule().

checkBatchDimsAtFrontInLayout()

static void at::checkBatchDimsAtFrontInLayout ( IntArrayRef  physical_strides, int64_t  num_batch_dims  )

static

Definition at line 473 of file BatchingRegistrations.cpp.

References c10::ArrayRef< T >::begin(), c10::ArrayRef< T >::end(), at::namedinference::num_batch_dims(), and TORCH_CHECK.

Referenced by as_strided_batching_rule().

checkContiguous()

TORCH_API void at::checkContiguous	(	CheckedFrom	c,
		const TensorGeometryArg &	t
	)

Definition at line 55 of file TensorUtils.cpp.

References c, at::Tensor::is_contiguous(), and TORCH_CHECK.

Referenced by at::native::_embedding_bag_backward(), at::native::_embedding_bag_per_sample_weights_backward_cpu_template(), and checkAllContiguous().

checkDefined()

TORCH_API void at::checkDefined	(	CheckedFrom	c,
		const TensorArg &	t
	)

Definition at line 202 of file TensorUtils.cpp.

References c, at::Tensor::defined(), and TORCH_CHECK.

Referenced by checkAllDefined().

checkDeviceType() [1/2]

TORCH_API void at::checkDeviceType	(	CheckedFrom	c,
		at::ArrayRef< Tensor >	tensors,
		at::DeviceType	device_type
	)

Definition at line 238 of file TensorUtils.cpp.

References c, checkDeviceType(), caffe2.python.gradient_check_test::device_type, and caffe2::tensors.

checkDeviceType() [2/2]

void at::checkDeviceType	(	CheckedFrom	c,
		const Tensor &	t,
		DeviceType	device_type
	)

Definition at line 230 of file TensorUtils.cpp.

References c, at::Tensor::defined(), at::Tensor::device(), caffe2.python.gradient_check_test::device_type, TORCH_CHECK, and c10::Device::type().

Referenced by _apply_preamble(), checkDeviceType(), at::native::einsum(), at::native::inner(), and at::native::inner_out().

checkDim() [1/2]

TORCH_API void at::checkDim	(	CheckedFrom	c,
		const Tensor &	tensor,
		const char *	name,
		int	pos,
		int64_t	dim
	)

Definition at line 22 of file TensorUtils.cpp.

References c, dim, at::Tensor::dim(), name, tensor, and TORCH_CHECK.

Referenced by at::native::adaptive_avg_pool1d(), at::native::adaptive_max_pool1d(), at::native::avg_pool1d(), at::native::bmm_out_or_baddbmm_(), checkSize(), at::native::embedding_renorm_cpu_(), and at::native::max_pool1d_with_indices().

checkDim() [2/2]

TORCH_API void at::checkDim	(	CheckedFrom	c,
		const TensorGeometryArg &	t,
		int64_t	dim
	)

Definition at line 41 of file TensorUtils.cpp.

References c, dim, at::Tensor::dim(), and TORCH_CHECK.

checkDimRange()

TORCH_API void at::checkDimRange	(	CheckedFrom	c,
		const TensorGeometryArg &	t,
		int64_t	dim_start,
		int64_t	dim_end
	)

Definition at line 47 of file TensorUtils.cpp.

References c, at::Tensor::dim(), and TORCH_CHECK.

checkDL()

static void* at::checkDL ( void *  x )

static

Definition at line 21 of file DynamicLibrary.cpp.

References AT_ERROR, and bench_ops::x.

Referenced by at::DynamicLibrary::DynamicLibrary(), and at::DynamicLibrary::sym().

checked_dense_tensor_list_unwrap()

static std::vector<TensorImpl*> at::checked_dense_tensor_list_unwrap ( ArrayRef< Tensor >  tensors, const char *  name, int  pos, DeviceType  device_type, ScalarType  scalar_type  )

inlinestatic

Definition at line 51 of file Utils.h.

References AT_ERROR, caffe2.python.gradient_check_test::device_type, name, detail::scalar_type(), and caffe2::tensors.

checked_dense_tensor_unwrap()

static TensorImpl* at::checked_dense_tensor_unwrap ( const Tensor &  expr, const char *  name, int  pos, const char *  api, bool  allowNull, DeviceType  device_type, ScalarType  scalar_type  )

inlinestatic

Definition at line 29 of file Utils.h.

References AT_ERROR, at::Tensor::defined(), at::Tensor::device(), caffe2.python.gradient_check_test::device_type, at::Tensor::layout(), name, at::Tensor::scalar_type(), detail::scalar_type(), c10::Device::type(), and at::Tensor::unsafeGetTensorImpl().

Referenced by at::native::legacy::cuda::_th_copy_ignoring_overlaps_(), at::native::legacy::cuda::_th_cross_kernel(), at::native::legacy::cuda::_th_cross_kernel_out(), at::native::legacy::cuda::_th_gels(), at::native::legacy::cpu::_th_gels(), at::native::legacy::cuda::_th_gels_out(), at::native::legacy::cpu::_th_gels_out(), at::native::legacy::cuda::_th_geqrf(), at::native::legacy::cpu::_th_geqrf(), at::native::legacy::cuda::_th_geqrf_out(), at::native::legacy::cpu::_th_geqrf_out(), at::native::legacy::cpu::_th_histc(), at::native::legacy::cpu::_th_histc_out(), at::native::legacy::cuda::_th_index_copy_(), at::native::legacy::cpu::_th_index_copy_(), at::native::legacy::cuda::_th_masked_fill_(), at::native::legacy::cuda::_th_masked_fill_bool_(), at::native::legacy::cuda::_th_mode(), at::native::legacy::cpu::_th_mode(), at::native::legacy::cuda::_th_mode_out(), at::native::legacy::cpu::_th_mode_out(), at::native::legacy::cpu::_th_nonzero(), at::native::legacy::cpu::_th_nonzero_out(), at::native::legacy::cpu::_th_ormqr(), at::native::legacy::cpu::_th_ormqr_out(), at::native::legacy::cuda::_th_put_(), at::native::legacy::cpu::_th_put_(), at::native::legacy::cuda::_th_renorm(), at::native::legacy::cpu::_th_renorm(), at::native::legacy::cuda::_th_renorm_(), at::native::legacy::cpu::_th_renorm_(), at::native::legacy::cuda::_th_renorm_out(), at::native::legacy::cpu::_th_renorm_out(), at::native::legacy::cuda::_th_sort(), at::native::legacy::cuda::_th_sort_out(), at::native::legacy::cpu::_th_std(), at::native::legacy::cuda::_th_topk(), at::native::legacy::cuda::_th_topk_out(), at::native::legacy::cpu::_th_var(), at::native::legacy::cuda::_thnn_conv2d_backward(), at::native::legacy::cuda::_thnn_conv2d_backward_out(), at::native::legacy::cuda::_thnn_conv2d_forward(), at::native::legacy::cuda::_thnn_conv2d_forward_out(), at::native::legacy::cuda::_thnn_conv_depthwise2d_backward(), at::native::legacy::cuda::_thnn_conv_depthwise2d_backward_out(), at::native::legacy::cuda::_thnn_conv_depthwise2d_forward(), at::native::legacy::cuda::_thnn_conv_depthwise2d_forward_out(), at::native::legacy::cuda::_thnn_glu_backward(), at::native::legacy::cuda::_thnn_glu_backward_out(), at::native::legacy::cuda::_thnn_glu_forward(), at::native::legacy::cuda::_thnn_glu_forward_out(), at::native::legacy::cuda::_thnn_log_sigmoid_backward(), at::native::legacy::cuda::_thnn_log_sigmoid_backward_out(), at::native::legacy::cuda::_thnn_log_sigmoid_forward(), at::native::legacy::cuda::_thnn_log_sigmoid_forward_out(), at::native::legacy::cuda::_thnn_multi_margin_loss_backward(), at::native::legacy::cuda::_thnn_multi_margin_loss_backward_out(), at::native::legacy::cuda::_thnn_multi_margin_loss_forward(), at::native::legacy::cuda::_thnn_multi_margin_loss_forward_out(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_backward(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_backward_out(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_forward(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_forward_out(), at::native::legacy::cuda::_thnn_nll_loss2d_backward(), at::native::legacy::cuda::_thnn_nll_loss2d_backward_out(), at::native::legacy::cuda::_thnn_nll_loss2d_forward(), at::native::legacy::cuda::_thnn_nll_loss2d_forward_out(), at::native::legacy::cuda::_thnn_nll_loss_backward(), at::native::legacy::cuda::_thnn_nll_loss_backward_out(), at::native::legacy::cuda::_thnn_nll_loss_forward(), at::native::legacy::cuda::_thnn_nll_loss_forward_out(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward_(), and at::native::legacy::cuda::_thnn_rrelu_with_noise_forward_out().

checkLayout() [1/2]

TORCH_API void at::checkLayout	(	CheckedFrom	c,
		at::ArrayRef< Tensor >	tensors,
		at::Layout	layout
	)

Definition at line 252 of file TensorUtils.cpp.

References c, checkLayout(), caffe2::layout, and caffe2::tensors.

checkLayout() [2/2]

TORCH_API void at::checkLayout	(	CheckedFrom	c,
		const Tensor &	t,
		Layout	layout
	)

Definition at line 244 of file TensorUtils.cpp.

References c, at::Tensor::defined(), at::Tensor::layout(), caffe2::layout, and TORCH_CHECK.

Referenced by _apply_preamble(), and checkLayout().

checkNumel()

TORCH_API void at::checkNumel	(	CheckedFrom	c,
		const TensorGeometryArg &	t,
		int64_t	numel
	)

Definition at line 109 of file TensorUtils.cpp.

References c, at::Tensor::numel(), numel(), and TORCH_CHECK.

checkRecordAllFunctions()

TORCH_API bool at::checkRecordAllFunctions

(

)

Definition at line 495 of file record_function.cpp.

Referenced by at::ThreadLocalState::ThreadLocalState().

checkSameDim()

TORCH_API void at::checkSameDim	(	CheckedFrom	c,
		const TensorGeometryArg &	t1,
		const TensorGeometryArg &	t2
	)

Definition at line 194 of file TensorUtils.cpp.

References c, pyspybench::t1, pyspybench::t2, and TORCH_CHECK.

checkSameGPU()

TORCH_API void at::checkSameGPU	(	CheckedFrom	c,
		const TensorArg &	t1,
		const TensorArg &	t2
	)

Definition at line 130 of file TensorUtils.cpp.

References AT_ERROR, c, pyspybench::t1, pyspybench::t2, and TORCH_CHECK.

Referenced by checkAllSameGPU().

checkSameNumel() [1/2]

void at::checkSameNumel	(	CheckedFrom	c,
		const TensorArg &	t1,
		const TensorArg &	t2
	)

Definition at line 117 of file TensorUtils.cpp.

References c, pyspybench::t1, pyspybench::t2, and TORCH_CHECK.

Referenced by checkAllSameNumel().

checkSameNumel() [2/2]

TORCH_API void at::checkSameNumel	(	CheckedFrom	c,
		const TensorGeometryArg &	t1,
		const TensorGeometryArg &	t2
	)

checkSameSize()

TORCH_API void at::checkSameSize	(	CheckedFrom	c,
		const TensorArg &	t1,
		const TensorArg &	t2
	)

Definition at line 97 of file TensorUtils.cpp.

References c, pyspybench::t1, pyspybench::t2, and TORCH_CHECK.

Referenced by checkAllSameSize(), at::native::dequantize_tensor_per_channel_affine(), at::native::dequantize_tensor_per_channel_float_qparams(), at::native::dequantize_tensor_per_tensor_affine(), at::native::log_softmax_backward_cpu(), at::native::quantize_tensor_per_channel_affine(), at::native::quantize_tensor_per_channel_float_qparams(), at::native::quantize_tensor_per_tensor_affine(), at::native::softmax_backward_cpu(), and at::native::softmax_backward_sparse_input_preprocessing().

checkSameType()

TORCH_API void at::checkSameType	(	CheckedFrom	c,
		const TensorArg &	t1,
		const TensorArg &	t2
	)

Definition at line 154 of file TensorUtils.cpp.

References c, pyspybench::t1, pyspybench::t2, and TORCH_CHECK.

Referenced by at::native::_embedding_bag_backward(), at::native::_embedding_bag_cpu_impl(), and checkAllSameType().

checkScalarType()

TORCH_API void at::checkScalarType	(	CheckedFrom	c,
		const TensorArg &	t,
		ScalarType	ty
	)

Definition at line 162 of file TensorUtils.cpp.

References c, at::Tensor::scalar_type(), TORCH_CHECK, at::Tensor::toString(), and c10::toString().

checkScalarTypes()

TORCH_API void at::checkScalarTypes	(	CheckedFrom	c,
		const TensorArg &	t,
		at::ArrayRef< ScalarType >	l
	)

Definition at line 170 of file TensorUtils.cpp.

References AT_ERROR, c10::ArrayRef< T >::begin(), c, c10::ArrayRef< T >::end(), at::Tensor::scalar_type(), at::Tensor::toString(), and c10::toString().

Referenced by at::native::_embedding_bag_backward(), at::native::_embedding_bag_cpu_impl(), at::native::_embedding_bag_dense_backward_cpu(), at::native::_embedding_bag_per_sample_weights_backward_cpu_template(), at::native::embedding(), at::native::embedding_dense_backward_cpu(), at::native::embedding_renorm_cpu_(), and at::native::embedding_sparse_backward().

checkSize() [1/2]

TORCH_API void at::checkSize	(	CheckedFrom	c,
		const TensorGeometryArg &	t,
		int64_t	dim,
		int64_t	size
	)

Definition at line 77 of file TensorUtils.cpp.

References c, dim, size(), at::Tensor::size(), and TORCH_CHECK.

checkSize() [2/2]

TORCH_API void at::checkSize	(	CheckedFrom	c,
		const TensorGeometryArg &	t,
		IntArrayRef	sizes
	)

Definition at line 69 of file TensorUtils.cpp.

References c, checkDim(), c10::ArrayRef< T >::equals(), sizes, at::Tensor::sizes(), and TORCH_CHECK.

chunk_batching_rule()

std::vector<Tensor> at::chunk_batching_rule	(	const Tensor &	self,
		int64_t	chunks,
		int64_t	dim
	)

Definition at line 182 of file BatchingRegistrations.cpp.

References at::native::chunk(), dim, and at::MultiBatchVmapTransform::logicalToPhysical().

Referenced by TORCH_LIBRARY_IMPL().

clamp_batching_rule()

Tensor at::clamp_batching_rule	(	const Tensor &	self,
		optional< Scalar >	min,
		optional< Scalar >	max
	)

Definition at line 190 of file BatchingRegistrations.cpp.

References c10::aten::clamp(), at::MultiBatchVmapTransform::logicalToPhysical(), max, and min.

Referenced by TORCH_LIBRARY_IMPL().

clamp_max_batching_rule()

Tensor at::clamp_max_batching_rule	(	const Tensor &	self,
		Scalar	max
	)

Definition at line 202 of file BatchingRegistrations.cpp.

References at::native::clamp_max(), at::MultiBatchVmapTransform::logicalToPhysical(), and max.

Referenced by TORCH_LIBRARY_IMPL().

clamp_min_batching_rule()

Tensor at::clamp_min_batching_rule	(	const Tensor &	self,
		Scalar	min
	)

Definition at line 196 of file BatchingRegistrations.cpp.

References at::native::clamp_min(), at::MultiBatchVmapTransform::logicalToPhysical(), and min.

Referenced by TORCH_LIBRARY_IMPL().

clearCallbacks()

TORCH_API void at::clearCallbacks

(

)

Definition at line 389 of file record_function.cpp.

References matmul_dlmc_bench::m.

Referenced by main(), and THPAutograd_initExtension().

clearGlobalCallbacks()

TORCH_API void at::clearGlobalCallbacks

(

)

clearGlobalCallbacks removes all global callbacks WARNING: not thread safe

Definition at line 381 of file record_function.cpp.

clearThreadLocalCallbacks()

TORCH_API void at::clearThreadLocalCallbacks

(

)

clearThreadLocalCallbacks removes all thread local callbacks

Definition at line 385 of file record_function.cpp.

clone_batching_rule()

Tensor at::clone_batching_rule	(	const Tensor &	self,
		optional< MemoryFormat >	memory_format
	)

Definition at line 718 of file BatchingRegistrations.cpp.

References at::native::clone(), at::MultiBatchVmapTransform::logicalToPhysical(), makeBatched(), c10::memory_format(), TORCH_CHECK, TORCH_INTERNAL_ASSERT, and unsafeGetBatchedImpl().

Referenced by TORCH_LIBRARY_IMPL().

collapse_dims()

template<typename T >

std::pair<int64_t, int64_t> at::collapse_dims ( T *  sizes, T *  strides, int64_t  dims, const int  excludeDim = -1  )

inline

Definition at line 28 of file CPUApplyUtils.h.

References caffe2::dims, caffe2::int64_t, sizes, strides, and TORCH_CHECK.

Referenced by at::strided_tensor_iter< T >::strided_tensor_iter(), and at::strided_tensor_iter_fixed< T, N >::strided_tensor_iter_fixed().

comparison_pointwise_batching_rule()

template<typename F , F Func>

Tensor at::comparison_pointwise_batching_rule	(	const Tensor &	self,
		const Tensor &	other
	)

Definition at line 1010 of file BatchingRegistrations.cpp.

References at::BroadcastingVmapTransform::logicalToPhysical(), at::meta::other, and tensor.

computeFrontBatchDimsFromLevels()

static BatchDims at::computeFrontBatchDimsFromLevels ( std::bitset< kVmapNumLevels >  levels_bitset )

static

Definition at line 82 of file VmapTransforms.cpp.

References dim, c10::SmallVectorImpl< T >::emplace_back(), caffe2::int64_t, kVmapNumLevels, and caffe2.python.onnx.frontend::level.

Referenced by at::VmapPhysicalToLogicalMap::apply().

computeIndex()

static SmallVector<indexing::TensorIndex,kVmapStaticDimVecSize> at::computeIndex ( int64_t  linear_idx, IntArrayRef  sizes  )

static

Definition at line 13 of file BatchedFallback.cpp.

References matmul_dlmc_bench::end, caffe2::it, c10::SmallVectorTemplateBase< T, isPodLike >::push_back(), at::native::remainder(), c10::SmallVectorImpl< T >::reserve(), fastrnns.custom_lstms::reverse(), and sizes.

Referenced by batchedTensorForLoopFallback(), and batchedTensorInplaceForLoopFallback().

contiguous_batching_rule()

Tensor at::contiguous_batching_rule	(	const Tensor &	self,
		MemoryFormat	memory_format
	)

Definition at line 445 of file BatchingRegistrations.cpp.

References at::MultiBatchVmapTransform::logicalToPhysical(), c10::memory_format(), and TORCH_CHECK.

Referenced by TORCH_LIBRARY_IMPL().

CPU()

static DeprecatedTypeProperties& at::CPU ( ScalarType  s )

inlinestatic

Definition at line 247 of file Context.h.

References c10::CPU, at::DeprecatedTypePropertiesRegistry::getDeprecatedTypeProperties(), and globalDeprecatedTypePropertiesRegistry().

Referenced by torch::jit::initJITBindings(), and torch::jit::Unpickler::readInstruction().

cpu_fixed_free()

static at::cpu_fixed_free ( void *  state, void *  allocation  )

static

Definition at line 22 of file CPUFixedAllocator.h.

References caffe2::function, and state.

cpu_fixed_malloc()

static at::cpu_fixed_malloc ( void *  , ptrdiff_t    )

static

Definition at line 14 of file CPUFixedAllocator.h.

References AT_ERROR.

cpu_fixed_realloc()

static at::cpu_fixed_realloc ( void *  , void *  , ptrdiff_t    )

static

Definition at line 18 of file CPUFixedAllocator.h.

References AT_ERROR.

CPU_tensor_apply2()

template<typename scalar1 , typename scalar2 , typename Op >

void at::CPU_tensor_apply2 ( Tensor  tensor1, Tensor  tensor2, const Op  op  )

inline

Definition at line 345 of file CPUApplyUtils.h.

References _apply_preamble(), _max_dim_tensors(), apply_op(), and at::Tensor::numel().

CPU_tensor_apply3()

template<typename scalar1 , typename scalar2 , typename scalar3 , typename Op >

void at::CPU_tensor_apply3 ( Tensor  tensor1, Tensor  tensor2, Tensor  tensor3, const Op  op  )

inline

Definition at line 367 of file CPUApplyUtils.h.

References _apply_preamble(), _max_dim_tensors(), apply_op(), and at::Tensor::numel().

CPU_tensor_apply4()

template<typename scalar1 , typename scalar2 , typename scalar3 , typename scalar4 , typename Op >

void at::CPU_tensor_apply4 ( Tensor  tensor1, Tensor  tensor2, Tensor  tensor3, Tensor  tensor4, const Op  op  )

inline

Definition at line 395 of file CPUApplyUtils.h.

References _apply_preamble(), _max_dim_tensors(), apply_op(), and at::Tensor::numel().

createBatchDimBitset()

std::bitset<kVmapMaxTensorDims> at::createBatchDimBitset ( BatchDimsRef  bdims )

inline

Definition at line 117 of file BatchedTensorImpl.h.

Referenced by at::BatchedTensorImpl::actualDim(), and permuteBatchDimsToFront().

CreateContext()

std::unique_ptr<at::BaseContext> at::CreateContext ( const at::Device &  device )

inline

Definition at line 157 of file context_base.h.

References device.

Referenced by caffe2::ContextFromProto(), caffe2::python::TensorFetcher::FetchTensor(), and caffe2::TensorSerializer::Serialize().

createVmapLevelsBitset()

std::bitset<kVmapNumLevels> at::createVmapLevelsBitset ( BatchDimsRef  bdims )

inline

Definition at line 126 of file BatchedTensorImpl.h.

Referenced by batchedTensorInplaceForLoopFallback(), getLevelsAndLargestLogicalDim(), getPhysicalTensorAndLevels(), inplaceIsVmapCompatible(), and at::MultiBatchVmapTransform::logicalToPhysical().

CUDA()

static DeprecatedTypeProperties& at::CUDA ( ScalarType  s )

inlinestatic

Definition at line 252 of file Context.h.

References c10::CUDA, at::DeprecatedTypePropertiesRegistry::getDeprecatedTypeProperties(), and globalDeprecatedTypePropertiesRegistry().

default_names()

DimnameList at::default_names

(

size_t

len

)

Definition at line 29 of file NamedTensor.cpp.

References kMaxNamedTensorDim, c10::aten::len(), TORCH_INTERNAL_ASSERT, and at::Dimname::wildcard().

Referenced by at::impl::get_names().

defaultfloat()

std::ios_base& at::defaultfloat ( std::ios_base &  __base )

inline

Definition at line 18 of file Formatting.cpp.

Referenced by __printFormat(), print(), and printScale().

deleter()

void at::deleter

(

DLManagedTensor *

arg

)

Definition at line 181 of file DLConvertor.cpp.

References setup::arg.

Referenced by torch::jit::tensorexpr::LoopNest::eliminateDeadStores(), torch::from_blob(), fromDLPack(), c10::InefficientStdFunctionContext::makeDataPtr(), and toDLPack().

device_of() [1/2]

optional<Device> at::device_of ( const Tensor &  t )

inline

Return the Device of a Tensor, if the Tensor is defined.

Definition at line 17 of file DeviceGuard.h.

References at::Tensor::defined(), at::Tensor::device(), c10::make_optional(), and c10::nullopt.

Referenced by at::native::legacy::cuda::_thnn_conv2d_backward(), at::native::legacy::cuda::_thnn_conv2d_backward_out(), at::native::legacy::cuda::_thnn_conv2d_forward(), at::native::legacy::cuda::_thnn_conv2d_forward_out(), at::native::legacy::cuda::_thnn_conv_depthwise2d_backward(), at::native::legacy::cuda::_thnn_conv_depthwise2d_backward_out(), at::native::legacy::cuda::_thnn_conv_depthwise2d_forward(), at::native::legacy::cuda::_thnn_conv_depthwise2d_forward_out(), at::native::legacy::cuda::_thnn_glu_backward(), at::native::legacy::cuda::_thnn_glu_backward_out(), at::native::legacy::cuda::_thnn_glu_forward(), at::native::legacy::cuda::_thnn_glu_forward_out(), at::native::legacy::cuda::_thnn_log_sigmoid_backward(), at::native::legacy::cuda::_thnn_log_sigmoid_backward_out(), at::native::legacy::cuda::_thnn_log_sigmoid_forward(), at::native::legacy::cuda::_thnn_log_sigmoid_forward_out(), at::native::legacy::cuda::_thnn_multi_margin_loss_backward(), at::native::legacy::cuda::_thnn_multi_margin_loss_backward_out(), at::native::legacy::cuda::_thnn_multi_margin_loss_forward(), at::native::legacy::cuda::_thnn_multi_margin_loss_forward_out(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_backward(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_backward_out(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_forward(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_forward_out(), at::native::legacy::cuda::_thnn_nll_loss2d_backward(), at::native::legacy::cuda::_thnn_nll_loss2d_backward_out(), at::native::legacy::cuda::_thnn_nll_loss2d_forward(), at::native::legacy::cuda::_thnn_nll_loss2d_forward_out(), at::native::legacy::cuda::_thnn_nll_loss_backward(), at::native::legacy::cuda::_thnn_nll_loss_backward_out(), at::native::legacy::cuda::_thnn_nll_loss_forward(), at::native::legacy::cuda::_thnn_nll_loss_forward_out(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward_(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward_out(), torch::autograd::InputBuffer::add(), device_of(), torch::autograd::dispatch_contiguous(), torch::autograd::dispatch_copy_(), torch::autograd::dispatch_invert(), torch::autograd::dispatch_nonzero(), torch::autograd::dispatch_nonzero_numpy(), torch::autograd::dispatch_range(), torch::autograd::dispatch_to_Bool(), torch::autograd::dispatch_to_CComplexDouble(), torch::autograd::dispatch_to_CDouble(), torch::autograd::dispatch_to_CLong(), torch::autograd::Engine::evaluate_function(), tools.codegen.dest.register_dispatch_key.RegisterDispatchKey::gen_unstructured(), at::Tensor::index(), at::Tensor::index_put_(), torch::autograd::THPVariable_getitem(), torch::autograd::THPVariable_new(), torch::autograd::THPVariable_new_ones(), torch::autograd::THPVariable_new_tensor(), and torch::autograd::THPVariable_setitem().

device_of() [2/2]

optional<Device> at::device_of ( TensorList  t )

inline

Return the Device of a TensorList, if the list is non-empty and the first Tensor is defined.

(This function implicitly assumes that all tensors in the list have the same device.)

Definition at line 28 of file DeviceGuard.h.

References device_of(), and c10::nullopt.

diagonal_backward_batching_rule()

Tensor at::diagonal_backward_batching_rule	(	const Tensor &	grad,
		IntArrayRef	input_sizes,
		int64_t	offset,
		int64_t	dim1,
		int64_t	dim2
	)

Definition at line 390 of file BatchingRegistrations.cpp.

References getGradInputPhysicalDim(), grad, at::MultiBatchVmapTransform::logicalToPhysical(), caffe2.perfkernels.hp_emblookup_codegen::offset, and caffe2::zeros.

Referenced by TORCH_LIBRARY_IMPL().

diagonal_batching_rule()

Tensor at::diagonal_batching_rule	(	const Tensor &	self,
		int64_t	offset,
		int64_t	dim1,
		int64_t	dim2
	)

Definition at line 382 of file BatchingRegistrations.cpp.

References at::native::diagonal(), at::MultiBatchVmapTransform::logicalToPhysical(), and caffe2.perfkernels.hp_emblookup_codegen::offset.

Referenced by TORCH_LIBRARY_IMPL().

dim_list_to_bitset()

static std::bitset<dim_bitset_size> at::dim_list_to_bitset ( IntArrayRef  dims, int64_t  ndims  )

inlinestatic

Definition at line 15 of file WrapDimUtilsMulti.h.

References dim, dim_bitset_size, caffe2::dims, caffe2::int64_t, maybe_wrap_dim(), and TORCH_CHECK.

Referenced by at::native::_sparse_sum(), at::native::_sparse_sum_backward_cpu(), at::native::_trilinear(), at::namedinference::compute_included_idxs(), at::native::flip_cpu(), at::native::make_dim_mask(), at::native::squeeze_multiple(), at::native::sumproduct_pair(), at::native::tensordot(), and torch::autograd::generated::details::unsqueeze_multiple().

dimname_to_position()

TORCH_API int64_t at::dimname_to_position	(	const Tensor &	tensor,
		Dimname	dim
	)

Definition at line 16 of file NamedTensorUtils.cpp.

References dim, at::Tensor::has_names(), caffe2::it, at::Tensor::names(), microbenchmarks::names, tensor, toDimnameRepr(), TORCH_CHECK, and WILDCARD.

Referenced by at::native::_sparse_log_softmax(), at::native::_sparse_softmax(), at::native::cat(), at::native::cat_out(), at::native::cummax(), at::native::cummax_out(), at::native::cummin(), at::native::cummin_out(), at::native::cumprod(), at::native::cumprod_(), at::native::cumprod_out(), at::native::cumsum(), at::native::cumsum_(), at::native::cumsum_out(), at::native::diagonal(), dimnames_to_positions(), at::native::flatten(), at::native::index_fill(), at::native::index_fill_(), at::native::kthvalue(), at::native::kthvalue_out(), at::native::log_softmax(), at::native::logcumsumexp(), at::native::logcumsumexp_out(), at::native::max(), at::native::max_out(), at::native::median(), at::native::median_out(), at::native::min(), at::native::min_out(), at::native::mode(), at::native::mode_out(), at::native::nanmedian(), at::native::nanmedian_out(), at::native::prod(), at::native::prod_out(), at::native::select(), at::native::size(), at::native::softmax(), at::native::squeeze(), at::native::stride(), at::native::transpose(), at::native::unbind(), and at::native::unflatten().

dimnames_to_positions()

TORCH_API std::vector< int64_t > at::dimnames_to_positions	(	const Tensor &	tensor,
		DimnameList	dims
	)

Definition at line 30 of file NamedTensorUtils.cpp.

References dimname_to_position(), caffe2::dims, name, and tensor.

Referenced by at::native::flatten(), at::native::logsumexp(), at::native::logsumexp_out(), at::native::mean(), at::native::mean_out(), at::native::mean_out_quantized_cpu(), at::native::mean_quantized_cpu(), at::native::norm(), at::native::norm_out(), at::native::std(), at::native::std_mean(), at::native::std_out(), at::native::sum(), at::native::sum_out(), at::native::var(), at::native::var_mean(), and at::native::var_out().

divup()

int64_t at::divup ( int64_t  x, int64_t  y  )

inline

Definition at line 8 of file Parallel.h.

References bench_ops::x, and at::native::metal::mpscnn::y.

Referenced by at::internal::calc_num_tasks_and_chunk_size(), at::native::PoolingParams1D::valid_output_end(), and at::native::PoolingParams1D::valid_output_start().

dot_batching_rule()

Tensor at::dot_batching_rule	(	const Tensor &	self,
		const Tensor &	other
	)

Definition at line 797 of file BatchingRegistrations.cpp.

References dim, at::Tensor::dim(), isBatchedTensor(), at::MultiBatchVmapTransform::logicalToPhysical(), torch.backends.cuda::matmul, at::meta::other, at::Tensor::sizes(), tensor, TORCH_CHECK, TORCH_INTERNAL_ASSERT, and at::native::unsqueeze().

Referenced by TORCH_LIBRARY_IMPL().

enableRecordFunction()

TORCH_API void at::enableRecordFunction

(

bool

enable

)

enableRecordFunction enables RecordFunction thread locally

Definition at line 399 of file record_function.cpp.

References torch.jit._state::enable().

Referenced by main(), at::RecordFunctionGuard::RecordFunctionGuard(), torch::jit::mobile::InterpreterState::run(), THPAutograd_initExtension(), and at::RecordFunctionGuard::~RecordFunctionGuard().

exp()

template<typename T >

C10_HOST_DEVICE T at::exp ( T  x )

inline

Definition at line 60 of file NumericUtils.h.

References exp(), value, and bench_ops::x.

Referenced by at::native::_kl_div_log_target(), exp(), exp< double >(), at::transformation::log_normal(), at::native::logsumexp_out_impl(), at::native::poisson_nll_loss(), at::native::soft_margin_loss_backward_out(), and TORCH_LIBRARY_IMPL().

exp< double >()

template<>

C10_HOST_DEVICE double at::exp< double > ( double  x )

inline

Definition at line 71 of file NumericUtils.h.

References exp(), and bench_ops::x.

expand_batching_rule()

Tensor at::expand_batching_rule	(	const Tensor &	self,
		IntArrayRef	size,
		bool	implicit
	)

Definition at line 144 of file BatchingRegistrations.cpp.

References c10::SmallVectorTemplateCommon< T, typename >::begin(), c10::aten::copy(), at::MultiBatchVmapTransform::logicalToPhysical(), size(), TORCH_CHECK, and TORCH_INTERNAL_ASSERT.

Referenced by TORCH_LIBRARY_IMPL().

expand_inplace() [1/4]

std::tuple<Tensor> at::expand_inplace ( const Tensor &  tensor, const Tensor &  to_expand  )

inline

Definition at line 48 of file ExpandUtils.h.

References c10::ArrayRef< T >::equals(), torch.jit._trace::make_tuple(), at::Tensor::sizes(), and tensor.

Referenced by at::native::templates::cuda::bernoulli_kernel(), at::indexing::copy_to(), expand_inplace(), torch::utils::map2_(), torch::utils::map_(), at::native::masked_fill__cuda(), and at::native::masked_scatter__cpu().

expand_inplace() [2/4]

std::tuple<Tensor> at::expand_inplace ( const Tensor &  tensor, const Tensor &  to_expand, const char *  api_name  )

inline

Definition at line 56 of file ExpandUtils.h.

References check_defined(), expand_inplace(), and tensor.

expand_inplace() [3/4]

std::tuple<Tensor, Tensor> at::expand_inplace ( const Tensor &  tensor, const Tensor &  to_expand1, const Tensor &  to_expand2  )

inline

Definition at line 61 of file ExpandUtils.h.

References c10::ArrayRef< T >::equals(), torch.jit._trace::make_tuple(), at::Tensor::sizes(), and tensor.

expand_inplace() [4/4]

std::tuple<Tensor, Tensor> at::expand_inplace ( const Tensor &  tensor, const Tensor &  to_expand1, const Tensor &  to_expand2, const char *  api_name  )

inline

Definition at line 71 of file ExpandUtils.h.

References check_defined(), expand_inplace(), and tensor.

expand_outplace() [1/5]

std::tuple<Tensor, Tensor> at::expand_outplace ( const Tensor &  to_expand1, const Tensor &  to_expand2  )

inline

Definition at line 77 of file ExpandUtils.h.

References c10::ArrayRef< T >::equals(), infer_size(), torch.jit._trace::make_tuple(), and at::Tensor::sizes().

Referenced by at::native::broadcast_tensors(), expand_outplace(), at::native::make_info(), at::native::masked_fill(), at::native::masked_scatter(), at::native::masked_select_out_impl_cpu(), and at::native::where().

expand_outplace() [2/5]

std::tuple<Tensor, Tensor> at::expand_outplace ( const Tensor &  to_expand1, const Tensor &  to_expand2, const char *  api_name  )

inline

Definition at line 88 of file ExpandUtils.h.

References check_defined(), and expand_outplace().

expand_outplace() [3/5]

std::tuple<Tensor, Tensor, Tensor> at::expand_outplace ( const Tensor &  to_expand1, const Tensor &  to_expand2, const Tensor &  to_expand3  )

inline

Definition at line 93 of file ExpandUtils.h.

References c10::ArrayRef< T >::equals(), infer_size(), torch.jit._trace::make_tuple(), and at::Tensor::sizes().

expand_outplace() [4/5]

std::tuple<Tensor, Tensor, Tensor> at::expand_outplace ( const Tensor &  to_expand1, const Tensor &  to_expand2, const Tensor &  to_expand3, const char *  api_name  )

inline

Definition at line 108 of file ExpandUtils.h.

References check_defined(), and expand_outplace().

expand_outplace() [5/5]

std::vector<Tensor> at::expand_outplace ( TensorList  to_expand )

inline

Definition at line 129 of file ExpandUtils.h.

References infer_size(), c10::ArrayRef< T >::size(), sizes, and c10::ArrayRef< T >::vec().

expand_size() [1/2]

std::tuple<Tensor> at::expand_size ( const Tensor &  to_expand, IntArrayRef  sizes  )

inline

Definition at line 116 of file ExpandUtils.h.

References c10::ArrayRef< T >::equals(), torch.jit._trace::make_tuple(), sizes, and at::Tensor::sizes().

Referenced by at::native::_sparse_sum_backward_cpu(), at::native::addbmm_out(), at::native::addmm_cpu_out(), at::native::addmm_out_sparse_dense_cpu(), at::native::addmm_sparse_dense_cpu(), at::native::baddbmm_out_cpu(), at::native::build_addr_iter(), and expand_size().

expand_size() [2/2]

std::tuple<Tensor> at::expand_size ( const Tensor &  to_expand, IntArrayRef  sizes, const char *  api_name  )

inline

Definition at line 124 of file ExpandUtils.h.

References check_defined(), expand_size(), and sizes.

fill_inplace_scalar_batching_rule()

Tensor& at::fill_inplace_scalar_batching_rule	(	Tensor &	self,
		Scalar	value
	)

Definition at line 236 of file BatchingRegistrations.cpp.

References at::MultiBatchVmapTransform::logicalToPhysical(), and value.

Referenced by TORCH_LIBRARY_IMPL().

fill_inplace_tensor_batching_rule()

Tensor& at::fill_inplace_tensor_batching_rule	(	Tensor &	self,
		const Tensor &	value
	)

Definition at line 242 of file BatchingRegistrations.cpp.

References isBatchedTensor(), at::MultiBatchVmapTransform::logicalToPhysical(), at::BroadcastingVmapTransform::logicalToPhysical(), tensor, and value.

Referenced by TORCH_LIBRARY_IMPL().

find_split_dim()

static int at::find_split_dim ( TensorIteratorBase &  iter )

static

Chooses a dimension over which to parallelize.

Prefers the outer-most dimension thats larger than the number of available threads.

Definition at line 70 of file TensorIteratorReduce.cpp.

References AT_ASSERT, dim, get_num_threads(), iter, and shape.

Referenced by at::TensorIteratorBase::foreach_reduced_elt(), and parallel_dim_reduction().

forward() [1/2]

void at::forward ( int64_t  offset )

inline

Definition at line 288 of file CPUApplyUtils.h.

Referenced by torch.distributed.nn.api.remote_module._RemoteModule::__init__(), torch::autograd::Function< T >::apply(), apply_op(), torch::python::bind_module(), torch.fx.graph.Graph::call_module(), torch.fx.symbolic_trace.Tracer::call_module(), torch.utils.checkpoint::checkpoint_sequential(), torch.quantization.fx.utils::collect_producer_nodes(), torch._jit_internal::export(), torch::onnx::export(), fastrnns.factory::flatten_list(), torch.jit._async::fork(), forward(), torch.jit._freeze::freeze(), operator_benchmark.benchmark_test_generator::generate_pt_tests_from_op_list(), torch::nn::AnyModule::get_(), audio_text_models::get_deepspeech(), vision_models::get_detr(), vision_models::get_fcn_resnet(), audio_text_models::get_multiheadattn(), torch.distributed.pipeline.sync.phony::get_phony(), vision_models::get_resnet18(), ppl_models::get_robust_regression(), ppl_models::get_simple_regression(), audio_text_models::get_transformer(), audio_text_models::get_wav2letter(), torch._jit_internal::ignore(), torch.distributed.pipeline.sync.checkpoint::is_recomputing(), torch::jit::isinstance(), fastrnns.factory::layernorm_pytorch_lstm_creator(), torch.fx.subgraph_rewriter::replace_pattern(), operator_benchmark.benchmark_caffe2.Caffe2OperatorTestCase::run_forward(), torch.jit._serialization::save(), torch.jit._script::script(), torch.distributed.pipeline.sync.skip.skippable::skippable(), torch.jit._trace::trace(), torch.jit._trace::trace_module(), torch._jit_internal::unused(), fastrnns.factory::varlen_pytorch_lstm_creator(), and torch::detail::wrap_pybind_function_impl_().

forward() [2/2]

template<typename Arg , typename... Args>

void at::forward ( int64_t  offset, Arg &  iter, Args &...  iter_tail  )

inline

Definition at line 291 of file CPUApplyUtils.h.

References forward(), caffe2::int64_t, iter, and caffe2.perfkernels.hp_emblookup_codegen::offset.

from_blob() [1/4]

Tensor at::from_blob ( void *  data, IntArrayRef  sizes, const std::function< void(void *)> &  deleter, const TensorOptions &  options = {}  )

inline

Definition at line 95 of file Functions.h.

from_blob() [2/4]

Tensor at::from_blob ( void *  data, IntArrayRef  sizes, const TensorOptions &  options = {}  )

inline

Definition at line 126 of file Functions.h.

from_blob() [3/4]

Tensor at::from_blob ( void *  data, IntArrayRef  sizes, IntArrayRef  strides, const std::function< void(void *)> &  deleter, const TensorOptions &  options = {}, const c10::optional< Device >  target_device = c10::nullopt  )

inline

Definition at line 69 of file Functions.h.

Referenced by torch::jit::tensorexpr::constructTensors(), caffe2::Workspace::CopyForwardedTensors(), at::sparse::flatten_indices(), torch::from_blob(), fromDLPack(), at::native::index_select_sparse(), parse_conv_serialized_state(), serialize_conv(), caffe2::ATenOp< Context >::tensorWrapping(), and PackedEmbeddingBagWeight::unpack().

from_blob() [4/4]

Tensor at::from_blob ( void *  data, IntArrayRef  sizes, IntArrayRef  strides, const TensorOptions &  options = {}  )

inline

Definition at line 103 of file Functions.h.

fromDLPack()

TORCH_API Tensor at::fromDLPack

(

const DLManagedTensor *

src

)

Definition at line 208 of file DLConvertor.cpp.

References deleter(), device, from_blob(), getATenDevice(), caffe2::src, and toScalarType().

Referenced by THPModule_fromDLPack().

geometry_is_contiguous()

TORCH_API bool at::geometry_is_contiguous	(	IntArrayRef	sizes,
		IntArrayRef	strides
	)

Definition at line 267 of file TensorUtils.cpp.

References dim, caffe2::int64_t, sizes, and strides.

Referenced by at::TensorGeometry::is_contiguous().

get_cxx_flags()

TORCH_API std::string at::get_cxx_flags

(

)

Definition at line 201 of file Version.cpp.

References caffe2::GetBuildOptions(), and TORCH_CHECK.

Referenced by THPModule_cxxFlags().

get_device()

int64_t at::get_device

(

Tensor

self

)

Referenced by torch.distributed.pipeline.sync.copy.Copy::backward(), torch.distributed.pipeline.sync.copy.Copy::forward(), torch.distributed.pipeline.sync.skip.portal.PortalCopy::forward(), torch.nn.parallel._functions.ReduceAddCoalesced::forward(), torch.nn.parallel._functions.Broadcast::forward(), torch::cuda::nccl::detail::get_communicators(), torch.nn.parallel.parallel_apply::parallel_apply(), and torch::autograd::THPVariable_get_device().

get_generator_or_default()

template<typename T >

static T* at::get_generator_or_default ( const c10::optional< Generator > &  gen, const Generator &  default_gen  )

inlinestatic

Utility function used in tensor implementations, which supplies the default generator to tensors, if an input generator is not supplied.

The input Generator* is also static casted to the backend generator type (CPU/CUDAGeneratorImpl etc.)

Definition at line 127 of file Utils.h.

References tools.autograd.gen_python_functions::gen().

get_mkl_version()

TORCH_API std::string at::get_mkl_version

(

)

Definition at line 21 of file Version.cpp.

References version.

Referenced by get_parallel_info(), and show_config().

get_mkldnn_version()

TORCH_API std::string at::get_mkldnn_version

(

)

Definition at line 37 of file Version.cpp.

Referenced by get_parallel_info(), and show_config().

get_num_interop_threads()

TORCH_API int at::get_num_interop_threads

(

)

Referenced by get_parallel_info(), main(), and THPModule_getNumInteropThreads().

get_num_threads()

TORCH_API int at::get_num_threads

(

)

Referenced by at::native::_cat_out_cpu(), at::internal::calc_num_tasks_and_chunk_size(), at::native::can_use_native_serial_stack(), find_split_dim(), at::TensorIteratorBase::for_each(), at::TensorIteratorBase::foreach_reduced_elt(), get_parallel_info(), main(), at::native::masked_select_out_impl_cpu(), at::TensorIteratorBase::parallel_reduce(), THPModule_getNumThreads(), and two_pass_reduction().

get_openmp_version()

TORCH_API std::string at::get_openmp_version

(

)

Definition at line 55 of file Version.cpp.

Referenced by get_parallel_info(), and show_config().

get_overlap_status() [1/2]

TORCH_API MemOverlapStatus at::get_overlap_status	(	const Tensor &	a,
		const Tensor &	b
	)

Definition at line 39 of file MemoryOverlap.cpp.

References caffe2.contrib.aten.docs.sample::a, at::native::metal::mpscnn::b, and get_overlap_status().

Referenced by at::native::_cat_out_cpu(), assert_no_overlap(), assert_no_partial_overlap(), at::native::can_use_native_serial_stack(), and get_overlap_status().

get_overlap_status() [2/2]

TORCH_API MemOverlapStatus at::get_overlap_status	(	TensorImpl *	a,
		TensorImpl *	b
	)

Definition at line 43 of file MemoryOverlap.cpp.

References caffe2.contrib.aten.docs.sample::a, at::native::metal::mpscnn::b, FULL, NO, PARTIAL, and TOO_HARD.

get_parallel_info()

std::string at::get_parallel_info

(

)

Definition at line 45 of file ParallelCommon.cpp.

References get_mkl_version(), get_mkldnn_version(), get_num_interop_threads(), get_num_threads(), and get_openmp_version().

Referenced by torch::throughput_benchmark::detail::BenchmarkHelper< Input, Output, Model >::benchmark(), main(), and THPModule_parallelInfo().

get_qtensorimpl()

TORCH_API QTensorImpl * at::get_qtensorimpl

(

const Tensor &

self

)

Definition at line 72 of file Quantizer.cpp.

References at::native::is_quantized(), fastrnns.scratch::requires_grad, TORCH_CHECK, and TORCH_INTERNAL_ASSERT.

Referenced by at::native::alias_with_sizes_and_strides(), at::native::as_strided_qtensorimpl(), at::native::dequantize_quant(), at::native::equal_quantized_cpu(), new_qtensor(), at::native::q_per_channel_axis(), at::native::q_per_channel_scales(), at::native::q_per_channel_zero_points(), at::native::q_scale_quant(), at::native::q_zero_point_quant(), at::native::qscheme_quant(), at::Tensor::quantizer(), set_quantizer_(), at::native::squeeze_qtensor(), and at::native::unsqueeze_qtensor().

get_record_function_tls_()

TORCH_API const RecordFunctionTLS & at::get_record_function_tls_

(

)

Definition at line 58 of file record_function.cpp.

Referenced by at::ThreadLocalState::ThreadLocalState().

get_sub_byte_tensor_size()

int64_t at::get_sub_byte_tensor_size	(	int64_t	size_bytes,
		at::ScalarType	t
	)

Definition at line 80 of file Quantizer.cpp.

References std::ceil(), and caffe2::int64_t.

Referenced by new_qtensor().

get_thread_num()

TORCH_API int at::get_thread_num

(

)

getATenDevice()

static Device at::getATenDevice ( const DLContext &  ctx )

static

Definition at line 96 of file DLConvertor.cpp.

References c10::CPU, c10::CUDA, torch.distributed.pipeline.sync._balance::Device, DLContext::device_id, DLContext::device_type, c10::HIP, kDLCPU, kDLGPU, kDLOpenCL, kDLROCM, caffe2::OPENCL, and std::to_string().

Referenced by fromDLPack().

getCPUAllocator()

TORCH_API Allocator * at::getCPUAllocator

(

)

Definition at line 235 of file Context.cpp.

References c10::GetCPUAllocator().

Referenced by at::detail::empty_cpu(), new_qtensor(), and torch::distributed::rpc::wireDeserialize().

getDeprecatedTypeProperties()

static DeprecatedTypeProperties& at::getDeprecatedTypeProperties ( Backend  p, ScalarType  s  )

inlinestatic

Definition at line 242 of file Context.h.

References at::DeprecatedTypePropertiesRegistry::getDeprecatedTypeProperties(), globalDeprecatedTypePropertiesRegistry(), and diagnose_protobuf::p.

Referenced by torch::autograd::VariableType::allTypesForBackends(), torch::jit::python::operator<<(), and torch::utils::options_from_string().

getDLContext()

TORCH_API DLContext at::getDLContext	(	const Tensor &	tensor,
		const int64_t &	device_id
	)

Definition at line 68 of file DLConvertor.cpp.

References c10::CPU, c10::CUDA, at::Tensor::device(), DLContext::device_id, DLContext::device_type, c10::HIP, kDLCPU, kDLGPU, kDLOpenCL, kDLROCM, caffe2::OPENCL, c10::Device::str(), tensor, and c10::Device::type().

Referenced by toDLPack().

getDLDataType()

TORCH_API DLDataType at::getDLDataType

(

const Tensor &

t

)

Definition at line 10 of file DLConvertor.cpp.

References c10::BFloat16, c10::Bool, c10::ComplexDouble, c10::Double, caffe2::dtype, at::Tensor::element_size(), c10::Float, c10::Half, c10::Int, kDLFloat, kDLInt, kDLUInt, at::Tensor::scalar_type(), and c10::Undefined.

Referenced by toDLPack().

getGradInputPhysicalDim()

static int64_t at::getGradInputPhysicalDim ( int64_t  dim, IntArrayRef  input_sizes, int64_t  num_batch_dims  )

static

Definition at line 350 of file BatchingRegistrations.cpp.

References dim, maybe_wrap_dim(), at::namedinference::num_batch_dims(), and c10::ArrayRef< T >::size().

Referenced by diagonal_backward_batching_rule(), select_backward_batching_rule(), and slice_backward_batching_rule().

getLevelsAndLargestLogicalDim()

static std::pair<std::bitset<kVmapNumLevels>,int64_t> at::getLevelsAndLargestLogicalDim ( TensorList  logical_tensors )

static

Definition at line 229 of file VmapTransforms.cpp.

References createVmapLevelsBitset(), at::Tensor::dim(), caffe2::int64_t, maybeGetBatchedImpl(), c10::ArrayRef< T >::size(), tensor, and TORCH_INTERNAL_ASSERT.

Referenced by at::BroadcastingVmapTransform::logicalToPhysical().

getNumGPUs()

static size_t at::getNumGPUs ( )

inlinestatic

Definition at line 275 of file Context.h.

References at::detail::getCUDAHooks(), at::detail::getHIPHooks(), at::CUDAHooksInterface::getNumGPUs(), at::HIPHooksInterface::getNumGPUs(), hasCUDA(), and hasHIP().

getPhysicalTensorAndLevels()

static std::pair<Tensor,std::bitset<kVmapNumLevels> > at::getPhysicalTensorAndLevels ( const Tensor &  self )

static

Definition at line 98 of file VmapTransforms.cpp.

References createVmapLevelsBitset(), maybeGetBatchedImpl(), and permuteBatchDimsToFront().

Referenced by alignBatchDimsAtFront().

globalContext()

TORCH_API Context & at::globalContext

(

)

Definition at line 31 of file Context.cpp.

Referenced by at::native::_choose_qparams_per_tensor(), at::native::_convolution(), at::native::legacy::cuda::_th_copy_ignoring_overlaps_(), at::native::legacy::cuda::_th_cross_kernel(), at::native::legacy::cuda::_th_cross_kernel_out(), at::native::legacy::cuda::_th_gels(), at::native::legacy::cuda::_th_gels_out(), at::native::legacy::cuda::_th_geqrf(), at::native::legacy::cuda::_th_geqrf_out(), at::native::legacy::cuda::_th_index_copy_(), at::native::legacy::cuda::_th_masked_fill_(), at::native::legacy::cuda::_th_masked_fill_bool_(), at::native::legacy::cuda::_th_mode(), at::native::legacy::cuda::_th_mode_out(), at::native::legacy::cuda::_th_put_(), at::native::legacy::cuda::_th_renorm(), at::native::legacy::cuda::_th_renorm_(), at::native::legacy::cuda::_th_renorm_out(), at::native::legacy::cuda::_th_sort(), at::native::legacy::cuda::_th_sort_out(), at::native::legacy::cuda::_th_topk(), at::native::legacy::cuda::_th_topk_out(), at::native::legacy::cuda::_thnn_conv2d_backward(), at::native::legacy::cuda::_thnn_conv2d_backward_out(), at::native::legacy::cuda::_thnn_conv2d_forward(), at::native::legacy::cuda::_thnn_conv2d_forward_out(), at::native::legacy::cuda::_thnn_conv_depthwise2d_backward(), at::native::legacy::cuda::_thnn_conv_depthwise2d_backward_out(), at::native::legacy::cuda::_thnn_conv_depthwise2d_forward(), at::native::legacy::cuda::_thnn_conv_depthwise2d_forward_out(), at::native::legacy::cuda::_thnn_glu_backward(), at::native::legacy::cuda::_thnn_glu_backward_out(), at::native::legacy::cuda::_thnn_glu_forward(), at::native::legacy::cuda::_thnn_glu_forward_out(), at::native::legacy::cuda::_thnn_log_sigmoid_backward(), at::native::legacy::cuda::_thnn_log_sigmoid_backward_out(), at::native::legacy::cuda::_thnn_log_sigmoid_forward(), at::native::legacy::cuda::_thnn_log_sigmoid_forward_out(), at::native::legacy::cuda::_thnn_multi_margin_loss_backward(), at::native::legacy::cuda::_thnn_multi_margin_loss_backward_out(), at::native::legacy::cuda::_thnn_multi_margin_loss_forward(), at::native::legacy::cuda::_thnn_multi_margin_loss_forward_out(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_backward(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_backward_out(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_forward(), at::native::legacy::cuda::_thnn_multilabel_margin_loss_forward_out(), at::native::legacy::cuda::_thnn_nll_loss2d_backward(), at::native::legacy::cuda::_thnn_nll_loss2d_backward_out(), at::native::legacy::cuda::_thnn_nll_loss2d_forward(), at::native::legacy::cuda::_thnn_nll_loss2d_forward_out(), at::native::legacy::cuda::_thnn_nll_loss_backward(), at::native::legacy::cuda::_thnn_nll_loss_backward_out(), at::native::legacy::cuda::_thnn_nll_loss_forward(), at::native::legacy::cuda::_thnn_nll_loss_forward_out(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward_(), at::native::legacy::cuda::_thnn_rrelu_with_noise_forward_out(), at::native::adaptive_avg_pool2d_quantized_cpu(), at::native::adaptive_avg_pool3d_out_quantized_cpu(), at::Context::alertNotDeterministic(), torch::autograd::VariableType::allCUDATypes(), at::native::avg_pool2d_quantized_cpu(), at::cuda::blas::bgemm< at::Half >(), at::cuda::blas::bgemm< c10::complex< double > >(), at::cuda::blas::bgemm< c10::complex< float > >(), at::cuda::blas::bgemm< double >(), at::cuda::blas::bgemm< float >(), torch::jit::tensorexpr::CudaCodeGen::CompileToNVRTC(), at::native::convolution(), at::native::cudnn_convolution_deprecated2(), at::native::cudnn_convolution_transpose_deprecated2(), at::native::cudnn_is_acceptable(), deserialize_conv(), torch::jit::fuser::cuda::FusedKernelCUDA::FusedKernelCUDA(), at::cuda::blas::gemm< at::Half >(), at::cuda::blas::gemm< c10::complex< double > >(), at::cuda::blas::gemm< c10::complex< float > >(), at::cuda::blas::gemm< double >(), at::cuda::blas::gemm< float >(), at::cuda::blas::gemv< c10::complex< double > >(), at::cuda::blas::gemv< c10::complex< float > >(), at::cuda::blas::gemv< double >(), at::cuda::blas::gemv< float >(), tools.codegen.dest.register_dispatch_key.RegisterDispatchKey::gen_unstructured(), at::cuda::getCurrentCUDABlasHandle(), at::cuda::getPinnedMemoryAllocator(), at::native::hardsigmoid_quantized_cpu(), hasCUDA(), hasHIP(), hasLAPACK(), hasMAGMA(), hasMKL(), hasMKLDNN(), hasOpenMP(), hasXLA(), at::native::index_copy_(), init(), isPinned(), manual_seed(), at::native::mean_out_quantized_cpu(), new_qtensor(), torch::jit::fuser::cuda::nvrtc(), torch::jit::tensorexpr::nvrtc(), torch::jit::fuser::cuda::executor_utils::nvrtcCompile(), pytorch_jni::PytorchJni::preModuleLoadSetupOnce(), at::native::quantized_hardswish(), at::native::quantized_max_pool2d(), register_linear_params(), torch::jit::RegisterCudaFuseGraph::registerPass(), at::native::relu_quantized_cpu(), torch::jit::fuser::cuda::FusionExecutor::runFusion(), at::native::sigmoid_quantized_cpu(), at::native::tanh_quantized_cpu(), THCPModule_initExtension(), THPModule_allowTF32CuBLAS(), THPModule_allowTF32CuDNN(), THPModule_are_vmap_fallback_warnings_enabled(), THPModule_benchmarkCuDNN(), THPModule_deterministicAlgorithms(), THPModule_deterministicCuDNN(), THPModule_isEnabledXNNPACK(), THPModule_qEngine(), THPModule_set_display_vmap_fallback_warnings_mode(), THPModule_setAllowTF32CuBLAS(), THPModule_setAllowTF32CuDNN(), THPModule_setBenchmarkCuDNN(), THPModule_setDeterministicAlgorithms(), THPModule_setDeterministicCuDNN(), THPModule_setFlushDenormal(), THPModule_setQEngine(), THPModule_setUserEnabledCuDNN(), THPModule_setUserEnabledMkldnn(), THPModule_supportedQEngines(), THPModule_userEnabledCuDNN(), THPModule_userEnabledMkldnn(), torch::jit::to_dispatch(), at::native::ConvParams::use_mkldnn(), and warnFallback().

globalDeprecatedTypePropertiesRegistry()

TORCH_API DeprecatedTypePropertiesRegistry & at::globalDeprecatedTypePropertiesRegistry

(

)

Definition at line 28 of file DeprecatedTypePropertiesRegistry.cpp.

Referenced by at::Tensor::C10_DEPRECATED_MESSAGE(), CPU(), CUDA(), getDeprecatedTypeProperties(), HIP(), at::DeprecatedTypeProperties::toBackend(), and at::DeprecatedTypeProperties::toScalarType().

has_internal_overlap() [1/2]

TORCH_API MemOverlap at::has_internal_overlap

(

const Tensor &

tensor

)

Definition at line 6 of file MemoryOverlap.cpp.

References has_internal_overlap(), tensor, and at::Tensor::unsafeGetTensorImpl().

Referenced by at::native::_debug_has_internal_overlap(), at::native::_index_put_impl_(), assert_no_internal_overlap(), has_internal_overlap(), at::native::index_fill_(), at::native::masked_fill__cuda(), and at::native::masked_fill_impl_cpu().

has_internal_overlap() [2/2]

TORCH_API MemOverlap at::has_internal_overlap

(

TensorImpl *

t

)

Definition at line 10 of file MemoryOverlap.cpp.

References AT_ASSERT, at::Tensor::is_contiguous(), c10::kStrided, at::Tensor::layout(), NO, sizes, at::Tensor::sizes(), strides, at::Tensor::strides(), TOO_HARD, and YES.

has_names()

bool at::has_names ( TensorList  tensors )

inline

Definition at line 13 of file NamedTensorUtils.h.

References caffe2::tensors.

Referenced by at::namedinference::compute_broadcast_outnames(), at::namedinference::compute_cat_outnames(), at::namedinference::compute_cdist_outnames(), at::namedinference::compute_matmul_outnames(), at::namedinference::propagate_names_for_expand(), at::native::resize_(), at::native::resize_named_tensor_(), at::native::select(), and at::native::unflatten().

hasCallbacks()

TORCH_API bool at::hasCallbacks

(

)

Definition at line 354 of file record_function.cpp.

References matmul_dlmc_bench::m.

Referenced by torch::jit::InterpreterStateImpl::checkAndStartRecordFunction().

hasCUDA()

static bool at::hasCUDA ( )

inlinestatic

Definition at line 262 of file Context.h.

References globalContext(), and at::Context::hasCUDA().

Referenced by getNumGPUs(), manual_seed(), torch::jit::RegisterCudaFuseGraph::registerPass(), and show_config().

hasGlobalCallbacks()

TORCH_API bool at::hasGlobalCallbacks

(

)

hasGlobalCallbacks returns whether there're global callbacks registered with pushGlobalCallback

Definition at line 359 of file record_function.cpp.

Referenced by torch::jit::mobile::InterpreterState::run(), and shouldRunRecordFunction().

hasHIP()

static bool at::hasHIP ( )

inlinestatic

Definition at line 266 of file Context.h.

References globalContext(), and at::Context::hasHIP().

Referenced by getNumGPUs(), and torch::jit::RegisterCudaFuseGraph::registerPass().

hasLAPACK()

static bool at::hasLAPACK ( )

inlinestatic

Definition at line 302 of file Context.h.

References globalContext(), and at::Context::hasLAPACK().

Referenced by initModule().

hasMAGMA()

static bool at::hasMAGMA ( )

inlinestatic

Definition at line 306 of file Context.h.

References globalContext(), and at::Context::hasMAGMA().

Referenced by THCPModule_initExtension().

hasMKL()

static bool at::hasMKL ( )

inlinestatic

Definition at line 298 of file Context.h.

References globalContext(), and at::Context::hasMKL().

Referenced by at::native::bmm_out_or_baddbmm_(), and initModule().

hasMKLDNN()

static bool at::hasMKLDNN ( )

inlinestatic

Definition at line 310 of file Context.h.

References globalContext(), and at::Context::hasMKLDNN().

Referenced by initModule().

hasOpenMP()

static bool at::hasOpenMP ( )

inlinestatic

Definition at line 294 of file Context.h.

References globalContext(), and at::Context::hasOpenMP().

Referenced by initModule().

hasThreadLocalCallbacks()

TORCH_API bool at::hasThreadLocalCallbacks

(

)

hasThreadLocalCallbacks returns whether there're callbacks registered with addThreadLocalCallback

Definition at line 363 of file record_function.cpp.

hasXLA()

static bool at::hasXLA ( )

inlinestatic

Definition at line 270 of file Context.h.

References globalContext(), and at::Context::hasXLA().

HIP()

static DeprecatedTypeProperties& at::HIP ( ScalarType  s )

inlinestatic

Definition at line 257 of file Context.h.

References at::DeprecatedTypePropertiesRegistry::getDeprecatedTypeProperties(), globalDeprecatedTypePropertiesRegistry(), and c10::HIP.

in_parallel_region()

TORCH_API bool at::in_parallel_region

(

)

Referenced by at::TensorIteratorBase::foreach_reduced_elt(), parallel_for(), parallel_reduce(), and at::TensorIteratorBase::parallel_reduce().

infer_dense_strides()

TORCH_API std::vector< int64_t > at::infer_dense_strides	(	IntArrayRef	tensor_sizes,
		IntArrayRef	tensor_strides
	)

Definition at line 103 of file ExpandUtils.cpp.

References matmul_dlmc_bench::comparison, c10::attr::idx(), caffe2::int64_t, c10::attr::perm(), c10::ArrayRef< T >::size(), std::swap(), and TORCH_CHECK.

Referenced by denseAndNonOverlapping(), and at::native::empty_like().

infer_size() [1/2]

TORCH_API std::vector< int64_t > at::infer_size	(	IntArrayRef	a,
		IntArrayRef	b
	)

Definition at line 6 of file ExpandUtils.cpp.

References caffe2.contrib.aten.docs.sample::a, at::native::metal::mpscnn::b, caffe2::int64_t, caffe2.perfkernels.hp_emblookup_codegen::offset, and TORCH_CHECK.

Referenced by at::native::_linalg_broadcast_batch_dims(), at::native::cdist_impl(), at::TensorIteratorBase::compute_shape(), expand_outplace(), torch::jit::fuser::getMapSize(), at::native::masked_select_backward(), at::native::matmul(), at::native::reshape(), at::native::vulkan::detail::reshape_copy(), at::native::reshape_out(), THPModule_inferSize(), at::native::view(), and at::native::vulkan::aten::view().

infer_size() [2/2]

std::vector<int64_t> at::infer_size ( IntArrayRef  shape, int64_t  numel  )

inline

Definition at line 12 of file InferSize.h.

References AT_ERROR, dim, caffe2::int64_t, numel(), caffe2::python::res, shape, c10::SmallVectorTemplateCommon< T, typename >::size(), and TORCH_CHECK.

inferExpandGeometry()

TORCH_API std::tuple< std::vector< int64_t >, std::vector< int64_t > > at::inferExpandGeometry	(	IntArrayRef	tensor_sizes,
		IntArrayRef	tensor_strides,
		IntArrayRef	sizes
	)

Definition at line 33 of file ExpandUtils.cpp.

References dim, caffe2::int64_t, caffe2.perfkernels.hp_emblookup_codegen::offset, c10::ArrayRef< T >::size(), size(), sizes, stride(), and TORCH_CHECK.

Referenced by at::native::expand().

init()

static void at::init ( void  )

inlinestatic

Definition at line 236 of file Context.h.

References globalContext().

Referenced by initModule(), at::internal::lazy_init_num_threads(), and at::native::Descriptor< T, ctor, dtor >::mut_desc().

init_num_threads()

TORCH_API void at::init_num_threads

(

)

Referenced by initModule(), at::internal::lazy_init_num_threads(), main(), at::PTThreadPool::PTThreadPool(), torch::autograd::Engine::reentrant_thread_init(), and torch::autograd::Engine::thread_init().

initialTensorOptions()

TensorOptions at::initialTensorOptions ( )

inline

Definition at line 10 of file InitialTensorOptions.h.

References c10::TensorOptions::dtype(), c10::kCPU, c10::kFloat, c10::kStrided, c10::TensorOptions::layout(), and c10::TensorOptions::requires_grad().

Referenced by at::native::hspmm_out_sparse_cpu(), and torch::jit::InferShapeTypeForUninitializedOutput().

inplaceIsVmapCompatible()

TORCH_API bool at::inplaceIsVmapCompatible	(	const Tensor &	self,
		const Tensor &	other
	)

Definition at line 132 of file BatchedTensorImpl.cpp.

References createVmapLevelsBitset(), maybeGetBatchedImpl(), and at::meta::other.

Referenced by torch::autograd::AccumulateGrad::accumulateGrad(), torch::autograd::generated::details::binary_cross_entropy_double_backward(), torch::autograd::generated::details::binary_cross_entropy_double_backward_grad_output(), and torch::autograd::generated::details::symeig_backward().

internal_set_names_inplace() [1/2]

TORCH_API Tensor & at::internal_set_names_inplace	(	Tensor &	tensor,
		optional< DimnameList >	names
	)

Definition at line 19 of file NamedTensor.cpp.

References at::impl::internal_set_names_inplace(), microbenchmarks::names, tensor, and at::Tensor::unsafeGetTensorImpl().

Referenced by at::native::align(), at::native::align_to(), at::native::empty(), at::native::flatten(), at::native::refine_names(), at::native::rename(), at::native::rename_(), THPVariable_set_names(), and at::native::unflatten().

internal_set_names_inplace() [2/2]

TORCH_API Tensor & at::internal_set_names_inplace	(	Tensor &	tensor,
		std::vector< Dimname > &&	names,
		bool	validate_names
	)

Definition at line 24 of file NamedTensor.cpp.

References at::impl::internal_set_names_inplace(), microbenchmarks::names, tensor, and at::Tensor::unsafeGetTensorImpl().

intraop_default_num_threads()

int at::intraop_default_num_threads

(

)

Definition at line 94 of file ParallelCommon.cpp.

References c10::TaskThreadPoolBase::defaultNumThreads(), and TORCH_CHECK.

intraop_invoke()

template<typename F0 , typename F1 >

void at::intraop_invoke	(	const F0 &	f0,
		const F1 &	f1
	)

Definition at line 103 of file ParallelNativeTBB.h.

intraop_launch()

TORCH_API void at::intraop_launch

(

std::function< void()>

func

)

intraop_launch_future()

TORCH_API std::shared_ptr<c10::ivalue::Future> at::intraop_launch_future

(

std::function< void()>

func

)

is_allowed_dim_on_scalar_tensor()

static bool at::is_allowed_dim_on_scalar_tensor ( int64_t  dim )

static

Definition at line 53 of file BatchingRegistrations.cpp.

References dim.

Referenced by sum_batching_rule(), and transpose_int_batching_rule().

is_custom_op()

bool at::is_custom_op

(

const c10::OperatorName &

opName

)

Definition at line 28 of file ATenOpList.cpp.

References $, torch::jit::aten_ops, c10::OperatorName::name, ops, and c10::OperatorName::overload_name.

is_expandable_to()

static bool at::is_expandable_to ( IntArrayRef  shape, IntArrayRef  desired  )

inlinestatic

Definition at line 181 of file ExpandUtils.h.

References caffe2::int64_t, shape, c10::ArrayRef< T >::size(), c10::SmallVectorTemplateCommon< T, typename >::size(), size(), and caffe2::target().

Referenced by at::native::make_index_put_iterator(), at::native::sum_to_size(), and torch::autograd::validate_outputs().

isBatchedTensor()

bool at::isBatchedTensor ( const Tensor &  tensor )

inline

Definition at line 99 of file BatchedTensorImpl.h.

References c10::Batched, c10::DispatchKeySet::has(), c10::TensorImpl::key_set(), tensor, and at::Tensor::unsafeGetTensorImpl().

Referenced by dot_batching_rule(), fill_inplace_tensor_batching_rule(), makeBatched(), maybeGetBatchedImpl(), mm_batching_rule(), mv_batching_rule(), THPEngine_run_backward(), and at::VmapPhysicalView::VmapPhysicalView().

isInplaceOp()

static bool at::isInplaceOp ( const c10::FunctionSchema &  schema )

static

Definition at line 46 of file BatchedFallback.cpp.

References caffe2::it.

Referenced by batchedTensorForLoopFallback().

isPhysicalScalarTensor()

bool at::isPhysicalScalarTensor

(

const Tensor &

logical_tensor

)

Definition at line 73 of file BatchingRegistrations.cpp.

References at::Tensor::dim(), and maybeGetBatchedImpl().

Referenced by binary_pointwise_batching_rule().

isRecordFunctionEnabled()

TORCH_API bool at::isRecordFunctionEnabled

(

)

isRecordFunctionEnabled returns whether RecordFunction is enabled thread locally

Definition at line 395 of file record_function.cpp.

Referenced by torch::jit::mobile::InterpreterState::run().

iterate() [1/2]

void at::iterate ( int64_t  size )

inline

Definition at line 241 of file CPUApplyUtils.h.

Referenced by apply_op(), and iterate().

iterate() [2/2]

template<typename Arg , typename... Args>

void at::iterate ( int64_t  size, Arg &  iter, Args &...  iter_tail  )

inline

Definition at line 244 of file CPUApplyUtils.h.

References iter, iterate(), and size().

iterate_continue() [1/2]

bool at::iterate_continue ( )

inline

Definition at line 250 of file CPUApplyUtils.h.

Referenced by apply_op(), and iterate_continue().

iterate_continue() [2/2]

template<typename Arg , typename... Args>

bool at::iterate_continue ( Arg &  iter, Args &...  iter_tail  )

inline

Definition at line 255 of file CPUApplyUtils.h.

References iter, and iterate_continue().

iterate_overflow() [1/2]

void at::iterate_overflow ( )

inline

Definition at line 271 of file CPUApplyUtils.h.

Referenced by apply_op(), and iterate_overflow().

iterate_overflow() [2/2]

template<typename Arg , typename... Args>

void at::iterate_overflow ( Arg &  iter, Args &...  iter_tail  )

inline

Definition at line 274 of file CPUApplyUtils.h.

References caffe2::int64_t, iter, and iterate_overflow().

launch()

TORCH_API void at::launch

(

std::function< void()>

func

)

Referenced by _launch_tasks_tree(), torch::distributed::autograd::DistEngine::globalCpuThread(), launch_tasks(), and torch::distributed::autograd::DistEngine::runEngineAndAccumulateGradients().

legacy_cat_wrap_dim() [1/2]

static int64_t at::legacy_cat_wrap_dim ( int64_t  dim, const std::vector< std::vector< int64_t > > &  tensor_sizes  )

inlinestatic

Definition at line 63 of file WrapDimUtils.h.

References dim, maybe_wrap_dim(), and sizes.

Referenced by at::native::cat(), at::native::cat_out(), and torch::autograd::generated::details::cat_tensors_backward().

legacy_cat_wrap_dim() [2/2]

static int64_t at::legacy_cat_wrap_dim ( int64_t  dim, TensorList  tensors  )

inlinestatic

Definition at line 73 of file WrapDimUtils.h.

References dim, at::Tensor::dim(), maybe_wrap_dim(), at::Tensor::sizes(), tensor, and caffe2::tensors.

legacyExtractDispatchKey()

static DispatchKey at::legacyExtractDispatchKey ( const Tensor &  t )

inlinestatic

Definition at line 805 of file TensorBody.h.

References at::Tensor::key_set(), and legacyExtractDispatchKey().

Referenced by torch::utils::_sparse_coo_tensor_unsafe_ctor(), torch::utils::_validate_sparse_coo_tensor_args(), torch::autograd::applySlicing(), at::Tensor::C10_DEPRECATED_MESSAGE(), legacyExtractDispatchKey(), torch::utils::sparse_coo_tensor_ctor(), torch::tensors::Tensor_instancecheck(), torch::autograd::THPVariable_new(), torch::autograd::THPVariable_new_ones(), and torch::autograd::THPVariable_new_tensor().

log()

template<typename T >

C10_HOST_DEVICE T at::log ( T  x )

inline

Definition at line 76 of file NumericUtils.h.

References log(), value, and bench_ops::x.

Referenced by at::native::_kl_div_non_log_target(), at::transformation::exponential(), at::transformation::geometric(), log(), log< double >(), log_nnc_fast(), log_nnc_sleef(), at::native::poisson_nll_loss(), and TORCH_LIBRARY_IMPL().

log< double >()

template<>

C10_HOST_DEVICE double at::log< double > ( double  x )

inline

Definition at line 87 of file NumericUtils.h.

References log(), and bench_ops::x.

make_per_channel_affine_quantizer()

TORCH_API QuantizerPtr at::make_per_channel_affine_quantizer	(	const Tensor &	scales,
		const Tensor &	zero_points,
		int64_t	axis,
		ScalarType	scalar_type
	)

Definition at line 44 of file Quantizer.cpp.

References caffe2::axis, at::Tensor::contiguous(), c10::isFloatingType(), c10::kDouble, c10::kFloat, c10::kLong, at::Tensor::scalar_type(), detail::scalar_type(), and TORCH_CHECK.

Referenced by at::native::empty_per_channel_affine_quantized(), at::native::quantize_per_channel_cpu(), at::native::squeeze_qtensor(), and at::native::unsqueeze_qtensor().

make_per_tensor_affine_quantizer()

TORCH_API QuantizerPtr at::make_per_tensor_affine_quantizer	(	double	scale,
		int64_t	zero_point,
		ScalarType	scalar_type
	)

Definition at line 36 of file Quantizer.cpp.

References detail::scalar_type(), scale, and zero_point.

Referenced by at::native::empty_affine_quantized(), and at::native::quantize_per_tensor().

makeBatched()

TORCH_API Tensor at::makeBatched	(	const Tensor &	tensor,
		BatchDims	bdims
	)

Definition at line 105 of file BatchedTensorImpl.cpp.

References c10::SmallVectorTemplateCommon< T, typename >::begin(), at::Tensor::dim(), c10::SmallVectorTemplateCommon< T, typename >::end(), isBatchedTensor(), kVmapMaxTensorDims, kVmapNumLevels, tensor, TORCH_CHECK, and TORCH_INTERNAL_ASSERT.

Referenced by addBatchDim(), at::VmapPhysicalToLogicalMap::apply(), clone_batching_rule(), pow_scalar_Tensor_batching_rule(), at::native::remove_existing_batch_dim(), to_dtype_layout_batching_rule(), unwrap_and_call(), and unwrap_and_call_method().

manual_seed()

static void at::manual_seed ( uint64_t  seed )

inlinestatic

Definition at line 314 of file Context.h.

References c10::CPU, at::Context::defaultGenerator(), torch.distributed.pipeline.sync._balance::Device, tools.autograd.gen_python_functions::gen(), at::detail::getCUDAHooks(), at::CUDAHooksInterface::getNumGPUs(), globalContext(), hasCUDA(), c10::kCUDA, lock, and torch.cuda.random::seed().

Referenced by torch::jit::Node::hasSideEffects(), and torch.utils.data.dataset::random_split().

max_dim() [1/2]

int64_t at::max_dim ( )

inline

Definition at line 302 of file CPUApplyUtils.h.

Referenced by apply_op(), and max_dim().

max_dim() [2/2]

template<typename Arg , typename... Args>

int64_t at::max_dim ( Arg &  iter, Args &...  iter_tail  )

inline

Definition at line 307 of file CPUApplyUtils.h.

References iter, std::max(), and max_dim().

max_iterate_size() [1/2]

int64_t at::max_iterate_size ( )

inline

Definition at line 260 of file CPUApplyUtils.h.

References max.

Referenced by max_iterate_size().

max_iterate_size() [2/2]

template<typename Arg , typename... Args>

int64_t at::max_iterate_size ( Arg &  iter, Args &...  iter_tail  )

inline

Definition at line 265 of file CPUApplyUtils.h.

References iter, max_iterate_size(), and std::min().

maximum_indexable_location()

static optional<int64_t> at::maximum_indexable_location ( IntArrayRef  sizes, IntArrayRef  strides, int64_t  storage_offset  )

static

Definition at line 493 of file BatchingRegistrations.cpp.

References c10::nullopt, sizes, at::native::storage_size_for(), and strides.

Referenced by checkBasicAsStridedValidForSlice().

maybe_data_ptr() [1/2]

TORCH_API void * at::maybe_data_ptr

(

const Tensor &

tensor

)

Definition at line 258 of file TensorUtils.cpp.

References at::Tensor::data_ptr(), at::Tensor::defined(), and tensor.

maybe_data_ptr() [2/2]

TORCH_API void * at::maybe_data_ptr

(

const TensorArg &

tensor

)

Definition at line 262 of file TensorUtils.cpp.

References at::Tensor::data_ptr(), at::Tensor::defined(), and tensor.

maybe_wrap_dim() [1/4]

static int64_t at::maybe_wrap_dim ( int64_t  dim, const std::vector< std::vector< int64_t > > &  tensor_sizes  )

inlinestatic

Definition at line 25 of file WrapDimUtils.h.

References dim, maybe_wrap_dim(), and size().

maybe_wrap_dim() [2/4]

static int64_t at::maybe_wrap_dim ( int64_t  dim, int64_t  dim_post_expr, bool  wrap_scalar = true  )

inlinestatic

Definition at line 9 of file WrapDimUtils.h.

References dim, and c10::maybe_wrap_dim().

Referenced by at::native::_allocate_or_resize_output_with_indices(), at::native::_aminmax_out_impl(), at::native::_reduction_with_indices_allocate_or_resize_output(), torch::autograd::generated::details::_safe_size(), at::native::_stack(), at::native::_stack_cpu(), at::BatchedTensorImpl::actualDim(), at::native::all_out(), at::native::any_out(), at::native::argmax_out(), at::native::argmin_out(), at::native::cat_sparse(), at::native::count_nonzero(), at::native::cross_out(), at::native::cummax_out(), at::native::cummin_out(), at::native::cumprod_backward(), at::native::default_alldims(), at::native::diag_embed(), at::native::diagonal(), at::native::diff_check_compatible_shape(), dim_list_to_bitset(), torch::autograd::generated::details::fft_r2c_backward(), at::native::flatten(), torch::cuda::gather(), torch::cuda::gather_out(), getGradInputPhysicalDim(), torch::jit::fuser::getMapSize(), torch::jit::getNativeOperation(), at::VmapPhysicalView::getPhysicalDim(), at::VmapPhysicalView::getPhysicalDims(), torch::jit::tensorexpr::TensorExprKernel::getReductionInfo(), at::native::glu_backward_out(), at::native::glu_out(), at::native::index_add_cpu_(), at::native::index_copy_(), at::native::index_fill_(), at::native::index_select_out_cpu_(), at::native::index_select_sparse(), legacy_cat_wrap_dim(), at::native::log_softmax_backward_cpu(), at::native::log_softmax_cpu(), at::native::max_out_impl(), maybe_wrap_dim(), at::native::min_out_impl(), at::native::mode_out(), at::native::movedim(), at::native::narrow(), at::native::narrow_copy_dense_cpu_out(), at::native::permute(), torch::autograd::generated::details::permute_backwards(), torch::autograd::generated::details::prod_backward(), at::namedinference::propagate_names_except(), at::native::quantized_topk_out_cpu(), torch::cuda::scatter(), torch::cuda::scatter_out(), at::native::select(), c10::TensorImpl::size(), at::TensorGeometry::size(), at::native::slice(), at::native::softmax_backward_cpu(), at::native::softmax_backward_sparse_input_preprocessing(), at::native::softmax_cpu(), torch::autograd::generated::details::split_backward(), torch::autograd::generated::details::split_with_sizes_backward(), at::native::squeeze(), at::native::squeeze_(), at::native::stack(), stack_batching_rule(), at::native::stack_out(), c10::TensorImpl::stride(), at::MetalTensorImpl< OpaqueHandle >::stride(), at::VulkanOpaqueTensorImpl< OpaqueHandle >::stride(), at::TensorGeometry::stride(), at::native::tensor_split(), at::namedinference::TensorNames::TensorNames(), at::native::topk_out_cpu(), at::native::transpose(), at::native::transpose_(), at::native::trapz(), at::native::unbind(), at::native::unflatten(), at::native::unfold(), at::native::unsqueeze(), at::native::unsqueeze_(), unsqueeze_batching_rule(), at::native::unsqueeze_qtensor(), torch::autograd::generated::details::unsqueeze_to(), and wrap_all_dims().

maybe_wrap_dim() [3/4]

static int64_t at::maybe_wrap_dim ( int64_t  dim, TensorImpl *  tensor  )

inlinestatic

Definition at line 13 of file WrapDimUtils.h.

References dim, at::Tensor::dim(), maybe_wrap_dim(), and tensor.

maybe_wrap_dim() [4/4]

static int64_t at::maybe_wrap_dim ( int64_t  dim, TensorList  tensors  )

inlinestatic

Definition at line 17 of file WrapDimUtils.h.

References dim, maybe_wrap_dim(), and caffe2::tensors.

maybe_wrap_dims()

template<typename Container >

void at::maybe_wrap_dims ( Container &  dims, int64_t  dim_post_expr  )

inline

Definition at line 54 of file WrapDimUtils.h.

References caffe2::dims, and maybe_wrap_dims_n().

Referenced by at::native::_fft_fill_with_conjugate_symmetry_(), at::native::_linalg_norm_matrix_out(), at::native::_linalg_norm_vector_out(), at::native::_sparse_sum(), at::native::_sparse_sum_backward_cpu(), at::native::frobenius_norm_out(), at::native::norm_sparse(), and at::native::nuclear_norm_out().

maybe_wrap_dims_n()

static void at::maybe_wrap_dims_n ( int64_t *  dims, int64_t  ndims, int64_t  dim_post_expr  )

inlinestatic

Definition at line 34 of file WrapDimUtils.h.

References dim, caffe2::dims, caffe2::int64_t, max, min, and TORCH_CHECK_INDEX.

Referenced by maybe_wrap_dims().

maybeGetBatchedImpl()

BatchedTensorImpl* at::maybeGetBatchedImpl ( Tensor  tensor )

inline

Definition at line 109 of file BatchedTensorImpl.h.

References isBatchedTensor(), tensor, and unsafeGetBatchedImpl().

Referenced by at::native::_remove_batch_dim(), addBatchDim(), batchedTensorForLoopFallback(), batchedTensorInplaceForLoopFallback(), getLevelsAndLargestLogicalDim(), getPhysicalTensorAndLevels(), at::native::has_level(), inplaceIsVmapCompatible(), isPhysicalScalarTensor(), and at::MultiBatchVmapTransform::logicalToPhysical().

mm_batching_rule()

Tensor at::mm_batching_rule	(	const Tensor &	self,
		const Tensor &	other
	)

Definition at line 844 of file BatchingRegistrations.cpp.

References dim, at::Tensor::dim(), isBatchedTensor(), at::MultiBatchVmapTransform::logicalToPhysical(), torch.backends.cuda::matmul, at::meta::other, at::Tensor::sizes(), tensor, TORCH_CHECK, and TORCH_INTERNAL_ASSERT.

Referenced by TORCH_LIBRARY_IMPL().

movedim_batching_rule()

Tensor at::movedim_batching_rule	(	const Tensor &	self,
		IntArrayRef	source,
		IntArrayRef	destination
	)

Definition at line 399 of file BatchingRegistrations.cpp.

References at::MultiBatchVmapTransform::logicalToPhysical(), and c10::aten::movedim().

Referenced by TORCH_LIBRARY_IMPL().

mv_batching_rule()

Tensor at::mv_batching_rule	(	const Tensor &	self,
		const Tensor &	other
	)

Definition at line 760 of file BatchingRegistrations.cpp.

References dim, at::Tensor::dim(), isBatchedTensor(), at::MultiBatchVmapTransform::logicalToPhysical(), torch.backends.cuda::matmul, at::meta::other, at::Tensor::sizes(), tensor, TORCH_CHECK, TORCH_INTERNAL_ASSERT, and at::native::unsqueeze().

Referenced by TORCH_LIBRARY_IMPL().

new_empty_batching_rule()

Tensor at::new_empty_batching_rule	(	const Tensor &	self,
		IntArrayRef	size,
		c10::optional< ScalarType >	dtype,
		c10::optional< Layout >	layout,
		c10::optional< Device >	device,
		c10::optional< bool >	pin_memory
	)

Definition at line 937 of file BatchingRegistrations.cpp.

References device, caffe2::dtype, caffe2::layout, at::MultiBatchVmapTransform::logicalToPhysical(), at::native::pin_memory(), and size().

Referenced by TORCH_LIBRARY_IMPL().

new_empty_strided_batching_rule()

Tensor at::new_empty_strided_batching_rule	(	const Tensor &	self,
		IntArrayRef	size,
		IntArrayRef	stride,
		optional< ScalarType >	dtype,
		optional< Layout >	layout,
		optional< Device >	device,
		optional< bool >	pin_memory
	)

Definition at line 950 of file BatchingRegistrations.cpp.

References at::detail::defaultStrides(), device, caffe2::dtype, caffe2::layout, at::MultiBatchVmapTransform::logicalToPhysical(), at::native::pin_memory(), size(), at::native::storage_size_for(), stride(), and TORCH_CHECK.

Referenced by TORCH_LIBRARY_IMPL().

new_qtensor()

TORCH_API Tensor at::new_qtensor ( IntArrayRef  sizes, const TensorOptions &  options, QuantizerPtr  quantizer  )

inline

Definition at line 92 of file Quantizer.cpp.

References allocator, check_size_nonnegative(), c10::CUDA, caffe2::dtype, c10::TensorImpl::empty_tensor_restride(), get_qtensorimpl(), get_sub_byte_tensor_size(), getCPUAllocator(), at::CUDAHooksInterface::getCUDADeviceAllocator(), at::detail::getCUDAHooks(), c10::GetDefaultMobileCPUAllocator(), globalContext(), caffe2::int64_t, c10::isQIntType(), c10::memory_format(), caffe2.perfkernels.hp_emblookup_codegen::options, prod_intlist(), detail::scalar_type(), c10::TensorImpl::set_sizes_contiguous(), sizes, caffe2::storage, tensor, TORCH_CHECK, and c10::typeMetaToScalarType().

Referenced by at::native::empty_affine_quantized(), at::native::empty_per_channel_affine_quantized(), at::PerTensorAffineQuantizer::quantize(), at::PerChannelAffineQuantizer::quantize(), and at::PerChannelAffineFloatQParamsQuantizer::quantize().

new_zeros_batching_rule()

Tensor at::new_zeros_batching_rule	(	const Tensor &	self,
		IntArrayRef	size,
		optional< ScalarType >	dtype,
		optional< Layout >	layout,
		optional< Device >	device,
		optional< bool >	pin_memory
	)

Definition at line 919 of file BatchingRegistrations.cpp.

References c10::TensorOptions::device(), device, c10::TensorOptions::dtype(), caffe2::dtype, c10::TensorOptions::layout(), caffe2::layout, at::MultiBatchVmapTransform::logicalToPhysical(), caffe2.perfkernels.hp_emblookup_codegen::options, at::native::pin_memory(), c10::TensorOptions::pinned_memory(), and size().

Referenced by TORCH_LIBRARY_IMPL().

numel()

int64_t at::numel ( const Tensor &  tensor )

inline

Definition at line 133 of file Functions.h.

References at::Tensor::numel(), and tensor.

Referenced by _all_equal_numel(), apply_op(), at::native::calculate_quant_loss(), checkNumel(), at::native::choose_qparams_optimized(), at::detail::computeStride(), at::native::dot(), at::native::dot_check(), at::native::equal_quantized_cpu(), infer_size(), and at::native::vdot().

operator!=() [1/4]

bool at::operator!= ( const Dimname &  lhs, const Dimname &  rhs  )

inline

Definition at line 44 of file Dimname.h.

operator!=() [2/4]

static Tensor at::operator!= ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator!=() [3/4]

static Tensor at::operator!= ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator!=() [4/4]

static Tensor at::operator!= ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator%() [1/3]

static Tensor at::operator% ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator%() [2/3]

static Tensor at::operator% ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator%() [3/3]

static Tensor at::operator% ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator&() [1/3]

static Tensor at::operator& ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator&() [2/3]

static Tensor at::operator& ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator&() [3/3]

static Tensor at::operator& ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator*() [1/3]

static Tensor at::operator* ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator*() [2/3]

static Tensor at::operator* ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator*() [3/3]

static Tensor at::operator* ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator+() [1/3]

static Tensor at::operator+ ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator+() [2/3]

static Tensor at::operator+ ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator+() [3/3]

static Tensor at::operator+ ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator-() [1/3]

static Tensor at::operator- ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator-() [2/3]

static Tensor at::operator- ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator-() [3/3]

static Tensor at::operator- ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator/() [1/3]

static Tensor at::operator/ ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator/() [2/3]

static Tensor at::operator/ ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator/() [3/3]

static Tensor at::operator/ ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator<() [1/3]

static Tensor at::operator< ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator<() [2/3]

static Tensor at::operator< ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator<() [3/3]

static Tensor at::operator< ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator<<() [1/7]

std::ostream& at::operator<< ( std::ostream &  out, const BatchDim &  bdim  )

inline

Definition at line 134 of file BatchedTensorImpl.h.

References at::BatchDim::dim(), at::BatchDim::level(), and out.

operator<<() [2/7]

TORCH_API std::ostream & at::operator<<	(	std::ostream &	out,
		const DeprecatedTypeProperties &	t
	)

Definition at line 36 of file Formatting.cpp.

References out, and at::Tensor::toString().

operator<<() [3/7]

TORCH_API std::ostream & at::operator<<	(	std::ostream &	out,
		const Dimname &	dimname
	)

Definition at line 9 of file Dimname.cpp.

References c10::namespaces::dimname(), out, c10::Symbol::toUnqualString(), and WILDCARD.

operator<<() [4/7]

std::ostream & at::operator<<	(	std::ostream &	out,
		const Range &	range
	)

Definition at line 7 of file Range.cpp.

References out, and c10::prim::range().

operator<<() [5/7]

static std::ostream& at::operator<< ( std::ostream &  out, const Tensor &  t  )

inlinestatic

Definition at line 18 of file Formatting.h.

References out, and print().

operator<<() [6/7]

static std::ostream& at::operator<< ( std::ostream &  out, Scalar  s  )

inlinestatic

Definition at line 25 of file Formatting.h.

References out.

operator<<() [7/7]

TORCH_API std::ostream & at::operator<<	(	std::ostream &	out,
		TensorGeometryArg	t
	)

Definition at line 11 of file TensorUtils.cpp.

References at::Tensor::name(), and out.

operator<=() [1/3]

static Tensor at::operator<= ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator<=() [2/3]

static Tensor at::operator<= ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator<=() [3/3]

static Tensor at::operator<= ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator==() [1/4]

bool at::operator== ( const Dimname &  lhs, const Dimname &  rhs  )

inline

Definition at line 40 of file Dimname.h.

References at::Dimname::symbol().

operator==() [2/4]

static Tensor at::operator== ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator==() [3/4]

static Tensor at::operator== ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator==() [4/4]

static Tensor at::operator== ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator>() [1/3]

static Tensor at::operator> ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator>() [2/3]

static Tensor at::operator> ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator>() [3/3]

static Tensor at::operator> ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator>=() [1/3]

static Tensor at::operator>= ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator>=() [2/3]

static Tensor at::operator>= ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator>=() [3/3]

static Tensor at::operator>= ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator^() [1/3]

static Tensor at::operator^ ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator^() [2/3]

static Tensor at::operator^ ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator^() [3/3]

static Tensor at::operator^ ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator|() [1/3]

static Tensor at::operator| ( const Tensor &  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator|() [2/3]

static Tensor at::operator| ( const Tensor &  x, Scalar  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

operator|() [3/3]

static Tensor at::operator| ( Scalar  x, const Tensor &  y  )

inlinestatic

Definition at line 110 of file TensorOperators.h.

parallel_dim_reduction()

static void at::parallel_dim_reduction ( TensorIteratorBase &  iter, loop2d_t  loop  )

static

Definition at line 98 of file TensorIteratorReduce.cpp.

References AT_ASSERT, dim, matmul_dlmc_bench::end, find_split_dim(), caffe2::int64_t, iter, caffe2.python.helpers.control_ops::loop(), parallel_for(), and round_columns().

Referenced by at::TensorIteratorBase::parallel_reduce().

parallel_for()

template<class F >

void at::parallel_for ( const int64_t  begin, const int64_t  end, const int64_t  grain_size, const F &  f  )

inline

Definition at line 34 of file ParallelNative.h.

References at::internal::_parallel_run(), matmul_dlmc_bench::end, in_parallel_region(), caffe2::int64_t, torch.cuda.profiler::start(), and TORCH_CHECK.

Referenced by at::native::_embedding_bag_dense_backward_cpu_sum_mean(), at::native::_grid_sampler_2d_cpu_fallback(), at::native::_grid_sampler_2d_cpu_fallback_backward(), _launch_tasks_tree(), at::native::_sparse_sum_backward_cpu(), at::native::add_dense_sparse_worker_cpu(), at::native::apply_triu_tril(), at::native::apply_triu_tril_single(), at::native::baddbmm_cpu_kernel(), at::native::batch_norm_backward_cpu_template(), at::native::batch_norm_cpu_transform_input_template(), at::native::batch_norm_cpu_update_stats_template(), compute_cpu(), at::sparse::coo_to_csr(), at::native::copy_impl(), at::native::dim_apply(), caffe2::RMSNormOp< Context >::DoRunWithType(), at::native::flip_cpu_kernel(), at::TensorIteratorBase::for_each(), at::TensorIteratorBase::foreach_reduced_elt(), at::native::index_select_out_cpu_(), torch::nn::parallel::parallel_apply(), parallel_dim_reduction(), at::native::prelu_cpu_backward_kernel_multi_weights(), at::native::prelu_cpu_kernel_multi_weights(), at::native::prelu_cpu_kernel_share_weights(), PackedEmbeddingBagWeight::prepack(), at::native::range_cpu_out(), caffe2::RMSNormGradientOp< Context >::RMSNormBackward(), at::native::slow_conv2d_forward_out_cpu(), at::native::slow_conv3d_forward_out_cpu(), at::native::sparse_mask_out_cpu_kernel(), two_pass_reduction(), and PackedEmbeddingBagWeight::unpack().

parallel_reduce()

template<class scalar_t , class F , class SF >

scalar_t at::parallel_reduce ( const int64_t  begin, const int64_t  end, const int64_t  grain_size, const scalar_t  ident, const F &  f, const SF &  sf  )

inline

Definition at line 58 of file ParallelNative.h.

References at::internal::_parallel_run(), at::internal::calc_num_tasks_and_chunk_size(), matmul_dlmc_bench::end, in_parallel_region(), caffe2::int64_t, compare-fastrnn-results::results, scalar_t, torch.cuda.profiler::start(), and TORCH_CHECK.

Referenced by at::native::prelu_cpu_backward_kernel_share_weights().

permute_batching_rule()

Tensor at::permute_batching_rule	(	const Tensor &	self,
		IntArrayRef	dims
	)

Definition at line 326 of file BatchingRegistrations.cpp.

References caffe2::dims, c10::SmallVectorTemplateCommon< T, typename >::end(), c10::SmallVectorImpl< T >::insert(), caffe2::int64_t, at::MultiBatchVmapTransform::logicalToPhysical(), c10::SmallVectorTemplateBase< T, isPodLike >::push_back(), and c10::SmallVectorImpl< T >::reserve().

Referenced by TORCH_LIBRARY_IMPL().

permuteBatchDimsToFront()

static Tensor at::permuteBatchDimsToFront ( BatchedTensorImpl *  batched )

static

Definition at line 18 of file VmapTransforms.cpp.

References areBdimsAtFrontInOrder(), at::BatchedTensorImpl::bdims(), createBatchDimBitset(), c10::attr::idx(), caffe2::int64_t, ptr, c10::SmallVectorImpl< T >::reserve(), sizes, at::Tensor::sizes(), and at::BatchedTensorImpl::value().

Referenced by getPhysicalTensorAndLevels(), and at::MultiBatchVmapTransform::logicalToPhysical().

pow_scalar_Tensor_batching_rule()

Tensor at::pow_scalar_Tensor_batching_rule	(	Scalar	other,
		const Tensor &	self
	)

Definition at line 711 of file BatchingRegistrations.cpp.

References makeBatched(), at::meta::other, at::native::pow(), and unsafeGetBatchedImpl().

Referenced by TORCH_LIBRARY_IMPL().

print() [1/2]

static void at::print ( const Tensor &  t, int64_t  linesize = 80  )

inlinestatic

Definition at line 21 of file Formatting.h.

References print().

print() [2/2]

TORCH_API std::ostream & at::print	(	std::ostream &	stream,
		const Tensor &	tensor_,
		int64_t	linesize
	)

Definition at line 230 of file Formatting.cpp.

References __printFormat(), __printMatrix(), __printTensor(), at::Tensor::contiguous(), at::Tensor::data_ptr(), defaultfloat(), at::Tensor::defined(), at::Tensor::fw_grad(), caffe2::int64_t, at::Tensor::is_quantized(), at::Tensor::is_sparse(), c10::kCPU, c10::kDouble, c10::kPerChannelAffine, c10::kPerChannelAffineFloatQParams, c10::kPerTensorAffine, at::Tensor::ndimension(), at::Tensor::numel(), print(), printScale(), scale, at::Tensor::size(), at::Tensor::sizes(), stream, sz, tensor, at::Tensor::toString(), and c10::toString().

Referenced by torch.utils.benchmark.utils.valgrind_wrapper.timer_interface._ValgrindWrapper::__init__(), torchaudio_models.ScaledDotProduct::__init__(), torchaudio_models.MultiheadAttentionContainer::__init__(), torch.futures.Future::_add_done_callback(), update-caffe2-models.SomeClass::_download(), torch.nn.parallel.distributed::_dump_DDP_relevant_env_vars(), torch.utils.cpp_extension::_find_cuda_home(), torch.utils.cpp_extension::_find_rocm_home(), torch.utils.cpp_extension::_get_build_directory(), torch.utils.cpp_extension::_get_num_workers(), caffe2.python.onnx.backend.Caffe2Backend::_graph_to_net(), torch.utils.cpp_extension::_jit_compile(), torch.functional::_lu_impl(), operator_benchmark.benchmark_core.BenchmarkRunner::_measure_time(), torch.onnx.utils::_model_to_graph(), tools.nightly::_move_single(), tools.nightly::_nightly_version(), caffe2.python.onnx.backend.Caffe2Backend::_onnx_model_to_caffe2_net(), torch::nn::functional::_pad(), torch::nn::functional::_pad_circular(), torch.utils.cpp_extension::_prepare_ldflags(), update-caffe2-models.SomeClass::_prepare_model_data(), caffe2.python.rnn.rnn_cell_test_util::_prepare_rnn(), benchmark_ddp_rpc::_print_cont(), operator_benchmark.benchmark_core.BenchmarkRunner::_print_header(), operator_benchmark.benchmark_pytorch.PyTorchOperatorTestCase::_print_per_iter(), operator_benchmark.benchmark_core.BenchmarkRunner::_print_perf_result(), operator_benchmark.benchmark_core.BenchmarkRunner::_print_test_case_info(), torch.optim.lr_scheduler.ReduceLROnPlateau::_reduce_lr(), torch.distributed.rpc.internal::_run_function(), torch.jit._trace::_time(), caffe2.python.parallel_workers.WorkerCoordinator::_wait_finish(), torch.utils.cpp_extension::_write_ninja_file_and_build_library(), torch.utils.cpp_extension::_write_ninja_file_and_compile_objects(), torch.utils.tensorboard.writer.SummaryWriter::add_embedding(), torch.nn.modules.module.Module::apply(), torch.distributed.rpc.functions::async_execution(), torch.utils.tensorboard.summary::audio(), caffe2.experiments.python.convnet_benchmarks::Benchmark(), caffe2.python.convnet_benchmarks::Benchmark(), caffe2.contrib.nccl.nccl_ops_test::benchmark(), caffe2.contrib.nnpack.nnpack_ops_test::benchmark(), concat_benchmark::benchmark_concat(), utils::benchmark_module(), framework_overhead_benchmark::benchmark_simple_fn(), sparse_lengths_sum_nbit_benchmark::benchmark_sparse_lengths_sum(), sparse_normalize_benchmark::benchmark_sparse_normalize(), utils::benchmark_using_throughput_benchmark(), torch.nn.modules.module.Module::buffers(), caffe2.python.onnx.helper::c2_native_run_net(), update-models-from-caffe2::caffe2_to_onnx(), caffe2.python.device_checker.DeviceChecker::CheckNet(), caffe2.python.gradient_checker.GradientChecker::CheckSimple(), caffe2.python.device_checker.DeviceChecker::CheckSimple(), update-caffe2-models::cleanup(), torch.package.exporter.PackageExporter::close(), torch.futures::collect_all(), caffe2.python.rnn.lstm_comparison::Compare(), torch.quantization._numeric_suite::compare_model_outputs(), torch.quantization._numeric_suite_fx::compare_model_outputs_fx(), torch.quantization._numeric_suite::compare_model_stub(), torch.quantization._numeric_suite_fx::compare_model_stub_fx(), caffe2.python.lengths_reducer_fused_8bit_rowwise_ops_test::compare_rowwise(), torch.quantization._numeric_suite::compare_weights(), torch.quantization._numeric_suite_fx::compare_weights_fx(), torch.utils.hipify.hipify_python::compute_stats(), caffe2.python.lstm_benchmark::create_model(), torch._jit_internal::createResolutionCallbackFromFrame(), torch.nn.utils.prune::custom_from_mask(), caffe2.python.helpers.array_helpers::depth_concat(), caffe2.python.cnn.CNNModelHelper::DepthConcat(), caffe2.experiments.python.device_reduce_sum_bench.Benchmark::display(), tools.download_mnist::download(), update-models-from-caffe2::download_caffe2_model(), tools.clang_format_utils::download_clang_format(), update-caffe2-models::download_models(), update-models-from-caffe2::download_onnx_model(), caffe2.python.models.download::downloadFromURLToFile(), caffe2.python.models.download::downloadModel(), benchmark.Benchmark::dump_result(), package.tool.clang_coverage::export(), fast_nvcc::fast_nvcc_warn(), caffe2.python.helpers.fc::fc_sparse(), tools.clang_format_all::file_clang_formatted_correctly(), tools.clang_tidy::filter_files(), caffe2.python.mkl.rewrite_graph::fix_BoxWithNMSLimit(), torch.jit._freeze::freeze(), caffe2.python.fused_8bit_rowwise_conversion_ops_test::fused_rowwise_8bit_quantize_reference(), gen_vulkan_spv::genCppH(), gen_vulkan_glsl::genCppH(), tools.setup_helpers.cmake.CMake::generate(), caffe2.python.lstm_benchmark::generate_data(), update-caffe2-models::generate_models(), caffe2.python.docs.generator.OperatorDoc::generateInterface(), tools.clang_format_utils::get_and_check_clang_format(), caffe2.python.binarysize::GetSymbolTrie(), torch.nn.utils.prune::global_unstructured(), torch.utils.tensorboard._pytorch_graph::graph(), categorize.Categorizer::handle_commit(), torch.hub::help(), torch.utils.hipify.hipify_python::hipify(), torch.nn.utils.prune::identity(), caffe2.python.dyndep::InitOpsLibrary(), process::insert(), torch.nn.utils.prune::is_pruned(), torch::jit::isinstance(), package.tool.print_report::line_oriented_report(), torch.jit._serialization::load(), matmul_dlmc_bench::load_dataset(), caffe2.python.parallel_workers.Metrics::log_metrics(), tools.nightly::logging_manager(), benchmark::main(), diff::main(), launcher::main(), framework_overhead_benchmark::main(), compare::main(), functional_autograd_benchmark::main(), bench::main(), __main__::main(), caffe2.python.net_drawer::main(), caffe2.python.numa_benchmark::main(), tools.clang_tidy::main(), tools.download_mnist::main(), torch.distributed.launch::main(), torch.utils.bottleneck.__main__::main(), torch.utils.collect_env::main(), bench_gen::main(), caffe2.python.benchmark_generator::main(), caffe2.python.mint.app::main(), fused_rowwise_nbit_conversion_bench::main(), torch.utils.tensorboard.summary::make_video(), package.tool.clang_coverage::merge(), torch.nn.modules.module.Module::modules(), freeze.Freezer::msg(), torch.nn.modules.module.Module::named_buffers(), torch.nn.modules.module.Module::named_children(), torch.nn.modules.module.Module::named_modules(), torch.nn.modules.module.Module::named_parameters(), caffe2.experiments.python.convnet_benchmarks::net_DAG_Builder(), caffe2.experiments.python.SparseTransformer::netbuilder(), fast_nvcc::nvcc_dryrun_data(), torch.quantization._learnable_fake_quantize._LearnableFakeQuantize::observe_quant_params(), operator<<(), c10::operator<<(), caffe2.python.ideep.transform_ideep_net::Optimize(), torch.nn.modules.module.Module::parameters(), torch.utils.tensorboard._onnx_graph::parse(), torch.jit._state.EnabledProxy::parse_env(), package.tool.summarize_jsons::parse_json(), common::parse_pr_number(), torch.utils.hipify.hipify_python::preprocess(), torch.utils.hipify.hipify_python::preprocess_file_and_save_result(), torch.utils.hipify.hipify_python::preprocessor(), print(), torch.utils.bottleneck.__main__::print_autograd_prof_summary(), launcher::print_benchmark_results(), setup::print_box(), torch.utils.bottleneck.__main__::print_cprofile_summary(), fast_nvcc::print_dot_graph(), package.util.utils::print_error(), package.tool.print_report::print_file_oriented_report(), package.tool.print_report::print_file_summary(), fastrnns.bench::print_json_oss_format(), fastrnns.bench::print_json_pep_format(), package.util.utils::print_log(), torch.optim.lr_scheduler._LRScheduler::print_lr(), framework_overhead_benchmark::print_results(), fastrnns.bench::print_stderr(), torch.fx.graph.Graph::print_tabular(), package.tool.print_report::print_test_by_type(), package.tool.print_report::print_test_condition(), package.util.utils::print_time(), fast_nvcc::print_verbose_output(), torch.utils.tensorboard._pytorch_graph.GraphPy::printall(), caffe2.python.binarysize::PrintTrie(), torch.nn.utils.prune::random_structured(), caffe2.python.lazy_dyndep::RegisterOpsLibrary(), caffe2.python.nomnigraph::render(), setup::report(), torch.package.exporter.PackageExporter::require_module(), namespace_check::run(), caffe2.python.utils.DebugMode::run(), operator_benchmark.benchmark_core.BenchmarkRunner::run(), caffe2.python.timeout_guard.WatcherThread::run(), tools.setup_helpers.cmake.CMake::run(), torch.utils.bottleneck.__main__::run_autograd_prof(), record_function_bench::run_bench(), microbenchmarks::run_benchmarks(), tools.clang_format_all::run_clang_format(), tools.clang_format_all::run_clang_format_on_file(), coordinator.CoordinatorBase::run_coordinator(), torch.utils.bottleneck.__main__::run_cprofile(), torch.utils.bottleneck.__main__::run_env_analysis(), caffe2.python.models.seq2seq.translate::run_seq2seq_beam_decoder(), tools.clang_tidy::run_shell_command(), caffe2.python.muji_test.TestMuji::RunningAllreduceWithGPUs(), torch.package.exporter.PackageExporter::save_pickle(), torch.package.exporter.PackageExporter::save_source_string(), torch.jit._script::script(), generate_test_torchscripts::scriptAndSave(), torch.distributed.rpc.options.TensorPipeRpcBackendOptions::set_device_map(), operator_benchmark.benchmark_utils::set_mkl_threads(), operator_benchmark.benchmark_utils::set_omp_threads(), torch.futures.Future::set_result(), caffe2.python.models.download::signalHandler(), caffe2.python.workspace::StartImmediate(), caffe2.python.workspace::StartMint(), caffe2.contrib.aten.gen_op::supports(), benchmark::sweep(), fastrnns.profile::system(), caffe2.python.ideep.concat_split_op_test.TestConcatSplitOps::test_concat_with_TensorCPU(), caffe2.python.nomnigraph_test.TestBindings::test_convertToProto(), caffe2.python.ideep.convfusion_op_test.ConvFusionTest::test_convolution_affch_folding(), caffe2.python.ideep.convfusion_op_test.ConvFusionTest::test_convolution_bn_folding(), caffe2.python.ideep.convfusion_op_test.ConvFusionTest::test_convolution_grouped_sum_relu_fusion(), caffe2.python.ideep.convfusion_op_test.ConvFusionTest::test_convolution_relu_fusion(), caffe2.python.ideep.convfusion_op_test.ConvFusionTest::test_convolution_sum_fusion(), caffe2.python.ideep.convfusion_op_test.ConvFusionTest::test_convolution_sum_relu_fusion(), caffe2.python.hypothesis_test.TestOperators::test_dag_net_forking(), caffe2.python.ideep.conv_op_test.ConvTest::test_depthwise_convolution(), caffe2.python.ideep.elementwise_sum_op_test.ElementwiseSumTest::test_elementwise_sum_fallback(), caffe2.python.ideep.expanddims_squeeze_op_test.ExpandDimsSqueezeTest::test_expand_dims_fallback(), caffe2.python.ideep.fc_op_test.FcTest::test_fc_with_axis(), caffe2.python.ideep.fc_op_test.FcTest::test_fc_with_axis_w(), caffe2.python.models.imagenet_trainer_test_utils::test_forward_only_fast_simplenet(), caffe2.python.memonger_test.MemongerTest::test_forward_optim_tree_harder(), caffe2.quantization.server.fully_connected_fp16_test.FullyConnectedFP16Test::test_fully_connected(), caffe2.python.ideep.operator_fallback_op_test.TestFallbackOps::test_in_place(), caffe2.python.ideep.elementwise_sum_op_test.ElementwiseSumTest::test_int8_elementwise_sum(), caffe2.python.ideep.fc_op_test.FcTest::test_int8_fc_4_dims(), caffe2.python.ideep.pool_op_test.PoolTest::test_int8_pooling(), caffe2.python.ideep.relu_op_test.ReluTest::test_int8_relu(), caffe2.python.nomnigraph_test.TestBindings::test_match_graph(), caffe2.python.predictor.predictor_exporter_test.PredictorExporterTest::test_meta_net_def_net_runs(), caffe2.python.mkl.rewrite_graph_test.MKLRewriteTest::test_mkl_multi_output_rewrite(), caffe2.python.ideep.order_switch_op_test.OrderSwitchTest::test_nhwc2nchw(), caffe2.python.observer_test.TestObservers::test_observer_rnn_executor(), caffe2.python.optimizer_test.TestOptimizerContext::test_optimizer_context(), caffe2.python.ideep.pre_convert_test.PreConvertTest::test_preConvert(), caffe2.contrib.nnpack.nnpack_ops_test.NNPackOpsTest::test_relu_timings(), caffe2.python.ideep.shape_op_test.ShapeTest::test_shape(), caffe2.python.ideep.shape_op_test.ShapeTest::test_shape_with_axes(), caffe2.python.lengths_reducer_fused_8bit_rowwise_ops_test.TestLengthsReducerOpsFused8BitRowwise::test_sparse_lengths_mean(), caffe2.python.sparse_to_dense_test.TestSparseToDense::test_sparse_to_dense(), caffe2.python.hypothesis_test.TestOperators::test_sparse_to_dense(), caffe2.python.sparse_to_dense_test.TestSparseToDense::test_sparse_to_dense_with_data_to_infer_dim(), caffe2.python.ideep.spatial_bn_op_test.TestSpatialBN::test_spatialbn_train_mode(), caffe2.python.ideep.expanddims_squeeze_op_test.ExpandDimsSqueezeTest::test_squeeze_fallback(), caffe2.contrib.nccl.nccl_ops_test.NCCLOpsTest::test_timings(), caffe2.contrib.nnpack.nnpack_ops_test.NNPackOpsTest::test_timings(), caffe2.python.transformations_test.TestTransformations::test_transformer_FuseConvBNNoConvBiasDuplicatedName(), caffe2.python.core_gradients_test.TestGradientCalculation::test_two_grads(), caffe2.contrib.aten.aten_test.TestATen::test_unique(), caffe2.python.net_printer_test.TestNetPrinter::test_valid_job(), caffe2.python.optimizer_test.TestWeightDecay::test_weight_decay(), caffe2.python.hypothesis_test.TestOperators::test_weighted_sample_blobs_queue(), caffe2.python.core_gradients_test.TestGradientsAccumulationWithPassThroughGradients::testAccumulationRuns(), caffe2.python.core_gradients_test.TestGradientsAccumulationWithPassThroughGradients::testAddAndStaticConstant(), caffe2.python.muji_test.TestMuji::testAllreduceWithEightGPUs(), caffe2.python.muji_test.TestMuji::testAllreduceWithFourGPUs(), caffe2.python.muji_test.TestMuji::testAllreduceWithFourGPUsAndTwoGroups(), caffe2.python.muji_test.TestMuji::testAllreduceWithTwoGPUs(), caffe2.python.mkl.mkl_pool_speed_test.TestMKLBasic::testAveragePoolingSpeed(), caffe2.python.caffe_translator_test.TestNumericalEquivalence::testBlobs(), caffe2.python.mkl.mkl_speed_test.TestMKLBasic::testConvSpeed(), caffe2.python.mkl.mkl_fc_speed_test.TestMKLBasic::testFCSpeed(), caffe2.python.core_gradients_test.TestGradientCalculation::testGradientCalculationWithPrint(), caffe2.python.core_gradients_test.TestGradientCalculation::testGradientMappingUsingSumOp(), caffe2.python.mkl.mkl_LRN_speed_test.TestMKLBasic::testLRNSpeed(), caffe2.python.mkl.mkl_pool_speed_test.TestMKLBasic::testMaxPoolingSpeed(), caffe2.python.core_gradients_test.TestGradientCalculation::testMultiUseInputAndMultipleVersionsBig(), caffe2.python.observer_test.TestObservers::testObserver(), caffe2.python.mkl.mkl_speed_test.TestMKLBasic::testReLUSpeed(), caffe2.python.mkl.mkl_sbn_speed_test.TestMKLBasic::testSpatialBNTestingSpeed(), caffe2.python.mkl.mkl_sbn_speed_test.TestMKLBasic::testSpatialBNTrainingSpeed(), caffe2.python.core_gradients_test.TestGradientsAccumulationWithPassThroughGradients::testSubOpInMiddle(), caffe2.python.core_gradients_test.TestGradientCalculation::testVersionMismatch(), torch.futures.Future::then(), pipe::train(), caffe2.experiments.python.SparseTransformer::transFCRelu(), tools.download_mnist::unzip(), upload_scribe.ScribeUploader::upload(), update-caffe2-models::upload_models(), update-models-from-caffe2::upload_onnx_model(), caffe2.contrib.warpctc.ctc_ops_test.CTCOpsTest::verify_cost(), and caffe2.python.memonger::verify_graph_equality().

printScale()

static void at::printScale ( std::ostream &  stream, double  scale  )

static

Definition at line 135 of file Formatting.cpp.

References defaultfloat(), scale, and stream.

Referenced by __printMatrix(), and print().

prod_intlist() [1/2]

template<typename C , typename std::enable_if< std::is_integral< typename C::value_type >::value, int >::type = 0>

int64_t at::prod_intlist ( const C &  container )

inline

Definition at line 99 of file Utils.h.

References torch::autograd::accumulate(), and caffe2::int64_t.

Referenced by at::native::_cdist_backward(), at::native::vulkan::detail::add(), at::native::detail::as_cufft_embed(), batchedTensorForLoopFallback(), batchedTensorInplaceForLoopFallback(), at::native::vulkan::detail::VulkanTensor::Impl::buffer_size_for_sizes(), at::native::cdist_impl(), at::detail::computeStride(), at::detail::empty_cpu(), at::native::mobile::empty_with_tail_padding(), at::native::flatten(), at::native::group_norm(), torch::autograd::generated::details::infinitely_differentiable_native_layer_norm_backward(), at::native::matmul(), at::native::vulkan::detail::mul(), at::native::ConvParams::needs_64bit_indexing_no_split(), new_qtensor(), at::native::quantized_group_norm_impl(), torch::jit::resizeConstantScalarOrTensorToShape(), torch::autograd::generated::details::sparse_constructor_values_backward(), at::native::unflatten(), at::native::metal::unpack(), at::meta::upsample_nearest2d(), and at::native::zero_().

prod_intlist() [2/2]

template<typename Iter , typename std::enable_if< std::is_integral< typename std::iterator_traits< Iter >::value_type >::value, int >::type = 0>

int64_t at::prod_intlist ( Iter  begin, Iter  end  )

inline

Definition at line 105 of file Utils.h.

References torch::autograd::accumulate(), matmul_dlmc_bench::end, and caffe2::int64_t.

REGISTER_CONTEXT() [1/2]

at::REGISTER_CONTEXT	(	DeviceType::CPU	,
		caffe2::CPUContext
	)

REGISTER_CONTEXT() [2/2]

at::REGISTER_CONTEXT	(	DeviceType::IDEEP	,
		caffe2::IDEEPContext
	)

REGISTER_COPY_BYTES_FUNCTION() [1/4]

at::REGISTER_COPY_BYTES_FUNCTION	(	DeviceType::CPU	,
		DeviceType::CPU	,
		caffe2::CopyBytesWrapper
	)

REGISTER_COPY_BYTES_FUNCTION() [2/4]

at::REGISTER_COPY_BYTES_FUNCTION	(	DeviceType::CPU	,
		DeviceType::IDEEP	,
		CopyBytesWrapper
	)

REGISTER_COPY_BYTES_FUNCTION() [3/4]

at::REGISTER_COPY_BYTES_FUNCTION	(	DeviceType::IDEEP	,
		DeviceType::CPU	,
		CopyBytesWrapper
	)

REGISTER_COPY_BYTES_FUNCTION() [4/4]

at::REGISTER_COPY_BYTES_FUNCTION	(	DeviceType::IDEEP	,
		DeviceType::IDEEP	,
		CopyBytesWrapper
	)

releaseRecordAllFunctions()

TORCH_API void at::releaseRecordAllFunctions

(

)

Definition at line 491 of file record_function.cpp.

References TORCH_CHECK.

Referenced by at::CallbackManager::removeCallback(), and at::ThreadLocalStateGuard::~ThreadLocalStateGuard().

removeCallback()

TORCH_API void at::removeCallback

(

CallbackHandle

handle

)

removeCallback removes a callback given the handle returned by addThreadLocalCallback or addGlobalCallback;

WARNING: removing a global callback is not thread safe, no other code can run simultaneously

Definition at line 377 of file record_function.cpp.

References handle.

Referenced by torch.autograd.profiler::disableProfilerLegacy().

report_positional_error()

static void at::report_positional_error ( const Dimname &  name, const Dimname &  other_name, DimnameList  names, DimnameList  other_names, const char *  action  )

static

Definition at line 39 of file NamedTensorUtils.cpp.

References fastrnns.bench::action, name, microbenchmarks::names, and TORCH_CHECK.

Referenced by unify_from_right().

reportNYIDimnameOverload()

void at::reportNYIDimnameOverload ( const char *  op_name )

inline

Definition at line 34 of file NamedTensorUtils.h.

References TORCH_CHECK.

Referenced by at::native::all(), at::native::all_out(), at::native::any(), at::native::any_out(), at::native::argmax(), at::native::argmin(), at::native::argsort(), at::native::gather(), at::native::gather_out(), at::native::index_add(), at::native::index_add_(), at::native::index_copy(), at::native::index_copy_(), at::native::index_select(), at::native::index_select_out(), at::native::scatter(), at::native::scatter_(), at::native::scatter_add(), at::native::scatter_add_(), at::native::sort(), at::native::sort_out(), and at::native::squeeze_().

reshape_batching_rule()

Tensor at::reshape_batching_rule	(	const Tensor &	self,
		IntArrayRef	shape
	)

Definition at line 407 of file BatchingRegistrations.cpp.

References at::MultiBatchVmapTransform::logicalToPhysical(), and shape.

Referenced by TORCH_LIBRARY_IMPL().

round_columns()

static std::tuple<int64_t, int64_t> at::round_columns ( TensorIteratorBase &  iter, int  dim, int  multiple, int64_t  begin, int64_t  end  )

static

Definition at line 89 of file TensorIteratorReduce.cpp.

References dim, matmul_dlmc_bench::end, iter, and torch.jit._trace::make_tuple().

Referenced by parallel_dim_reduction().

safeStack()

static Tensor at::safeStack ( TensorList  tensors )

static

Definition at line 206 of file BatchedFallback.cpp.

References at::Tensor::defined(), torch.distributions.constraints::stack, caffe2::tensors, and TORCH_CHECK.

Referenced by batchedTensorForLoopFallback().

select_backward_batching_rule()

Tensor at::select_backward_batching_rule	(	const Tensor &	grad,
		IntArrayRef	input_sizes,
		int64_t	dim,
		int64_t	index
	)

Definition at line 354 of file BatchingRegistrations.cpp.

References dim, getGradInputPhysicalDim(), grad, index, at::MultiBatchVmapTransform::logicalToPhysical(), and caffe2::zeros.

Referenced by TORCH_LIBRARY_IMPL().

select_batching_rule()

Tensor at::select_batching_rule	(	const Tensor &	self,
		int64_t	dim,
		int64_t	index
	)

Definition at line 343 of file BatchingRegistrations.cpp.

References dim, index, and at::MultiBatchVmapTransform::logicalToPhysical().

Referenced by TORCH_LIBRARY_IMPL().

set_num_interop_threads()

TORCH_API void at::set_num_interop_threads

(

int

)

Referenced by main(), and THPModule_setNumInteropThreads().

set_num_threads()

TORCH_API void at::set_num_threads

(

int

)

Referenced by main(), and THPModule_setNumThreads().

set_quantizer_()

TORCH_API void at::set_quantizer_	(	const Tensor &	self,
		ConstQuantizerPtr	quantizer
	)

Definition at line 210 of file Quantizer.cpp.

References get_qtensorimpl(), and at::QTensorImpl::set_quantizer_().

Referenced by at::native::copy_impl().

set_record_function_tls_()

TORCH_API void at::set_record_function_tls_

(

const RecordFunctionTLS &

tls

)

Definition at line 62 of file record_function.cpp.

Referenced by at::ThreadLocalState::setThreadLocalState().

should_include_kernel_dtype()

constexpr bool at::should_include_kernel_dtype ( const char *  kernel_tag_str, at::ScalarType  scalar_type  )

inlineconstexpr

The method should_include_kernel_dtype() returns true/false based on whether the switching code for a specific dtype should be included based on build time constants generated from tracing model execution.

This method will be implmeneted via code-generation and included in this file when code-gen is ready.

Definition at line 24 of file Dispatch.h.

shouldRunRecordFunction()

TORCH_API bool at::shouldRunRecordFunction

(

bool *

pre_sampled

)

Definition at line 499 of file record_function.cpp.

References hasGlobalCallbacks().

Referenced by c10::Dispatcher::callBoxed(), c10::Dispatcher::callWithDispatchKey(), torch::jit::InterpreterStateImpl::checkAndStartRecordFunction(), torch::autograd::Node::operator()(), and runPureRecordFunctionBench().

show_config()

TORCH_API std::string at::show_config

(

)

Returns a detailed string describing the configuration PyTorch.

Definition at line 124 of file Version.cpp.

References get_mkl_version(), get_mkldnn_version(), get_openmp_version(), caffe2::GetBuildOptions(), at::detail::getCUDAHooks(), hasCUDA(), at::CUDAHooksInterface::showConfig(), and used_cpu_capability().

Referenced by THPModule_showConfig().

size()

int64_t at::size ( const Tensor &  tensor, int64_t  dim  )

inline

Definiti