Crate onednnl_sys

Crate onednnl_sys 

Source

Modules§

dnnl_accumulation_mode_t
dnnl_alg_kind_t
dnnl_cpu_isa_hints_t
dnnl_cpu_isa_t
dnnl_data_type_t
dnnl_engine_kind_t
dnnl_format_kind_t
dnnl_format_tag_t
dnnl_fpmath_mode_t
dnnl_graph_layout_type_t
dnnl_graph_op_attr_t
dnnl_graph_op_kind_t
dnnl_graph_partition_policy_t
dnnl_graph_tensor_property_t
dnnl_normalization_flags_t
dnnl_primitive_kind_t
dnnl_prop_kind_t
dnnl_query_t
dnnl_rnn_direction_t
dnnl_rnn_flags_t
dnnl_rounding_mode_t
dnnl_scratchpad_mode_t
dnnl_status_t
dnnl_stream_flags_t
ocl
sycl

Structs§

dnnl_engine
@struct dnnl_engine @brief An opaque structure to describe an engine.
dnnl_exec_arg_t
A structure that contains an index and a memory object, and is used to pass arguments to dnnl_primitive_execute().
dnnl_graph_allocator
An opaque structure to describe an allocator.
dnnl_graph_compiled_partition
An opaque structure to describe a compiled partition.
dnnl_graph_graph
An opaque structure to describe a graph.
dnnl_graph_inplace_pair_t
In-place pair definition. It can queried from a compiled partition indicating that an input and an output of the partition can share the same memory buffer for computation. In-place computation helps to reduce the memory footprint and improves cache locality. But since the library may not have a global view of user’s application, it’s possible that the tensor with input_id is used at other places in user’s computation graph. In this case, the user should take the in-place pair as a hint and pass a different memory buffer for output tensor to avoid overwriting the input memory buffer which will probably cause unexpected incorrect results.
dnnl_graph_logical_tensor_t
Logical tensor. It is based on an ID, a number of dimensions, dimensions themselves, element data type, tensor property and tensor memory layout.
dnnl_graph_op
An opaque structure to describe an operation.
dnnl_graph_partition
An opaque structure to describe a partition.
dnnl_graph_tensor
An opaque structure to describe a tensor.
dnnl_memory
@struct dnnl_memory An opaque structure to describe a memory.
dnnl_memory_desc
@struct dnnl_memory_desc An opaque structure to describe a memory descriptor.
dnnl_post_ops
@struct dnnl_post_ops @brief An opaque structure for a chain of post operations.
dnnl_primitive
@struct dnnl_primitive An opaque structure to describe a primitive.
dnnl_primitive_attr
@struct dnnl_primitive_attr @brief An opaque structure for primitive descriptor attributes.
dnnl_primitive_desc
@struct dnnl_primitive_desc @brief An opaque structure to describe a primitive descriptor.
dnnl_stream
@struct dnnl_stream An opaque structure to describe an execution stream.
dnnl_version_t
Structure containing version information as per Semantic Versioning

Constants§

BUILD_AMX
BUILD_AVX2
BUILD_AVX512
BUILD_BATCH_NORMALIZATION
BUILD_BINARY
BUILD_CONCAT
BUILD_CONVOLUTION
BUILD_DECONVOLUTION
BUILD_ELTWISE
BUILD_GEMM_AVX2
BUILD_GEMM_AVX512
BUILD_GEMM_KERNELS_ALL
BUILD_GEMM_KERNELS_NONE
BUILD_GEMM_SSE41
BUILD_GEN9
BUILD_GEN11
BUILD_GROUP_NORMALIZATION
BUILD_INFERENCE
BUILD_INNER_PRODUCT
BUILD_LAYER_NORMALIZATION
BUILD_LRN
BUILD_MATMUL
BUILD_POOLING
BUILD_PRELU
BUILD_PRIMITIVE_ALL
BUILD_PRIMITIVE_CPU_ISA_ALL
BUILD_PRIMITIVE_GPU_ISA_ALL
BUILD_REDUCTION
BUILD_REORDER
BUILD_RESAMPLING
BUILD_RNN
BUILD_SDPA
BUILD_SHUFFLE
BUILD_SOFTMAX
BUILD_SSE41
BUILD_SUM
BUILD_TRAINING
BUILD_XE2
BUILD_XE3
BUILD_XEHP
BUILD_XEHPC
BUILD_XEHPG
BUILD_XELP
DNNL_ARG_ATTR_DROPOUT_MASK
DNNL_ARG_ATTR_DROPOUT_PROBABILITY
DNNL_ARG_ATTR_DROPOUT_SEED
DNNL_ARG_ATTR_MULTIPLE_POST_OP_BASE
DNNL_ARG_ATTR_OUTPUT_SCALES
DNNL_ARG_ATTR_POST_OP_DW
DNNL_ARG_ATTR_ROUNDING_SEED
DNNL_ARG_ATTR_SCALES
DNNL_ARG_ATTR_ZERO_POINTS
DNNL_ARG_AUGRU_ATTENTION
DNNL_ARG_BIAS
DNNL_ARG_DIFF_AUGRU_ATTENTION
DNNL_ARG_DIFF_BIAS
DNNL_ARG_DIFF_DST
DNNL_ARG_DIFF_DST_0
DNNL_ARG_DIFF_DST_1
DNNL_ARG_DIFF_DST_2
DNNL_ARG_DIFF_DST_ITER
DNNL_ARG_DIFF_DST_ITER_C
DNNL_ARG_DIFF_DST_LAYER
DNNL_ARG_DIFF_SCALE
DNNL_ARG_DIFF_SHIFT
DNNL_ARG_DIFF_SRC
DNNL_ARG_DIFF_SRC_0
DNNL_ARG_DIFF_SRC_1
DNNL_ARG_DIFF_SRC_2
DNNL_ARG_DIFF_SRC_3
DNNL_ARG_DIFF_SRC_ITER
DNNL_ARG_DIFF_SRC_ITER_C
DNNL_ARG_DIFF_SRC_LAYER
DNNL_ARG_DIFF_WEIGHTS
DNNL_ARG_DIFF_WEIGHTS_0
DNNL_ARG_DIFF_WEIGHTS_1
DNNL_ARG_DIFF_WEIGHTS_2
DNNL_ARG_DIFF_WEIGHTS_3
DNNL_ARG_DIFF_WEIGHTS_ITER
DNNL_ARG_DIFF_WEIGHTS_LAYER
DNNL_ARG_DIFF_WEIGHTS_PEEPHOLE
DNNL_ARG_DIFF_WEIGHTS_PROJECTION
DNNL_ARG_DST
DNNL_ARG_DST_0
DNNL_ARG_DST_1
DNNL_ARG_DST_2
DNNL_ARG_DST_ITER
DNNL_ARG_DST_ITER_C
DNNL_ARG_DST_LAYER
DNNL_ARG_FROM
DNNL_ARG_MEAN
DNNL_ARG_MULTIPLE_DST
DNNL_ARG_MULTIPLE_SRC
DNNL_ARG_REDUCE
DNNL_ARG_SCALE
DNNL_ARG_SCRATCHPAD
DNNL_ARG_SHIFT
DNNL_ARG_SRC
DNNL_ARG_SRC_0
DNNL_ARG_SRC_1
DNNL_ARG_SRC_2
DNNL_ARG_SRC_3
DNNL_ARG_SRC_ITER
DNNL_ARG_SRC_ITER_C
DNNL_ARG_SRC_LAYER
DNNL_ARG_TO
DNNL_ARG_UNDEF
DNNL_ARG_VARIANCE
DNNL_ARG_WEIGHTS
DNNL_ARG_WEIGHTS_0
DNNL_ARG_WEIGHTS_1
DNNL_ARG_WEIGHTS_2
DNNL_ARG_WEIGHTS_3
DNNL_ARG_WEIGHTS_ITER
DNNL_ARG_WEIGHTS_LAYER
DNNL_ARG_WEIGHTS_PEEPHOLE
DNNL_ARG_WEIGHTS_PROJECTION
DNNL_ARG_WORKSPACE
DNNL_CPU_RUNTIME
DNNL_CPU_THREADING_RUNTIME
DNNL_GPU_RUNTIME
DNNL_GPU_VENDOR
DNNL_GRAPH_UNKNOWN_NDIMS
DNNL_JIT_PROFILE_LINUX_JITDUMP
DNNL_JIT_PROFILE_LINUX_JITDUMP_USE_TSC
DNNL_JIT_PROFILE_LINUX_PERF
DNNL_JIT_PROFILE_LINUX_PERFMAP
DNNL_JIT_PROFILE_NONE
DNNL_JIT_PROFILE_VTUNE
DNNL_MAX_NDIMS
DNNL_RUNTIME_DPCPP
DNNL_RUNTIME_NONE
DNNL_RUNTIME_OCL
DNNL_RUNTIME_OMP
DNNL_RUNTIME_S32_VAL_REP
@cond DO_NOT_DOCUMENT_THIS
DNNL_RUNTIME_SEQ
DNNL_RUNTIME_SYCL
DNNL_RUNTIME_TBB
DNNL_RUNTIME_THREADPOOL
DNNL_VENDOR_AMD
DNNL_VENDOR_GENERIC
DNNL_VENDOR_INTEL
DNNL_VENDOR_NONE
DNNL_VENDOR_NVIDIA
DNNL_VERSION_MAJOR
DNNL_VERSION_MINOR
DNNL_VERSION_PATCH

Statics§

DNNL_RUNTIME_F32_VAL_REP

Functions§

dnnl_augru_backward_primitive_desc_create
Creates a primitive descriptor for AUGRU backward propagation primitive.
dnnl_augru_forward_primitive_desc_create
Creates a primitive descriptor for AUGRU forward propagation primitive.
dnnl_batch_normalization_backward_primitive_desc_create
Creates a primitive descriptor for a batch normalization backward propagation primitive.
dnnl_batch_normalization_forward_primitive_desc_create
Creates a primitive descriptor for a batch normalization forward propagation primitive.
dnnl_binary_primitive_desc_create
Creates a primitive descriptor for a binary primitive.
dnnl_binary_primitive_desc_create_v2
Creates a primitive descriptor for a binary primitive with support of ternary operators.
dnnl_concat_primitive_desc_create
Creates a primitive descriptor for an out-of-place concatenation primitive.
dnnl_convolution_backward_data_primitive_desc_create
Creates a primitive descriptor for a convolution backward propagation primitive.
dnnl_convolution_backward_weights_primitive_desc_create
Creates a primitive descriptor for a convolution weights gradient primitive.
dnnl_convolution_forward_primitive_desc_create
Creates a primitive descriptor for a convolution forward propagation primitive.
dnnl_data_type_size
Returns the size of data type.
dnnl_deconvolution_backward_data_primitive_desc_create
Creates a primitive descriptor for a deconvolution backward propagation primitive.
dnnl_deconvolution_backward_weights_primitive_desc_create
Creates a primitive descriptor for a deconvolution weights gradient primitive.
dnnl_deconvolution_forward_primitive_desc_create
Creates a primitive descriptor for a deconvolution forward propagation primitive.
dnnl_eltwise_backward_primitive_desc_create
Creates a primitive descriptor for an eltwise backward propagation primitive.
dnnl_eltwise_forward_primitive_desc_create
Creates a primitive descriptor for an eltwise forward propagation primitive.
dnnl_engine_create
Creates an engine.
dnnl_engine_destroy
Destroys an engine.
dnnl_engine_get_count
Returns the number of engines of a particular kind.
dnnl_engine_get_kind
Returns the kind of an engine.
dnnl_gemm_s8s8s32
Performs integer matrix-matrix multiply on 8-bit signed matrix A, 8-bit signed matrix B, and 32-bit signed resulting matrix C.
dnnl_gemm_u8s8s32
Performs integer matrix-matrix multiply on 8-bit unsigned matrix A, 8-bit signed matrix B, and 32-bit signed resulting matrix C.
dnnl_get_cpu_isa_hints
Gets the ISA specific hints that library can follow. See #dnnl_cpu_isa_hints_t and #dnnl::cpu_isa_hints for the list of the values returned by the C and C++ API functions respectively.
dnnl_get_default_fpmath_mode
Returns the floating-point math mode that will be used by default for all subsequently created primitives.
dnnl_get_effective_cpu_isa
Gets the maximal ISA the library can dispatch to on the CPU. See #dnnl_cpu_isa_t and #dnnl::cpu_isa for the list of the values returned by the C and C++ API functions respectively.
dnnl_get_primitive_cache_capacity
Returns the number of primitives that can be held in the primitive cache at the same time.
dnnl_graph_add_op
Adds an operation into a graph. The API will return failure if the operator has already been added to the graph or the operation cannot pass the schema check in the library (eg. input and output numbers and data types, the attributes of the operation, etc.).
dnnl_graph_allocator_create
Creates a host allocator with the given allocation and deallocation call-back function pointers.
dnnl_graph_allocator_destroy
Destroys an allocator.
dnnl_graph_compiled_partition_create
Creates a new compiled partition handle.
dnnl_graph_compiled_partition_destroy
Destroys a compiled partition.
dnnl_graph_compiled_partition_execute
Executes a compiled partition.
dnnl_graph_compiled_partition_get_inplace_ports
Returns the hint of in-place pairs from a compiled partition. It indicates that an input and an output of the partition can share the same memory buffer for computation. In-place computation helps to reduce the memory footprint and improves cache locality. But since the library may not have a global view of user’s application, it’s possible that the tensor with input_id is used at other places in user’s computation graph. In this case, the user should take the in-place pair as a hint and pass a different memory buffer for output tensor to avoid overwriting the input memory buffer which will probably cause unexpected incorrect results.
dnnl_graph_compiled_partition_query_logical_tensor
Queries an input or output logical tensor according to tensor ID. If the tensor ID doesn’t belong to any input or output of the compiled partition, an error status #dnnl_invalid_arguments will be returned by the API.
dnnl_graph_get_compiled_partition_cache_capacity
Returns the number of compiled partitions that can be held in the compiled partition cache at the same time.
dnnl_graph_get_constant_tensor_cache
Return the enabling or disabling status of constant tensor cache.
dnnl_graph_get_constant_tensor_cache_capacity
Return the current capacity of constant tensor cache.
dnnl_graph_graph_create
Creates a new empty graph. A graph is associated to a specific engine kind. The partitions returned from the graph will inherit the engine kind of the graph.
dnnl_graph_graph_create_with_fpmath_mode
Creates a new empty graph with an engine kind and a floating-point math mode. All partitions returned from the graph will inherit the engine kind and floating-point math mode.
dnnl_graph_graph_destroy
Destroys a graph.
dnnl_graph_graph_filter
Filters a graph. Partitions will be claimed internally according to the capability of the library, the engine kind, and the policy.
dnnl_graph_graph_finalize
Finalizes a graph. It means users have finished adding operations into the graph and the graph is ready for partitioning. Adding a new operation into a finalized graph will return failures. Similarly, partitioning on a un-finalized graph will also return failures.
dnnl_graph_graph_get_fpmath_mode
Get the floating point math mode for a graph.
dnnl_graph_graph_get_partition_num
Returns the number of partitions of a graph. The API should be called after a partition is already filtered. Otherwise, the output number is zero.
dnnl_graph_graph_get_partitions
Returns the partitions from a filtered graph. Output partition instances will be written into the parameter partitions. Users need to make sure partitions is valid and has enough space to accept the partition instances. Each output partition instance should be destroyed via #dnnl_graph_partition_destroy explicitly after use.
dnnl_graph_graph_is_finalized
Checks if a graph is finalized.
dnnl_graph_graph_set_fpmath_mode
Set the floating point math mode for a graph.
dnnl_graph_logical_tensor_get_mem_size
Returns the memory size described by the logical tensor. If it’s a strided layout, the size will be calculated by dims and strides. If it’s an opaque layout, the size will be decided by layout_id.
dnnl_graph_logical_tensor_init
Initializes a logical tensor with id, data type, number of dimensions, layout type, and property. The logical tensor’s dims are unknown with this interface.
dnnl_graph_logical_tensor_init_with_dims
Initializes a logical tensor with basic information and dims. The logical tensor’s dimensions and layout will be initialized according to the input arguments.
dnnl_graph_logical_tensor_init_with_strides
Initializes a logical tensor with dimensions and strides provided by user.
dnnl_graph_logical_tensor_is_equal
Compares if two logical tenors are equal. Users can decide accordingly if layout reordering is needed for two logical tensors. The method will return true for below two circumstances:
dnnl_graph_make_engine_with_allocator
This API is a supplement for existing onednn engine API.
dnnl_graph_op_add_input
Adds input logical tensor to the op.
dnnl_graph_op_add_output
Adds output logical tensor to the op.
dnnl_graph_op_create
Initializes an op with unique id, kind, and name.
dnnl_graph_op_destroy
Destroys an op.
dnnl_graph_op_get_id
Returns the unique id of an op.
dnnl_graph_op_get_kind
Returns the kind of an op.
dnnl_graph_op_set_attr_bool
Sets boolean attribute to an op.
dnnl_graph_op_set_attr_f32
Sets floating point attribute to an op.
dnnl_graph_op_set_attr_s64
Sets integer attribute to an op.
dnnl_graph_op_set_attr_str
Sets string attribute to an op.
dnnl_graph_partition_compile
Compiles a partition with given input and output logical tensors. The output logical tensors can contain unknown dimensions. For this case, the compilation will deduce the output shapes according to input shapes. The output logical tensors can also have layout type any. The compilation will choose the optimal layout for output tensors. The optimal layout will be represented as an opaque layout ID saved in the output logical tensor.
dnnl_graph_partition_create_with_op
Creates a new partition with a given operator and engine kind. The API is used to create a partition from an operation directly without creating the graph and calling get_partitions(). The output partition contains only one operation specified by the parameter. The output partition instance should be destroyed via #dnnl_graph_partition_destroy after use.
dnnl_graph_partition_destroy
Destroys a partition.
dnnl_graph_partition_get_engine_kind
Returns the engine kind of a partition.
dnnl_graph_partition_get_id
Returns the ID of a partition.
dnnl_graph_partition_get_input_ports
Returns a list of input logical tensors from a partition.
dnnl_graph_partition_get_input_ports_num
Returns the number of input logical tensors of a partition.
dnnl_graph_partition_get_op_num
Returns the number of operations in a partition.
dnnl_graph_partition_get_ops
Returns the list of op IDs of the partition.
dnnl_graph_partition_get_output_ports
Returns a list of output logical tensors from a partition.
dnnl_graph_partition_get_output_ports_num
Returns the number of output logical tensors of a partition.
dnnl_graph_partition_is_supported
Returns the supporting status of a partition. Some operations may not be supported by the library under certain circumstances. During partitioning stage, unsupported partitions will be returned to users with each containing an unsupported operation. Users should check the supporting status of a partition before transforming the computation graph or compiling the partition.
dnnl_graph_set_compiled_partition_cache_capacity
Sets a number of compiled partitions that can be held in the compiled partition cache at the same time. The default capacity of compiled partition cache is 1024.
dnnl_graph_set_constant_tensor_cache
Control the enabling or disabling of constant tensor cache. This API must be called once before compilation stage. By default, constant tensor cache is disabled in the library.
dnnl_graph_set_constant_tensor_cache_capacity
Control the capacity for the constant tensor cache that used for specific engine kind. This API is thread safe and can be called multiple times at runtime. The capacity is set to zero by default which means the cache is disabled. When calling this API, the corresponding cache will be flushed. Setting capacity to 0 means to clear all cached tensors and disable cache. Once the capacity limit is reached, no new tensors will be cached. If there are multiple devices for an engine kind, the capacity set here is for each device.
dnnl_graph_tensor_create
Creates a tensor with logical tensor, engine, and data handle.
dnnl_graph_tensor_destroy
Destroys a tensor.
dnnl_graph_tensor_get_data_handle
Gets the data handle of a tensor.
dnnl_graph_tensor_get_engine
Returns the engine of a tensor object.
dnnl_graph_tensor_get_logical_tensor
Returns the logical tensor of a tensor object.
dnnl_graph_tensor_set_data_handle
Set data handle for a tensor.
dnnl_group_normalization_backward_primitive_desc_create
Creates a primitive descriptor for a group normalization backward propagation primitive.
dnnl_group_normalization_forward_primitive_desc_create
Creates a primitive descriptor for a group normalization forward propagation primitive.
dnnl_gru_backward_primitive_desc_create
Creates a primitive descriptor for GRU backward propagation primitive.
dnnl_gru_forward_primitive_desc_create
Creates a primitive descriptor for GRU forward propagation primitive.
dnnl_inner_product_backward_data_primitive_desc_create
Creates a primitive descriptor for an inner product backward propagation primitive.
dnnl_inner_product_backward_weights_primitive_desc_create
Creates a primitive descriptor for an inner product weights gradient primitive.
dnnl_inner_product_forward_primitive_desc_create
Creates a primitive descriptor for an inner product forward propagation primitive.
dnnl_layer_normalization_backward_primitive_desc_create
Creates a primitive descriptor for a layer normalization backward propagation primitive.
dnnl_layer_normalization_backward_primitive_desc_create_v2
Creates a primitive descriptor for a layer normalization backward propagation primitive with a user-provided data type for the scale and shift memory objects.
dnnl_layer_normalization_forward_primitive_desc_create
Creates a primitive descriptor for a layer normalization forward propagation primitive.
dnnl_layer_normalization_forward_primitive_desc_create_v2
Creates a primitive descriptor for a layer normalization forward propagation primitive with a user-provided data type for the scale and shift memory objects.
dnnl_lbr_augru_backward_primitive_desc_create
Creates a primitive descriptor for LBR AUGRU backward propagation primitive.
dnnl_lbr_augru_forward_primitive_desc_create
Creates a primitive descriptor for LBR AUGRU forward propagation primitive.
dnnl_lbr_gru_backward_primitive_desc_create
Creates a primitive descriptor for LBR GRU backward propagation primitive.
dnnl_lbr_gru_forward_primitive_desc_create
Creates a descriptor for LBR GRU forward propagation primitive.
dnnl_lrn_backward_primitive_desc_create
Creates a primitive descriptor for an LRN backward propagation primitive.
dnnl_lrn_forward_primitive_desc_create
Creates a primitive descriptor for an LRN forward propagation primitive.
dnnl_lstm_backward_primitive_desc_create
Creates a primitive descriptor for an LSTM backward propagation primitive.
dnnl_lstm_forward_primitive_desc_create
Creates a primitive descriptor for an LSTM forward propagation primitive.
dnnl_matmul_primitive_desc_create
Creates a primitive descriptor for a matrix multiplication primitive.
dnnl_memory_create
Creates a memory object.
dnnl_memory_desc_clone
Clones a memory descriptor. The resulting memory descriptor must be destroyed separately.
dnnl_memory_desc_create_submemory
@param memory_desc Output memory descriptor. @param parent_memory_desc An existing memory descriptor. @param dims Sizes of the region. @param offsets Offsets to the region from the encompassing memory object in each dimension @returns #dnnl_success on success and a status describing the error otherwise.
dnnl_memory_desc_create_with_blob
Creates a memory descriptor from a memory descriptor binary blob.
dnnl_memory_desc_create_with_strides
Creates a memory descriptor using dimensions and strides.
dnnl_memory_desc_create_with_tag
Creates a memory descriptor using dimensions and memory format tag.
dnnl_memory_desc_destroy
Destroys a memory descriptor.
dnnl_memory_desc_equal
Compares two memory descriptors.
dnnl_memory_desc_get_blob
Retrieves a binary blob associated with the given memory descriptor
dnnl_memory_desc_get_size
Returns the size of a memory descriptor.
dnnl_memory_desc_permute_axes
Creates a memory descriptor by permuting axes in an existing one.
dnnl_memory_desc_query
Queries a memory descriptor for various pieces of information.
dnnl_memory_desc_reshape
Creates a memory descriptor by reshaping an existing one. The new memory descriptor inherits the data type. This operation is valid only for memory descriptors that have format_kind #dnnl_blocked or #dnnl_format_kind_any.
dnnl_memory_destroy
Destroys a memory object.
dnnl_memory_get_data_handle
Returns memory object’s data handle.
dnnl_memory_get_engine
Returns the engine of a memory object.
dnnl_memory_get_memory_desc
Returns the memory descriptor for a memory object.
dnnl_memory_map_data
Maps a memory object and returns a host-side pointer to a memory buffer with a copy of its contents.
dnnl_memory_set_data_handle
Sets the underlying memory buffer.
dnnl_memory_unmap_data
Unmaps a memory object and writes back any changes made to the previously mapped memory buffer. The pointer to the mapped buffer must be obtained via the dnnl_memory_map_data() call.
dnnl_pooling_backward_primitive_desc_create
Creates a primitive descriptor for a pooling backward propagation primitive.
dnnl_pooling_forward_primitive_desc_create
Creates a primitive descriptor for a pooling forward propagation primitive.
dnnl_post_ops_append_binary
Appends a binary post-op.
dnnl_post_ops_append_dw
Appends a depthwise post-op convolution.
dnnl_post_ops_append_eltwise
Appends an elementwise post-op.
dnnl_post_ops_append_prelu
Appends a prelu forward post-op.
dnnl_post_ops_append_sum
Appends an accumulation v3 (sum) to post-ops. Prior to accumulating the result, a zero point is subtracted from the previous value and is multiplied by the scale.
dnnl_post_ops_clone
Clones post-ops primitive attribute.
dnnl_post_ops_create
Creates empty post-ops sequence.
dnnl_post_ops_destroy
Destroys post-ops.
dnnl_post_ops_get_kind
Returns the kind of a post-op entry.
dnnl_post_ops_get_params_binary
Returns the parameters of a binary post-op.
dnnl_post_ops_get_params_dw
Returns the parameters of an depthwise post-op.
dnnl_post_ops_get_params_eltwise
Returns the parameters of an elementwise post-op.
dnnl_post_ops_get_params_prelu
Returns the parameters of a prelu post-op.
dnnl_post_ops_get_params_sum
Returns the parameters of an accumulation (sum) post-op with zero point and data type parameter.
dnnl_post_ops_len
Returns the length of post-ops.
dnnl_prelu_backward_primitive_desc_create
Creates a primitive descriptor for a PReLU (leaky ReLU with trainable alpha parameter) backward propagation primitive.
dnnl_prelu_forward_primitive_desc_create
Creates a primitive descriptor for a PReLU (leaky ReLU with trainable alpha parameter) forward propagation primitive.
dnnl_primitive_attr_clone
Clones primitive attributes.
dnnl_primitive_attr_create
Creates an empty (default) primitive attributes with all the parameters set to their default values.
dnnl_primitive_attr_destroy
Destroys primitive attributes.
dnnl_primitive_attr_get_accumulation_mode
Returns the accumulation mode primitive attribute.
dnnl_primitive_attr_get_deterministic
Returns the deterministic primitive attribute value.
dnnl_primitive_attr_get_dropout
Returns probability for output dropout primitive attribute.
dnnl_primitive_attr_get_fpmath_mode
Returns the floating-point math mode primitive attribute.
dnnl_primitive_attr_get_fpmath_mode_v2
Returns the floating-point math mode primitive attribute.
dnnl_primitive_attr_get_post_ops
Returns primitive attributes post-ops.
dnnl_primitive_attr_get_rnn_data_qparams
Returns the quantization scale and shift parameters for RNN data tensors.
dnnl_primitive_attr_get_rnn_weights_projection_qparams
Returns the quantization scaling factors for RNN projection weights tensors.
dnnl_primitive_attr_get_rnn_weights_qparams
Returns the quantization scaling factors for RNN weights tensors.
dnnl_primitive_attr_get_rounding
Returns the rounding mode attribute value for a given argument
dnnl_primitive_attr_get_scratchpad_mode
Returns the primitive attributes scratchpad mode.
dnnl_primitive_attr_set_accumulation_mode
Sets the accumulation mode primitive attribute.
dnnl_primitive_attr_set_deterministic
Sets the deterministic primitive attribute value.
dnnl_primitive_attr_set_dropout
Sets probability for output dropout primitive attribute.
dnnl_primitive_attr_set_fpmath_mode
Sets the floating-point math mode primitive attributes.
dnnl_primitive_attr_set_fpmath_mode_v2
Sets the floating-point math mode primitive attributes.
dnnl_primitive_attr_set_post_ops
Sets primitive attributes post-ops.
dnnl_primitive_attr_set_rnn_data_qparams
Set quantization scale and shift parameters for RNN data tensors.
dnnl_primitive_attr_set_rnn_weights_projection_qparams
Sets quantization scaling factors for RNN projection weights tensors. The low-precision configuration of the RNN primitives expects input weights to use the signed 8-bit integer data type. The scaling factors are used to quantize floating-point data to signed integer and must be passed to RNN primitives using attributes.
dnnl_primitive_attr_set_rnn_weights_qparams
Sets quantization scaling factors for RNN weights tensors. The low-precision configuration of the RNN primitives expects input weights to use the signed 8-bit integer data type. The scaling factors are used to quantize floating-point data to signed integer and must be passed to RNN primitives using attributes.
dnnl_primitive_attr_set_rounding
Sets the rounding mode attribute value for a given argument
dnnl_primitive_attr_set_scales
Sets primitive attributes scaling factors for primitive operations for a given memory argument. The scaling factors must be passed at execution time as an argument with index #DNNL_ARG_ATTR_SCALES | arg.
dnnl_primitive_attr_set_scales_mask
Sets primitive attributes scaling factors for primitive operations for a given memory argument. The scaling factors must be passed at execution time as an argument with index #DNNL_ARG_ATTR_SCALES | arg.
dnnl_primitive_attr_set_scratchpad_mode
Sets primitive attributes scratchpad mode.
dnnl_primitive_attr_set_zero_points
Sets primitive attributes zero points for primitive operations for a given memory argument. The zero points must be passed at execution time as an argument with index #DNNL_ARG_ATTR_ZERO_POINTS | arg.
dnnl_primitive_attr_set_zero_points_mask
Sets primitive attributes zero points for primitive operations for a given memory argument. The zero points must be passed at execution time as an argument with index #DNNL_ARG_ATTR_ZERO_POINTS | arg.
dnnl_primitive_create
Creates a primitive.
dnnl_primitive_create_from_cache_blob
Creates a primitive from a cache blob.
dnnl_primitive_desc_clone
Clones a primitive descriptor. The resulting primitive descriptor must be destroyed separately.
dnnl_primitive_desc_destroy
Destroys a primitive descriptor.
dnnl_primitive_desc_get_attr
Returns a constant reference to the attributes of a primitive descriptor.
dnnl_primitive_desc_next_impl
Changes the primitive descriptor to point to the next available implementation.
dnnl_primitive_desc_query
Queries a primitive descriptor for various pieces of information.
dnnl_primitive_desc_query_md
Queries primitive descriptor for a memory descriptor.
dnnl_primitive_desc_query_s32
Queries primitive descriptor for a signed 32bit int.
dnnl_primitive_destroy
Destroys a primitive.
dnnl_primitive_execute
@note If any argument in @p args is padded (padded_dims > dims), the primitive execution will assume properly zero-padded input arguments, and produce zero-padded output arguments.
dnnl_primitive_get_cache_blob
Retrieves a cache blob associated with the given primitive.
dnnl_primitive_get_primitive_desc
Retrieves a constant reference to the primitive descriptor of a given primitive.
dnnl_reduction_primitive_desc_create
Creates a primitive descriptor for a reduction primitive.
dnnl_reorder_primitive_desc_create
Creates a primitive descriptor for a reorder primitive.
dnnl_resampling_backward_primitive_desc_create
Creates a primitive descriptor for a resampling backward propagation primitive.
dnnl_resampling_forward_primitive_desc_create
Creates a primitive descriptor for a resampling forward propagation primitive.
dnnl_set_cpu_isa_hints
Sets the hints flag for the CPU ISA. See #dnnl_cpu_isa_hints_t and #dnnl::cpu_isa_hints for the list of the values accepted by the C and C++ API functions respectively.
dnnl_set_default_fpmath_mode
Sets the floating-point math mode that will be used by default for all subsequently created primitives.
dnnl_set_jit_dump
Configures dumping of JIT-generated code.
dnnl_set_jit_profiling_flags
Sets library profiling flags. The flags define which profilers are supported.
dnnl_set_jit_profiling_jitdumpdir
Sets JIT dump output path. Only applicable to Linux and is only used when profiling flags have DNNL_JIT_PROFILE_LINUX_PERF bit set.
dnnl_set_max_cpu_isa
Sets the maximal ISA the library can dispatch to on the CPU. See #dnnl_cpu_isa_t and #dnnl::cpu_isa for the list of the values accepted by the C and C++ API functions respectively.
dnnl_set_primitive_cache_capacity
Sets a number of primitives that can be held in the primitive cache at a time.
dnnl_set_verbose
Configures verbose output to stdout.
dnnl_sgemm
Performs single-precision matrix-matrix multiply.
dnnl_shuffle_backward_primitive_desc_create
Creates a primitive descriptor for a shuffle backward propagation primitive
dnnl_shuffle_forward_primitive_desc_create
Creates a primitive descriptor for a shuffle forward propagation primitive
dnnl_softmax_backward_primitive_desc_create
Creates a primitive descriptor for a softmax backward propagation primitive.
dnnl_softmax_forward_primitive_desc_create
Creates a primitive descriptor for a softmax forward propagation primitive.
dnnl_stream_create
Creates an execution stream.
dnnl_stream_destroy
Destroys an execution stream.
dnnl_stream_get_engine
Returns the engine of a stream object.
dnnl_stream_wait
Waits for all primitives in the execution stream to finish computations.
dnnl_sum_primitive_desc_create
Creates a primitive descriptor for an (out-of-place) sum primitive.
dnnl_vanilla_rnn_backward_primitive_desc_create
Creates a primitive descriptor for vanilla RNN backward propagation primitive.
dnnl_vanilla_rnn_forward_primitive_desc_create
Creates a primitive descriptor for vanilla RNN forward propagation primitive.
dnnl_version
Returns library version information. @returns Pointer to a constant structure containing

Type Aliases§

const_dnnl_graph_allocator_t
A constant allocator handle.
const_dnnl_graph_compiled_partition_t
A constant compiled partition handle.
const_dnnl_graph_graph_t
A constant graph handle.
const_dnnl_graph_op_t
A constant operation handle.
const_dnnl_graph_partition_t
A constant partition handle.
const_dnnl_graph_tensor_t
A constant tensor handle.
const_dnnl_memory_desc_t
A memory descriptor handle.
const_dnnl_memory_t
A constant memory handle.
const_dnnl_post_ops_t
@brief A constant post operation chain handle.
const_dnnl_primitive_attr_t
@brief A constant primitive descriptor attributes handle.
const_dnnl_primitive_desc_t
@brief A constant primitive descriptor handle.
const_dnnl_primitive_t
A constant primitive handle.
const_dnnl_stream_t
A constant execution stream handle.
dnnl_dim_t
A type to describe tensor dimension.
dnnl_dims_t
A type to describe tensor dimensions.
dnnl_engine_t
@brief An engine handle.
dnnl_graph_allocator_t
An allocator handle.
dnnl_graph_compiled_partition_t
A compiled partition handle.
dnnl_graph_graph_t
A graph handle.
dnnl_graph_host_allocate_f
Allocation call-back function interface for host. For SYCL allocator, see #dnnl_graph_sycl_allocate_f.
dnnl_graph_host_deallocate_f
Deallocation call-back function interface for host. For SYCL allocator, see #dnnl_graph_sycl_deallocate_f.
dnnl_graph_op_t
An operation handle.
dnnl_graph_partition_t
A partition handle.
dnnl_graph_tensor_t
A tensor handle.
dnnl_memory_desc_t
A memory descriptor handle.
dnnl_memory_t
A memory handle.
dnnl_post_ops_t
@brief A post operation chain handle.
dnnl_primitive_attr_t
@brief A primitive descriptor attributes handle that controls primitive behavior.
dnnl_primitive_desc_t
@brief A primitive descriptor handle.
dnnl_primitive_t
A primitive handle.
dnnl_stream_t
An execution stream handle.

Unions§

_bindgen_ty_1
@cond DO_NOT_DOCUMENT_THIS Hex representation for a special quiet NAN (!= NAN from math.h)
dnnl_graph_logical_tensor_t__bindgen_ty_1