Developer Guide and Reference

  • 2022.1
  • 04/11/2022
  • Public Content
Contents

enum dnnl_query_t

Overview

Primitive descriptor query specification. More…
#include <dnnl_types.h> enum dnnl_query_t { dnnl_query_undef = 0, dnnl_query_engine, dnnl_query_primitive_kind, dnnl_query_num_of_inputs_s32, dnnl_query_num_of_outputs_s32, dnnl_query_time_estimate_f64, dnnl_query_memory_consumption_s64, dnnl_query_scratchpad_engine, dnnl_query_impl_info_str, dnnl_query_reorder_src_engine, dnnl_query_reorder_dst_engine, dnnl_query_prop_kind, dnnl_query_cache_blob_id_size_s64, dnnl_query_cache_blob_id, dnnl_query_some_d = 64, dnnl_query_op_d, dnnl_query_convolution_d, dnnl_query_deconvolution_d, dnnl_query_shuffle_d, dnnl_query_eltwise_d, dnnl_query_softmax_d, dnnl_query_pooling_d, dnnl_query_lrn_d, dnnl_query_batch_normalization_d, dnnl_query_layer_normalization_d, dnnl_query_inner_product_d, dnnl_query_rnn_d, dnnl_query_gemm_d, dnnl_query_binary_d, dnnl_query_logsoftmax_d, dnnl_query_matmul_d, dnnl_query_resampling_d, dnnl_query_pooling_v2_d, dnnl_query_reduction_d, dnnl_query_prelu_d, dnnl_query_softmax_v2_d, dnnl_query_some_md = 128, dnnl_query_src_md, dnnl_query_diff_src_md, dnnl_query_weights_md, dnnl_query_diff_weights_md, dnnl_query_dst_md, dnnl_query_diff_dst_md, dnnl_query_workspace_md, dnnl_query_scratchpad_md, dnnl_query_exec_arg_md = 255, dnnl_query_max = 0x7fff, };

Detailed Documentation

Primitive descriptor query specification.
For generic function dnnl_primitive_desc_query(), the type of result must agree with the queried argument. The correspondence table:
Query kind
Type of query result
dnnl_query_*_s32
int *
dnnl_query_*_s64
dnnl_dim_t * (same as int64_t )
dnnl_query_*_f64
double *
dnnl_query_*_str
const char **
dnnl_query_*_md
dnnl_query_*_<op>_d
const dnnl_<op>_desc_t **
dnnl_query_*_pd
dnnl_query_cache_blob_id
const uint8_t **
Rule of thumb: all opaque types and structures are returned by reference. All numbers are returned by value.
All returned references point to constant objects and are valid only during the lifetime of the queried primitive descriptor. Returned objects must not be destroyed by the user. If you need to keep the object longer than the lifetime of the queried primitive descriptor, use dnnl_primitive_desc_clone() to make a copy.
Enum Values
dnnl_query_undef
no query
dnnl_query_engine
execution engine
dnnl_query_primitive_kind
primitive kind
dnnl_query_num_of_inputs_s32
number of inputs expected
dnnl_query_num_of_outputs_s32
number of outputs expected
dnnl_query_time_estimate_f64
runtime estimation (seconds)
dnnl_query_memory_consumption_s64
memory consumption extra
dnnl_query_scratchpad_engine
(scratch) memory, additional to all inputs and outputs memory (bytes)
scratchpad engine engine to be used
dnnl_query_impl_info_str
for creating scratchpad memory
implementation name
dnnl_query_reorder_src_engine
source engine
dnnl_query_reorder_dst_engine
destination engine
dnnl_query_prop_kind
propagation kind
dnnl_query_cache_blob_id_size_s64
size of cache blob ID in bytes
dnnl_query_cache_blob_id
cache blob ID (pointer to array)
dnnl_query_some_d
stub
dnnl_query_op_d
op descriptor
dnnl_query_convolution_d
convolution descriptor
dnnl_query_deconvolution_d
deconvolution descriptor
dnnl_query_shuffle_d
shuffle descriptor
dnnl_query_eltwise_d
eltwise descriptor
dnnl_query_softmax_d
softmax descriptor
dnnl_query_pooling_d
pooling descriptor
dnnl_query_lrn_d
lrn descriptor
dnnl_query_batch_normalization_d
batch normalization descriptor
dnnl_query_layer_normalization_d
layer normalization descriptor
dnnl_query_inner_product_d
inner product descriptor
dnnl_query_rnn_d
rnn descriptor
dnnl_query_gemm_d
GEMM descriptor (internal)
dnnl_query_binary_d
binary descriptor
dnnl_query_logsoftmax_d
logsoftmax descriptor
dnnl_query_matmul_d
matrix multiplication (matmul) descriptor
dnnl_query_resampling_d
resampling descriptor
dnnl_query_pooling_v2_d
pooling version 2 descriptor
dnnl_query_reduction_d
reduction descriptor
dnnl_query_prelu_d
prelu descriptor
dnnl_query_softmax_v2_d
softmax version 2 descriptor
dnnl_query_some_md
stub
dnnl_query_src_md
source memory desc
dnnl_query_diff_src_md
source gradient memory desc
dnnl_query_weights_md
weights memory descriptor desc
dnnl_query_diff_weights_md
weights grad. memory desc
dnnl_query_dst_md
destination memory desc
dnnl_query_diff_dst_md
destination grad. memory desc
dnnl_query_workspace_md
workspace memory desc
dnnl_query_scratchpad_md
scratchpad memory desc
dnnl_query_exec_arg_md
memory desc of an execute argument

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.