Developer Guide and Reference

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

enum dnnl::query

Overview

Primitive descriptor query specification. More…
#include <dnnl.hpp> enum query { undef = dnnl_query_undef, engine = dnnl_query_engine, primitive_kind = dnnl_query_primitive_kind, num_of_inputs_s32 = dnnl_query_num_of_inputs_s32, num_of_outputs_s32 = dnnl_query_num_of_outputs_s32, time_estimate_f64 = dnnl_query_time_estimate_f64, memory_consumption_s64 = dnnl_query_memory_consumption_s64, scratchpad_engine = dnnl_query_scratchpad_engine, reorder_src_engine = dnnl_query_reorder_src_engine, reorder_dst_engine = dnnl_query_reorder_dst_engine, impl_info_str = dnnl_query_impl_info_str, prop_kind = dnnl_query_prop_kind, cache_blob_id_size_s64 = dnnl_query_cache_blob_id_size_s64, cache_blob_id = dnnl_query_cache_blob_id, op_d = dnnl_query_op_d, convolution_d = dnnl_query_convolution_d, deconvolution_d = dnnl_query_deconvolution_d, shuffle_d = dnnl_query_shuffle_d, eltwise_d = dnnl_query_eltwise_d, softmax_d = dnnl_query_softmax_d, pooling_d = dnnl_query_pooling_d, lrn_d = dnnl_query_lrn_d, batch_normalization_d = dnnl_query_batch_normalization_d, layer_normalization_d = dnnl_query_layer_normalization_d, inner_product_d = dnnl_query_inner_product_d, rnn_d = dnnl_query_rnn_d, binary_d = dnnl_query_binary_d, logsoftmax_d = dnnl_query_logsoftmax_d, matmul_d = dnnl_query_matmul_d, resampling_d = dnnl_query_resampling_d, reduction_d = dnnl_query_reduction_d, src_md = dnnl_query_src_md, diff_src_md = dnnl_query_diff_src_md, weights_md = dnnl_query_weights_md, diff_weights_md = dnnl_query_diff_weights_md, dst_md = dnnl_query_dst_md, diff_dst_md = dnnl_query_diff_dst_md, workspace_md = dnnl_query_workspace_md, scratchpad_md = dnnl_query_scratchpad_md, exec_arg_md = dnnl_query_exec_arg_md, };

Detailed Documentation

Primitive descriptor query specification.
In general, queries are not used with the C++ API because most queries are implemented as class members.
See dnnl_query_t for more information.
Enum Values
undef
no query
engine
execution engine
primitive_kind
primitive kind
num_of_inputs_s32
number of inputs expected
num_of_outputs_s32
number of outputs expected
time_estimate_f64
runtime estimation (seconds), unimplemented
memory_consumption_s64
memory required for scratchpad (bytes)
@sa @ref dev_guide_attributes_scratchpad
scratchpad_engine
scratchpad engine
engine to be used for creating scratchpad memory
reorder_src_engine
reorder source engine
reorder_dst_engine
reorder destination engine
impl_info_str
implementation name
prop_kind
propagation kind
cache_blob_id_size_s64
size of cache blob ID in bytes
cache_blob_id
cache blob ID (pointer to array)
op_d
operation descriptor
convolution_d
convolution descriptor
deconvolution_d
deconvolution descriptor
shuffle_d
shuffle descriptor
eltwise_d
eltwise descriptor
softmax_d
softmax descriptor
pooling_d
pooling descriptor
lrn_d
lrn descriptor
batch_normalization_d
batch normalization descriptor
layer_normalization_d
layer normalization descriptor
inner_product_d
inner product descriptor
rnn_d
rnn descriptor
binary_d
binary descriptor
logsoftmax_d
logsoftmax descriptor
matmul_d
matmul descriptor
resampling_d
resampling descriptor
reduction_d
reduction descriptor
src_md
source memory desc
diff_src_md
source gradient (diff) memory desc
weights_md
weights memory descriptor desc
diff_weights_md
weights gradient (diff) memory desc
dst_md
destination memory desc
diff_dst_md
destination gradient (diff) memory desc
workspace_md
workspace memory desc
scratchpad_md
scratchpad memory desc
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.