Developer Guide and Reference

  • 2023.0
  • 12/16/2022
  • Public Content
Contents

enum dnnl::primitive::kind

Overview

Kinds of primitives supported by the library. More…
#include <dnnl.hpp> enum kind { undef = dnnl_undefined_primitive, reorder = dnnl_reorder, shuffle = dnnl_shuffle, concat = dnnl_concat, sum = dnnl_sum, convolution = dnnl_convolution, deconvolution = dnnl_deconvolution, eltwise = dnnl_eltwise, lrn = dnnl_lrn, batch_normalization = dnnl_batch_normalization, inner_product = dnnl_inner_product, rnn = dnnl_rnn, binary = dnnl_binary, matmul = dnnl_matmul, resampling = dnnl_resampling, pooling = dnnl_pooling, reduction = dnnl_reduction, prelu = dnnl_prelu, softmax = dnnl_softmax, layer_normalization = dnnl_layer_normalization, };

Detailed Documentation

Kinds of primitives supported by the library.
Enum Values
undef
Undefined primitive.
reorder
A reorder primitive.
shuffle
A shuffle primitive.
concat
A (out-of-place) tensor concatenation primitive.
sum
A summation primitive.
convolution
A convolution primitive.
deconvolution
A deconvolution primitive.
eltwise
An element-wise primitive.
lrn
An LRN primitive.
batch_normalization
A batch normalization primitive.
inner_product
An inner product primitive.
rnn
An RNN primitive.
binary
A binary primitive.
matmul
A matmul (matrix multiplication) primitive.
resampling
A resampling primitive.
pooling
A pooling primitive.
reduction
A reduction primitive.
prelu
A PReLU primitive.
softmax
A softmax primitive.
layer_normalization
A layer normalization primitive.

Product and Performance Information

1

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