saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-05-14 20:48:00 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Revert Softmax optimizations using openmp. (#97)

Browse source 
Revert "Simpler unused parameter in #if defined() switch."
This reverts commit 6b00e6bb4d.

Revert "Better opemmp parallel group count calculation in Softmax parallel running."
This reverts commit c530064ebe.

Revert "Optimize softmax cpu by parallel using openmp."
This reverts commit e7bdfa00db.
This commit is contained in:
Zhang Lei 2018-12-04 16:21:27 -08:00 committed by GitHub
parent 7f0e947f96
commit 996d6ea4cd
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 8 additions and 50 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

58
onnxruntime/core/providers/cpu/math/softmax_shared.cc
View file

@ -36,47 +36,8 @@
#include "gsl/gsl_algorithm"
#include "gsl/gsl_util"
#if defined(_OPENMP)
#include <omp.h>
#endif
namespace onnxruntime {
common::Status SoftmaxCore(const int n,
const int d,
const float* Xdata,
float* Ydata,
const float* sum_multiplier,
float* rowmax) {
const int nd = n * d;
math::RowwiseMax<float, CPUMathUtil>(n, d, Xdata, rowmax, nullptr);
// Put the intermediate result X - max(X) into Y by first copying X to Y, and then subtracting max from each entry
gsl::copy(gsl::make_span(Xdata, nd), gsl::make_span(Ydata, nd));
math::Gemm<float, CPUMathUtil>(CblasNoTrans, CblasNoTrans, n, d, 1, -1, rowmax, sum_multiplier, 1, Ydata, nullptr);
// Exponentiation
math::Exp<float, CPUMathUtil>(nd, Ydata, Ydata, nullptr);
return Status::OK();
}
static int GetParallelGroupCount(int n, int d) {
#if defined(_OPENMP)
int omp_num_threads = omp_get_num_threads();
int group_count = std::min(omp_num_threads, n);
if (group_count <= 1) return 1;
// 2048 * sizeof(float) is size of 2 cache page
static const int min_elements_per_group = 2048;
int max_groups = gsl::narrow_cast<int>((int64_t{n} * d + min_elements_per_group-1) / min_elements_per_group);
return std::min(group_count, max_groups);
#else
(void)n;
(void)d;
return 1;
#endif
}
common::Status SoftmaxCPU(const int64_t N,
const int64_t D,
const float* Xdata,
@ -96,24 +57,21 @@ common::Status SoftmaxCPU(const int64_t N,
const int n = gsl::narrow_cast<int>(N);
const int d = gsl::narrow_cast<int>(D);
const int nd = gsl::narrow_cast<int>(N * D);
int parallel_group_count = GetParallelGroupCount(n, d);
int n_per_group = (n + (parallel_group_count-1)) / parallel_group_count;
math::RowwiseMax<float, CPUMathUtil>(n, d, Xdata, rowmax, nullptr);
#pragma omp parallel for
for (int i = 0; i < parallel_group_count; ++i) {
int s = n_per_group * i;
if (s < n) {
int c = (n - s >= n_per_group) ? n_per_group : (n-s);
SoftmaxCore(c, d, Xdata + (s*d), Ydata + (s*d), sum_multiplier, rowmax+s);
}
}
// Put the intermediate result X - max(X) into Y by first copying X to Y, and then subtracting max from each entry
gsl::copy(gsl::make_span(Xdata, nd), gsl::make_span(Ydata, nd));
math::Gemm<float, CPUMathUtil>(CblasNoTrans, CblasNoTrans, n, d, 1, -1, rowmax, sum_multiplier, 1, Ydata, nullptr);
// Exponentiation
math::Exp<float, CPUMathUtil>(nd, Ydata, Ydata, nullptr);
math::Gemv<float, CPUMathUtil>(CblasNoTrans, n, d, 1, Ydata, sum_multiplier, 0, scale, nullptr);
// Do division
if (!logarithmic) {
#pragma omp parallel for
for (int i = 0; i < N; ++i) {
for (int j = 0; j < D; ++j) {
Ydata[i * D + j] /= scale[i];