From 8f3ef9aa3fc287f5195ce5acb8e7b358dd0f0b20 Mon Sep 17 00:00:00 2001
From: Zhang Lei <zhalei@microsoft.com>
Date: Tue, 4 Dec 2018 15:40:05 -0800
Subject: [PATCH] Revert Softmax optimizations using openmp. Revert "Simpler
 unused parameter in #if defined() switch." This reverts commit
 6b00e6bb4df23e77da73b72c67ff98ba605712f1.

Revert "Better opemmp parallel group count calculation in Softmax parallel running."
This reverts commit c530064ebe912621ac5627bc486c62356fbd1eeb.

Revert "Optimize softmax cpu by parallel using openmp."
This reverts commit e7bdfa00db1d0e9909f1d0d5e159173e3fbd608f.
---
 .../core/providers/cpu/math/softmax_shared.cc | 58 +++----------------
 1 file changed, 8 insertions(+), 50 deletions(-)

diff --git a/onnxruntime/core/providers/cpu/math/softmax_shared.cc b/onnxruntime/core/providers/cpu/math/softmax_shared.cc
index 18f077d6c127d..32df249f362a0 100644
--- a/onnxruntime/core/providers/cpu/math/softmax_shared.cc
+++ b/onnxruntime/core/providers/cpu/math/softmax_shared.cc
@@ -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];