Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[WIP][DRAFT][onert] Optimized BatchMatMul in CPU backend #13919

Draft
wants to merge 1 commit into
base: master
Choose a base branch
from
Draft

[WIP][DRAFT][onert] Optimized BatchMatMul in CPU backend #13919

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
55 changes: 41 additions & 14 deletions compute/cker/include/cker/operation/reference/BatchMatMul.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,7 @@

#include "cker/Types.h"
#include "cker/Shape.h"
#include "cker/operation/optimized/Gemm.h"

namespace nnfw
{
Expand Down Expand Up @@ -73,7 +74,7 @@ inline void BatchMatMul(const Shape &lhs_shape, const float *lhs_data, const Sha
// Set params for each matrix multiply.
const int lhs_rows = extended_lhs_shape.Dims(3);
const int rhs_cols = extended_rhs_shape.Dims(4);
const int accum_depth = extended_lhs_shape.Dims(4);
// const int accum_depth = extended_lhs_shape.Dims(4);

for (int b0 = 0; b0 < batch_dim0; ++b0)
{
Expand All @@ -89,19 +90,45 @@ inline void BatchMatMul(const Shape &lhs_shape, const float *lhs_data, const Sha
const float *rhs_ptr2 = rhs_ptr1 + b2 * rhs_ext2;
float *out_ptr = output_data + ((b0 * batch_dim1 * batch_dim2) + b1 * batch_dim2 + b2) *
lhs_rows * rhs_cols;
for (int j = 0; j < rhs_cols; ++j)
{
for (int i = 0; i < lhs_rows; ++i)
{
float total = 0.f;
for (int k = 0; k < accum_depth; ++k)
{
total += lhs_ptr2[accum_depth * i + k] * rhs_ptr2[j * accum_depth + k];
}
int idx = lhs_rows * j + i;
out_ptr[idx] = total;
}
}

MatrixParams<float> rhs_params;
rhs_params.order = Order::kColMajor; // should be always like this? base it of adj_x & adj_y?
rhs_params.rows = rhs_cols;
rhs_params.cols = lhs_rows;
// How to determine this?
rhs_params.cache_policy = nnfw::cker::optimized::DefaultCachePolicy(false);

MatrixParams<float> lhs_params;
lhs_params.order = Order::kRowMajor; // should be always like this? base it of adj_x & adj_y?
lhs_params.rows = lhs_rows;
lhs_params.cols = rhs_cols;
// How to determine this?
lhs_params.cache_policy = nnfw::cker::optimized::DefaultCachePolicy(false);

MatrixParams<float> dst_params;
dst_params.order = Order::kColMajor;
dst_params.rows = lhs_rows;
dst_params.cols = rhs_cols;

GemmParams<float, float> gemm_params;

nnfw::cker::optimized::Gemm(lhs_params, lhs_ptr2, rhs_params, rhs_ptr2, dst_params, out_ptr,
gemm_params);

// for (int j = 0; j < rhs_cols; ++j)
// {
// for (int i = 0; i < lhs_rows; ++i)
// {
// float total = 0.f;
// for (int k = 0; k < accum_depth; ++k)
// {
// total += lhs_ptr2[accum_depth * i + k] * rhs_ptr2[j * accum_depth + k];
// }
// int idx = lhs_rows * j + i;
// out_ptr[idx] = total;
// }
// }

}
}
}
Expand Down
Loading