From 013155420c89f623f5d6510287e57eeb74d6a133 Mon Sep 17 00:00:00 2001
From: JordanHanotiaux <103147288+JordanHanotiaux@users.noreply.github.com>
Date: Mon, 19 May 2025 10:06:55 +0200
Subject: [PATCH] up
---
matrix_opencl.cpp | 10 +++++-----
matrix_opencl.hpp | 5 +++--
2 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/matrix_opencl.cpp b/matrix_opencl.cpp
index c367b45..190932e 100644
--- a/matrix_opencl.cpp
+++ b/matrix_opencl.cpp
@@ -85,7 +85,7 @@ const std::string kernel_source_matrix_mul = R"(
}
}
)";
-const std::string kernel_source_matrix_mul_V2 = R"(
+const std::string kernel_source_fast_matrix_mul = R"(
__kernel void fast_matrix_mul(__global const float* A, __global const float* B, __global float* C, int A_rows, int A_cols, int B_cols) {
int row = get_global_id(0);
int col = get_global_id(1);
@@ -191,8 +191,8 @@ void KernelCache::compileKernels(cl::Context context, const std::vector<cl::Devi
cl::Program prog_matrix_mul = loadAndBuildProgram(context, devices, kernel_source_matrix_mul, "matrix_mul");
kernel_matrix_mul = cl::Kernel(prog_matrix_mul, "matrix_mul");
- cl::Program prog_matrix_mul_V2 = loadAndBuildProgram(context, devices, kernel_source_matrix_mul_V2, "matrix_mul_V2");
- kernel_matrix_mul_v2 = cl::Kernel(prog_matrix_mul_V2, "matrix_mul_V2");
+ cl::Program prog_fast_matrix_mul = loadAndBuildProgram(context, devices, kernel_source_fast_matrix_mul, "fast_matrix_mul");
+ kernel_fast_matrix_mul = cl::Kernel(prog_fast_matrix_mul, "fast_matrix_mul");
cl::Program prog_sigmoid = loadAndBuildProgram(context, devices, kernel_source_sigmoid, "sigmoid");
kernel_sigmoid = cl::Kernel(prog_sigmoid, "sigmoid");
@@ -377,10 +377,10 @@ MatrixCL MatrixCL::operator*(const MatrixCL& other) const {
return result;
}
-MatrixCL MatrixCL::matrix_mul_V2(const MatrixCL& other) const {
+MatrixCL MatrixCL::fast_matrix_mul(const MatrixCL& other) const {
MatrixCL result(rows_, other.numCols(), context_, queue_);
- cl::Kernel kernel = kernels_->kernel_matrix_mul_v2;
+ cl::Kernel kernel = kernels_->kernel_fast_matrix_mul;
kernel.setArg(0, buffer_);
kernel.setArg(1, other.getBuffer());
kernel.setArg(2, result.getBuffer());
diff --git a/matrix_opencl.hpp b/matrix_opencl.hpp
index 6e9d5df..07c1bab 100644
--- a/matrix_opencl.hpp
+++ b/matrix_opencl.hpp
@@ -25,7 +25,7 @@ struct KernelCache {
cl::Kernel kernel_sub_mul;
cl::Kernel kernel_transpose;
cl::Kernel kernel_matrix_mul;
- cl::Kernel kernel_matrix_mul_v2;
+ cl::Kernel kernel_fast_matrix_mul;
cl::Kernel kernel_sigmoid;
cl::Kernel kernel_sigmoid_backward;
cl::Kernel kernel_bce_elementwise;
@@ -94,7 +94,8 @@ public:
// Matrix multiplication: C = A * B
MatrixCL operator*(const MatrixCL& other) const;
- MatrixCL matrix_mul_V2(const MatrixCL& other) const;
+ // Fast matrix multiplication: C = A * B (optimized for large matrices)
+ MatrixCL fast_matrix_mul(const MatrixCL& other) const;
// Transpose: returns a new Matrix that is the transpose (B = A^T)
MatrixCL transpose() const;
--
GitLab