wu-arch/kernels
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

sgemm_tcore: Support fp16 input generation in host code

Browse Source
This commit is contained in:
Hansung Kim
2024-07-29 17:18:35 -07:00
parent bca53a9c76
commit 5f342914bd
2 changed files with 4061 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4018
tests/regression/sgemm_tcore/half.hpp Normal file
View File

File diff suppressed because it is too large Load Diff

61
tests/regression/sgemm_tcore/main.cpp
View File

@@ -4,7 +4,12 @@
#include <string.h>
#include <vortex.h>
#include <vector>
#include <cassert>
#include "common.h"
#include "half.hpp"
using half_float::half;
using half_float::half_cast;
#define RT_CHECK(_expr) \
do { \
@@ -21,8 +26,8 @@
const char* kernel_file = "kernel.bin";
uint32_t count = 0;
std::vector<float> src_a_data;
std::vector<float> src_b_data;
template <typename T> std::vector<T> src_a_data;
template <typename T> std::vector<T> src_b_data;
std::vector<float> ref_data;
vx_device_h device = nullptr;
@@ -65,28 +70,45 @@ void cleanup() {
}
}
template <typename T>
void generate_source_matrix(uint32_t dim_m, uint32_t dim_n, uint32_t dim_k) {
src_a_data.resize(dim_m * dim_k);
src_b_data.resize(dim_k * dim_n);
static_assert(std::is_same_v<half, T> || std::is_same_v<float, T>,
"unsupported floating point datatype");
for (uint32_t i = 0; i < src_a_data.size(); ++i) {
src_a_data[i] = static_cast<float>(i);
std::cout << "A: " << i << ": value=" << src_a_data[i] << std::endl;
src_a_data<T>.resize(dim_m * dim_k);
src_b_data<T>.resize(dim_k * dim_n);
for (uint32_t i = 0; i < src_a_data<T>.size(); ++i) {
if constexpr (std::is_same_v<half, T>) {
src_a_data<T>[i] = half_cast<half>(static_cast<float>(i));
} else if (std::is_same_v<float, T>) {
src_a_data<T>[i] = static_cast<float>(i);
}
std::cout << "A: " << i << ": value=" << src_a_data<T>[i] << std::endl;
}
for (uint32_t i = 0; i < src_b_data.size(); ++i) {
src_b_data[i] = static_cast<float>(i);
std::cout << "B: " << i << ": value=" << src_b_data[i] << std::endl;
for (uint32_t i = 0; i < src_b_data<T>.size(); ++i) {
if constexpr (std::is_same_v<half, T>) {
src_b_data<T>[i] = half_cast<half>(static_cast<float>(i));
} else if (std::is_same_v<float, T>) {
src_b_data<T>[i] = static_cast<float>(i);
}
std::cout << "B: " << i << ": value=" << src_b_data<T>[i] << std::endl;
}
}
template <typename T>
void generate_reference_matmul(uint32_t dim_m, uint32_t dim_n, uint32_t dim_k) {
static_assert(std::is_same_v<half, T> || std::is_same_v<float, T>,
"unsupported floating point datatype");
ref_data.resize(dim_m * dim_n);
for (uint32_t i = 0; i < dim_m; ++i) {
for (uint32_t j = 0; j < dim_n; ++j) {
float ref = 0.0f;
for (uint32_t k = 0; k < dim_k; ++k) {
ref += src_a_data[dim_k * i + k] * src_b_data[dim_n * k + j];
ref += static_cast<float>(src_a_data<T>[dim_k * i + k]) *
static_cast<float>(src_b_data<T>[dim_n * k + j]);
}
ref_data.at(dim_n * i + j) = ref;
}
@@ -151,8 +173,9 @@ int main(int argc, char *argv[]) {
uint32_t dim_n = 64;
uint32_t dim_k = 64;
generate_source_matrix(dim_m, dim_n, dim_k);
generate_reference_matmul(dim_m, dim_n, dim_k);
using float_type = half;
generate_source_matrix<float_type>(dim_m, dim_n, dim_k);
generate_reference_matmul<float_type>(dim_m, dim_n, dim_k);
std::cout << "write reference output" << std::endl;
std::ofstream ref_file("reference.c.bin", std::ios::binary | std::ios::out);
@@ -163,9 +186,9 @@ int main(int argc, char *argv[]) {
ref_file.write(reinterpret_cast<char *>(ref_data.data()), ref_data.size() * sizeof(ref_data[0]));
ref_file.close();
uint32_t src_a_buf_size = src_a_data.size() * sizeof(src_a_data[0]);
uint32_t src_b_buf_size = src_b_data.size() * sizeof(src_b_data[0]);
uint32_t dst_buf_size = ref_data.size() * sizeof(src_a_data[0]);
uint32_t src_a_buf_size = src_a_data<float_type>.size() * sizeof(src_a_data<float_type>[0]);
uint32_t src_b_buf_size = src_b_data<float_type>.size() * sizeof(src_b_data<float_type>[0]);
uint32_t dst_buf_size = ref_data.size() * sizeof(src_a_data<float_type>[0]);
std::cout << "buffer size: " << dst_buf_size << " bytes" << std::endl;
@@ -225,7 +248,8 @@ int main(int argc, char *argv[]) {
{
{
auto buf_ptr = staging_buf.data();
memcpy(buf_ptr, src_a_data.data(), src_a_data.size() * sizeof(float));
memcpy(buf_ptr, src_a_data<float_type>.data(),
src_a_data<float_type>.size() * sizeof(float_type));
RT_CHECK(vx_copy_to_dev(device, kernel_arg.addr_a, staging_buf.data(),
src_a_buf_size));
@@ -242,7 +266,8 @@ int main(int argc, char *argv[]) {
}
{
auto buf_ptr = staging_buf.data();
memcpy(buf_ptr, src_b_data.data(), src_b_data.size() * sizeof(float));
memcpy(buf_ptr, src_b_data<float_type>.data(),
src_b_data<float_type>.size() * sizeof(float_type));
RT_CHECK(vx_copy_to_dev(device, kernel_arg.addr_b, staging_buf.data(),
src_b_buf_size));