ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-18 20:09:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

[CK_TILE] Move hipmalloc/memcpy calls out of gpu reference gemm (#1743)

Browse Source 
* [CK_TILE] Move hipmalloc/memcpy calls out of gpu reference gemm

* [CK_TILE] Move hipmalloc/memcpy calls out of gpu reference gemm - review changes

* [CK_TILE] Move hipmalloc/memcpy calls out of gpu reference gemm - review fix

[ROCm/composable_kernel commit: f6c4d614e3]
This commit is contained in:
aledudek
2024-12-18 09:45:58 +01:00
committed by GitHub
parent 521a2796d0
commit f5727b5f11
3 changed files with 68 additions and 156 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
example/ck_tile/03_gemm/run_gemm_example.inc
View File

@@ -161,14 +161,39 @@ int run_gemm_example_with_layouts(int argc,
c_m_n_gpu_ref.SetZero();
c_m_n_gpu_buf_ref.SetZero();
ADataType* d_A;
BDataType* d_B;
CDataType* d_C;
ck_tile::hip_check_error(hipMalloc(&d_A, M * K * sizeof(ADataType)));
ck_tile::hip_check_error(hipMalloc(&d_B, N * K * sizeof(BDataType)));
ck_tile::hip_check_error(hipMalloc(&d_C, M * N * sizeof(CDataType)));
ck_tile::hip_check_error(hipMemcpy(d_A,
a_m_k_dev_buf.GetDeviceBuffer(),
M * K * sizeof(ADataType),
hipMemcpyHostToDevice));
ck_tile::hip_check_error(hipMemcpy(d_B,
b_k_n_dev_buf.GetDeviceBuffer(),
N * K * sizeof(BDataType),
hipMemcpyHostToDevice));
ck_tile::reference_gemm_gpu<ADataType,
BDataType,
AccDataType,
CDataType,
ALayout,
BLayout,
CLayout>(
a_m_k_dev_buf, b_k_n_dev_buf, c_m_n_gpu_buf_ref, M, N, K, stride_A, stride_B, stride_C);
CLayout>(d_A, d_B, d_C, M, N, K, stride_A, stride_B, stride_C);
ck_tile::hip_check_error(hipMemcpy(c_m_n_gpu_buf_ref.GetDeviceBuffer(),
d_C,
M * N * sizeof(CDataType),
hipMemcpyDeviceToHost));
ck_tile::hip_check_error(hipFree(d_A));
ck_tile::hip_check_error(hipFree(d_B));
ck_tile::hip_check_error(hipFree(d_C));
c_m_n_gpu_buf_ref.FromDevice(c_m_n_gpu_ref.data());
pass = ck_tile::check_err(c_m_n_dev_result, c_m_n_gpu_ref);

33
example/ck_tile/16_batched_gemm/run_batched_gemm_example.inc
View File

@@ -188,15 +188,33 @@ int run_batched_gemm_example_with_layouts(int argc,
c_m_n_gpu_ref.SetZero();
c_m_n_gpu_buf_ref.SetZero();
ADataType* d_A;
BDataType* d_B;
CDataType* d_C;
ck_tile::hip_check_error(hipMalloc(&d_A, batch_count * M * K * sizeof(ADataType)));
ck_tile::hip_check_error(hipMalloc(&d_B, batch_count * N * K * sizeof(BDataType)));
ck_tile::hip_check_error(hipMalloc(&d_C, batch_count * M * N * sizeof(CDataType)));
ck_tile::hip_check_error(hipMemcpy(d_A,
a_m_k_dev_buf.GetDeviceBuffer(),
batch_count * M * K * sizeof(ADataType),
hipMemcpyHostToDevice));
ck_tile::hip_check_error(hipMemcpy(d_B,
b_k_n_dev_buf.GetDeviceBuffer(),
batch_count * N * K * sizeof(BDataType),
hipMemcpyHostToDevice));
ck_tile::reference_batched_gemm_gpu<ADataType,
BDataType,
AccDataType,
CDataType,
ALayout,
BLayout,
CLayout>(a_m_k_dev_buf,
b_k_n_dev_buf,
c_m_n_gpu_buf_ref,
CLayout>(d_A,
d_B,
d_C,
M,
N,
K,
@@ -208,6 +226,15 @@ int run_batched_gemm_example_with_layouts(int argc,
batch_stride_C,
batch_count);
ck_tile::hip_check_error(hipMemcpy(c_m_n_gpu_buf_ref.GetDeviceBuffer(),
d_C,
batch_count * M * N * sizeof(CDataType),
hipMemcpyDeviceToHost));
ck_tile::hip_check_error(hipFree(d_A));
ck_tile::hip_check_error(hipFree(d_B));
ck_tile::hip_check_error(hipFree(d_C));
c_m_n_gpu_buf_ref.FromDevice(c_m_n_gpu_ref.data());
pass = ck_tile::check_err(c_m_n_dev_result, c_m_n_gpu_ref);