ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 21:21:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

Enable gemm_dl and other kernels on Navi3x. (#714)

Browse Source 
* enable dl kernels on navi3

* do not build xdl tests and examples on Navi

* run tests before building everything on jenkins

* disable gemm_bilinear on gfx1030

* add gpu targets to installer on Navi

* put tests in the same order as before

* reduce the number of navi targets in CI

* build CI installed for gfx940 as well

* only build for MI300 during QA runs
This commit is contained in:
Illia Silin
2023-05-23 09:23:16 -07:00
committed by GitHub
parent 3cff340423
commit d821d1e54f
39 changed files with 291 additions and 210 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
include/ck/tensor_operation/gpu/device/impl/device_convnd_bwd_data_nwc_kxc_nwk_dl.hpp
View File

@@ -1393,7 +1393,9 @@ struct DeviceConvNdBwdDataNwcKxcNwk_Dl
static bool IsSupportedArgument(const Argument& arg)
{
// check device
if(!(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030"))
if(!(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030" ||
ck::get_device_name() == "gfx1100" || ck::get_device_name() == "gfx1101" ||
ck::get_device_name() == "gfx1102"))
{
return false;
}

4
include/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp
View File

@@ -485,7 +485,9 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
static bool IsSupportedArgument(const Argument& arg)
{
if(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030")
if(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030" ||
ck::get_device_name() == "gfx1100" || ck::get_device_name() == "gfx1101" ||
ck::get_device_name() == "gfx1102")
{
return GridwiseGemm::CheckValidity(
arg.a_grid_desc_k0_m_k1_, arg.b_grid_desc_k0_n_k1_, arg.c_grid_desc_m_n_);

8
include/ck/tensor_operation/gpu/device/impl/device_gemm_multiple_d_dl.hpp
View File

@@ -50,8 +50,9 @@ __global__ void
const CGridDesc_M0_M10_M11_N0_N10_N11 e_grid_desc_m0_m10_m11_n0_n10_n11,
const Block2CTileMap block_2_ctile_map)
{
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx906__) || defined(__gfx908__) || \
defined(__gfx90a__) || defined(__gfx940__) || defined(__gfx1030__))
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx906__) || defined(__gfx908__) || \
defined(__gfx90a__) || defined(__gfx940__) || defined(__gfx1030__) || defined(__gfx1100__) || \
defined(__gfx1101__) || defined(__gfx1102__))
constexpr index_t shared_block_size =
GridwiseGemm::GetSharedMemoryNumberOfByte() / sizeof(ABDataType);
@@ -553,7 +554,8 @@ struct DeviceGemmMultipleD_Dl : public DeviceGemmMultipleD<ALayout,
{
if(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx908" ||
ck::get_device_name() == "gfx90a" || ck::get_device_name() == "gfx1030" ||
ck::get_device_name() == "gfx940")
ck::get_device_name() == "gfx940" || ck::get_device_name() == "gfx1100" ||
ck::get_device_name() == "gfx1101" || ck::get_device_name() == "gfx1102")
{
return GridwiseGemm::CheckValidity(
arg.a_grid_desc_k0_m_k1_, arg.b_grid_desc_k0_n_k1_, arg.e_grid_desc_m_n_);

4
include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_gnwc_gkxc_gnwk_dl.hpp
View File

@@ -1027,7 +1027,9 @@ struct DeviceGroupedConvBwdWeightGnwcGkxcGnwk_Dl
static bool IsSupportedArgument(const Argument& arg)
{
// check device
if(!(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030"))
if(!(ck::get_device_name() == "gfx906" || ck::get_device_name() == "gfx1030" ||
ck::get_device_name() == "gfx1100" || ck::get_device_name() == "gfx1101" ||
ck::get_device_name() == "gfx1102"))
{
return false;
}

5
include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_multiple_d_dl.hpp
View File

@@ -39,8 +39,9 @@ __global__ void
const BElementwiseOperation b_element_op,
const CDEElementwiseOperation cde_element_op)
{
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx906__) || defined(__gfx908__) || \
defined(__gfx90a__) || defined(__gfx1030__))
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx906__) || defined(__gfx908__) || \
defined(__gfx90a__) || defined(__gfx1030__) || defined(__gfx1100__) || defined(__gfx1101__) || \
defined(__gfx1102__) || defined(__gfx940__))
__shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
const index_t block_id = get_block_1d_id();