Unverified Commit cd167e49 authored by Chao Liu's avatar Chao Liu Committed by GitHub
Browse files

Compile for gfx908 and gfx90a (#130)

* adding compilation for multiple targets

* fix build

* clean

* update Jekinsfile

* update readme

* update Jenkins

* use ck::half_t instead of ushort for bf16

* rename enum classes

* clean

* rename

* clean
parent ecf337ba
......@@ -367,7 +367,7 @@ void device_convolution_backward_weight_implicit_gemm_v4r4r5_xdlops_atomic_nhwc_
TIn,
TAcc,
TWei,
InMemoryDataOperationEnum_t::AtomicAdd,
InMemoryDataOperationEnum::AtomicAdd,
decltype(out_gemmkbatch_gemmk0_gemmm_gemmk1_grid_desc),
decltype(in_gemmkbatch_gemmk0_gemmn_gemmk1_grid_desc),
decltype(wei_gemmm_gemmn_grid_desc),
......
......@@ -138,7 +138,7 @@ void device_convolution_forward_implicit_gemm_v4r4_dlops_nchw_kcyx_nkhw(
TInWei,
TAcc,
TOut,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(wei_gemmk_gemmm_grid_desc),
decltype(in_gemmk_gemmn_grid_desc),
decltype(out_gemmm_gemmn_grid_desc),
......
......@@ -202,7 +202,7 @@ void device_convolution_forward_implicit_gemm_v4r4r2_dlops_nhwc_kyxc_nhwk(
TInWei,
TAcc,
TOut,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(in_gemmk0_gemmm_gemmk1_grid_desc),
decltype(wei_gemmk0_gemmn_gemmk1_grid_desc),
decltype(out_gemmm_gemmn_grid_desc),
......
......@@ -167,7 +167,7 @@ void device_convolution_forward_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw(
TInWei,
TAcc,
TOut,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(wei_gemmk0_gemmm_gemmk1_grid_desc),
decltype(in_gemmk0_gemmn_gemmk1_grid_desc),
decltype(out_gemmm_gemmn_grid_desc),
......
......@@ -522,7 +522,7 @@ void device_convolution_forward_implicit_gemm_v4r4r4_xdlops_nhwc_kyxc_nhwk(
TInWei,
TAcc,
TOut,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(in_gemmk0_gemmm_gemmk1_grid_desc),
decltype(wei_gemmk0_gemmn_gemmk1_grid_desc),
decltype(out_gemmm_gemmn_grid_desc),
......
......@@ -6,7 +6,7 @@
template <typename TInWei,
typename TAcc,
typename TOut,
ck::ActivTypeEnum_t activ_type,
ck::ActivTypeEnum activ_type,
typename InLengths,
typename WeiLengths,
typename OutLengths,
......
......@@ -182,7 +182,7 @@ void device_convolution_forward_implicit_gemm_v6r1_dlops_nchw_kcyx_nkhw(
TInWei,
TAcc,
TOut,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(wei_grid_desc_gk0_gm0_gm1_gk1),
decltype(in_grid_desc_gk0_gn0_gn1_gk1),
decltype(out_grid_desc_gm0_gm1_gn0_gn1),
......
......@@ -6,7 +6,7 @@
template <typename TInWei,
typename TAcc,
typename TOut,
ck::ActivTypeEnum_t activ_type,
ck::ActivTypeEnum activ_type,
typename InLengths,
typename WeiLengths,
typename MaxLengths,
......
......@@ -398,7 +398,7 @@ void device_gemm_xdlops_km_kn_mn(const Tensor<ABType>& a_k_m,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -202,7 +202,7 @@ void device_gemm_xdlops_km_kn_nm(const Tensor<ABType>& a_k_m,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -398,7 +398,7 @@ void device_gemm_xdlops_km_nk_mn(const Tensor<ABType>& a_k_m,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -202,7 +202,7 @@ void device_gemm_xdlops_km_nk_nm(const Tensor<ABType>& a_k_m,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -398,7 +398,7 @@ void device_gemm_xdlops_mk_kn_mn(const Tensor<ABType>& a_m_k,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -230,7 +230,7 @@ void device_gemm_xdlops_mk_kn_nm(const Tensor<ABType>& a_m_k,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -499,7 +499,7 @@ void device_gemm_xdlops_mk_nk_mn(const Tensor<ABType>& a_m_k,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -286,7 +286,7 @@ void device_gemm_xdlops_mk_nk_nm(const Tensor<ABType>& a_m_k,
ABType,
AccType,
CType,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_k0_m_k1_grid_desc),
decltype(b_k0_n_k1_grid_desc),
decltype(c_m_n_grid_desc),
......
......@@ -10,7 +10,7 @@ template <ck::index_t BlockSize,
typename FloatAB,
typename FloatAcc,
typename FloatC,
ck::InMemoryDataOperationEnum_t CGlobalMemoryDataOperation,
ck::InMemoryDataOperationEnum CGlobalMemoryDataOperation,
typename AGridDesc_GK0_GM0_GM1_GK1,
typename BGridDesc_GK0_GN0_GN1_GK1,
typename CGridDesc_GM0_GM1_GN0_GN1,
......
......@@ -27,7 +27,7 @@ template <ck::index_t BlockSize,
ck::index_t ABlockTransferDstScalarPerVector_E2,
ck::index_t BThreadTransferSrcScalarPerVector_E2,
ck::index_t CThreadTransferDstScalarPerVector_K,
ck::ActivTypeEnum_t activ_type>
ck::ActivTypeEnum activ_type>
struct DriverDynamicConvolutionForwardImplicitGemmDlops_v5r1_nc0hwc1_kc0yxc1_nk0hwk1_add
{
template <typename... Wei,
......@@ -294,7 +294,7 @@ struct DriverDynamicConvolutionForwardImplicitGemmDlops_v5r1_nc0hwc1_kc0yxc1_nk0
FloatAB,
FloatAcc,
FloatC,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_e0_e1_k_e2_grid_desc),
decltype(b_e0_e1_n_ho_wo_e2_grid_desc),
decltype(c_k_n_hop_wop_grid_desc),
......
......@@ -27,7 +27,7 @@ template <ck::index_t BlockSize,
ck::index_t ABlockTransferDstScalarPerVector_E2,
ck::index_t BThreadTransferSrcScalarPerVector_E2,
ck::index_t CThreadTransferDstScalarPerVector_K,
ck::ActivTypeEnum_t activ_type>
ck::ActivTypeEnum activ_type>
struct DriverDynamicConvolutionForwardImplicitGemmDlops_v5r1_nc0hwc1_kc0yxc1_nk0hwk1_outpad
{
template <typename... Wei,
......@@ -260,7 +260,7 @@ struct DriverDynamicConvolutionForwardImplicitGemmDlops_v5r1_nc0hwc1_kc0yxc1_nk0
FloatAB,
FloatAcc,
FloatC,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_e0_e1_k_e2_grid_desc),
decltype(b_e0_e1_n_ho_wo_e2_grid_desc),
decltype(c_k_n_hop_wop_grid_desc),
......
......@@ -27,7 +27,7 @@ template <ck::index_t BlockSize,
ck::index_t ABlockTransferDstScalarPerVector_E2,
ck::index_t BThreadTransferSrcScalarPerVector_E2,
ck::index_t CThreadTransferDstScalarPerVector_K,
ck::ActivTypeEnum_t activ_type>
ck::ActivTypeEnum activ_type>
struct DriverDynamicConvolutionForwardImplicitGemmDlops_v5r1_nc0hwc1_kc0yxc1_nk0hwk1_maxpool
{
template <typename... Wei,
......@@ -305,7 +305,7 @@ struct DriverDynamicConvolutionForwardImplicitGemmDlops_v5r1_nc0hwc1_kc0yxc1_nk0
FloatAB,
FloatAcc,
FloatC,
InMemoryDataOperationEnum_t::Set,
InMemoryDataOperationEnum::Set,
decltype(a_e0_e1_k_e2_grid_desc),
decltype(b_e0_e1_n_ho_wo_e2_grid_desc),
decltype(c_k_n_hop_wop_grid_desc),
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment