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

library/include/ck/library/obselete_driver_offline/device_convolution_backward_weight_implicit_gemm_v4r4r5_xdlops_atomic_nhwc_kyxc_nhwk.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_convolution_forward_implicit_gemm_v4r4_dlops_nchw_kcyx_nkhw.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_convolution_forward_implicit_gemm_v4r4r2_dlops_nhwc_kyxc_nhwk.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_convolution_forward_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_convolution_forward_implicit_gemm_v4r4r4_xdlops_nhwc_kyxc_nhwk.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_convolution_forward_implicit_gemm_v5r1_dlops_nc0hwc1_kc0yxc1_nk0hwk1.hpp

@@ -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,

library/include/ck/library/obselete_driver_offline/device_convolution_forward_implicit_gemm_v6r1_dlops_nchw_kcyx_nkhw.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_convolution_maxpool_forward_implicit_gemm_v5r1_dlops_nc0hwc1_kc0yxc1_nk0hwk1.hpp

@@ -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,

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_km_kn_mn.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_km_kn_nm.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_km_nk_mn.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_km_nk_nm.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_mk_kn_mn.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_mk_kn_nm.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_mk_nk_mn.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/device_gemm_xdlops_mk_nk_nm.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/driver_contraction_dlops_v1r2.hpp

@@ -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,

library/include/ck/library/obselete_driver_offline/driver_convolution_add_forward_implicit_gemm_v5r1_dlops_nc0hwc1_kc0yxc1_nk0hwk1.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/driver_convolution_forward_implicit_gemm_v5r1_dlops_nc0hwc1_kc0yxc1_nk0hwk1.hpp

@@ -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),

library/include/ck/library/obselete_driver_offline/driver_convolution_maxpool_forward_implicit_gemm_v5r1_dlops_nc0hwc1_kc0yxc1_nk0hwk1.hpp

@@ -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),