Merge remote-tracking branch 'origin/develop' into aosewski/ggemm_multi_d2

include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_multi_abd_xdl_fixed_nk.hpp

@@ -45,8 +45,7 @@ __global__ void
                                          const BElementwiseOperation b_element_op,
                                          const CDEElementwiseOperation cde_element_op)
 {
-#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__) || \
-    defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__))
+#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx9__))
     __shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
 
     const index_t KBatch = 1;

include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_multiple_d_dl.hpp

@@ -553,7 +553,7 @@ struct DeviceGroupedGemmMultipleD_Dl : public DeviceGroupedGemm<ALayout,
 
             for(std::size_t i = 0; i < arg.gemm_desc_kernel_arg_.size(); i++)
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     std::cout << "group: " << i << " arg.a_grid_desc_k0_m_k1_{"
                               << arg.gemm_desc_kernel_arg_[i].a_grid_desc_k0_m_k1_.GetLength(I0)

include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_multiple_d_xdl_cshuffle_tile_loop.hpp

@@ -620,7 +620,7 @@ struct DeviceGroupedGemmMultipleDXdlCShuffleTileLoop
                  GridwiseGemm::template CheckTensorTransfersValidity<ALayout, BLayout, ELayout>(
                      M, N, K)))
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     std::cout << "The provided GEMM problem size (M,N,K) [" << M << "," << N << ","
                               << K << "] are not supported by current template parameters!"

include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_xdl.hpp

@@ -514,7 +514,7 @@ struct DeviceGroupedGemm_Xdl : public DeviceGroupedGemm<ALayout,
 
             for(std::size_t i = 0; i < arg.gemm_desc_kernel_arg_.size(); i++)
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     std::cout << "group: " << i << " arg.a_grid_desc_ak0_m_ak1_{"
                               << arg.gemm_desc_kernel_arg_[i].a_grid_desc_ak0_m_ak1_.GetLength(I0)

include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_xdl_splitk_cshuffle.hpp

@@ -529,7 +529,7 @@ struct DeviceGroupedGemmXdlSplitKCShuffle : public DeviceGroupedGemmSplitK<ALayo
         if((ck::type_convert<ck::index_t>(arg.gemm_kernel_args_.size()) +
             arg.skipped_group_count_) != arg.group_count_)
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "The group count is not equal to sum of skipped groups "
                              "and kernel args size!"
@@ -545,7 +545,7 @@ struct DeviceGroupedGemmXdlSplitKCShuffle : public DeviceGroupedGemmSplitK<ALayo
             bool group_arg_valid = GridwiseGemm::CheckValidity(a);
             if(not group_arg_valid)
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     std::cout << "[" << __func__ << "] group id: " << i
                               << " has invalid GridwiseGemm settings!" << std::endl;

include/ck/tensor_operation/gpu/grid/gridwise_fpAintB_gemm_wmma.hpp

@@ -50,8 +50,7 @@ __global__ void
                                  const CElementwiseOperation c_element_op,
                                  const Block2CTileMap block_2_ctile_map)
 {
-#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1100__) || defined(__gfx1101__) || \
-    defined(__gfx1102__))
+#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx11__))
     __shared__ char p_shared[GridwiseGemm::SharedMemTrait::lds_size];
 
     GridwiseGemm::template Run<HasMainKBlockLoop>(p_a_grid,
@@ -80,7 +79,7 @@ __global__ void
     ignore = b_element_op;
     ignore = c_element_op;
     ignore = block_2_ctile_map;
-#endif // end of if (defined(__gfx1100__))
+#endif // end of if (defined(__gfx11__))
 }
 
 // Assume B is Col-Major