Merge branch 'amd-develop' into amd-master

32806d5f · Jun Liu · e70a4d19 · d0f355a3 · 32806d5f · 32806d5f
Commit 32806d5f authored Dec 27, 2023 by Jun Liu
20 changed files
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward_scaleadd_scaleadd_relu.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward_scaleadd_scaleadd_relu.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/groupnorm_bwd_data.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/groupnorm_bwd_data.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/layernorm_bwd_data.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/layernorm_bwd_data.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/normalization_fwd.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/normalization_fwd.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/normalization_fwd_swish.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/normalization_fwd_swish.hpp
@@ -19,7 +19,7 @@ namespace instance {
 // FP16
 void add_device_normalization_fwd_rank_5_3_swish_f16_instances(
-    std::vector<std::unique_ptr<DeviceNormalizationFwd<F16, F16, F16, F16, F32, Swish, 5, 3>>>&);
+    std::vector<std::unique_ptr<DeviceNormalizationFwd<F16, F16, F16, F16, F16, Swish, 5, 3>>>&);
 // FP32
 void add_device_normalization_fwd_rank_5_3_swish_f32_instances(
@@ -61,7 +61,7 @@ struct DeviceOperationInstanceFactory<
        if constexpr(is_same_v<XDataType, F16> && is_same_v<GammaDataType, F16> &&
                     is_same_v<BetaDataType, F16> && is_same_v<YDataType, F16> &&
-                     is_same_v<SaveMeanInvStdDataType, F32>)
+                     is_same_v<SaveMeanInvStdDataType, F16>)
        {
            if constexpr(Rank == 5 && NumReduceDim == 3)
            {

--- a/library/include/ck/library/tensor_operation_instance/gpu/permute_scale.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/permute_scale.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f8_f16_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f8_f16_mk_kn_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f8_f16_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f8_f16_mk_nk_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_default_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_default_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_padded_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_padded_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_lds_direct_load_f16_f16_f16_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_lds_direct_load_f16_f16_f16_mk_nk_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_bf16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_f16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_f32_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_int8_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_scaleadd_scaleadd_relu/xdl/device_grouped_conv3d_fwd_xdl_scaleadd_scaleadd_relu_ndhwgc_gkzyxc_ndhwgk_int8_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/normalization_bwd_data/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/normalization_bwd_data/CMakeLists.txt
--- a/library/src/tensor_operation_instance/gpu/normalization_bwd_data/device_groupnorm_bwd_data_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/normalization_bwd_data/device_groupnorm_bwd_data_f32_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/normalization_bwd_data/device_layernorm2d_bwd_data_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/normalization_bwd_data/device_layernorm2d_bwd_data_f16_instance.cpp