Merge branch 'develop' into lwpck-1038

ec2fbe1f · Rostyslav Geyyer · GitHub · bd5f5a0d · e8cddfdc · ec2fbe1f
Unverified Commit ec2fbe1f authored Nov 17, 2023 by Rostyslav Geyyer Committed by GitHub Nov 17, 2023
20 changed files
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_groupnorm_bwd.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_groupnorm_bwd.hpp
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm_bwd.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm_bwd.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm.hpp
@@ -377,7 +377,7 @@ struct DeviceOperationInstanceFactory<
            if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Row> &&
                         is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f32_f32_f32_mk_kn_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f32_f32_f32_mk_kn_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f32_f32_f32_mk_kn_mn_instances(op_ptrs);
 #endif
@@ -386,7 +386,7 @@ struct DeviceOperationInstanceFactory<
            else if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Col> &&
                              is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f32_f32_f32_mk_nk_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f32_f32_f32_mk_nk_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f32_f32_f32_mk_nk_mn_instances(op_ptrs);
 #endif
@@ -395,7 +395,7 @@ struct DeviceOperationInstanceFactory<
            else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Row> &&
                              is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f32_f32_f32_km_kn_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f32_f32_f32_km_kn_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f32_f32_f32_km_kn_mn_instances(op_ptrs);
 #endif
@@ -404,7 +404,7 @@ struct DeviceOperationInstanceFactory<
            else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Col> &&
                              is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f32_f32_f32_km_nk_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f32_f32_f32_km_nk_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f32_f32_f32_km_nk_mn_instances(op_ptrs);
 #endif
@@ -418,7 +418,7 @@ struct DeviceOperationInstanceFactory<
            if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Row> &&
                         is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f16_f16_f16_mk_kn_mn_instances(op_ptrs);
                add_device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instances(op_ptrs);
@@ -430,7 +430,7 @@ struct DeviceOperationInstanceFactory<
            else if constexpr(is_same_v<ALayout, Row> && is_same_v<BLayout, Col> &&
                              is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f16_f16_f16_mk_nk_mn_instances(op_ptrs);
                add_device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instances(op_ptrs);
@@ -443,7 +443,7 @@ struct DeviceOperationInstanceFactory<
            else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Row> &&
                              is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f16_f16_f16_km_kn_mn_instances(op_ptrs);
                add_device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instances(op_ptrs);
@@ -455,7 +455,7 @@ struct DeviceOperationInstanceFactory<
            else if constexpr(is_same_v<ALayout, Col> && is_same_v<BLayout, Col> &&
                              is_same_v<CLayout, Row>)
            {
-                add_device_gemm_xdl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
+                /// add_device_gemm_xdl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
 #ifdef DL_KERNELS
                add_device_gemm_dl_f16_f16_f16_km_nk_mn_instances(op_ptrs);
                add_device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instances(op_ptrs);

--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_scaleadd_ab_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_scaleadd_ab_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_scaleadd_scaleadd_relu_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_scaleadd_scaleadd_relu_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward_scaleadd_ab.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_forward_scaleadd_ab.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/quantization/grouped_convolution_bias_forward_perchannel_quantization.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_bias_forward_perchannel_quantization.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_bias_forward_perlayer_quantization.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_bias_forward_perlayer_quantization.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_forward_perchannel_quantization.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_forward_perchannel_quantization.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_forward_perlayer_quantization.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/quantization/grouped_convolution_forward_perlayer_quantization.hpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/xdl/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/xdl/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instance.cpp
@@ -10,18 +10,18 @@ namespace device {
 namespace instance {

 void add_device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_bf16_instances(
-    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleD<1,
-                                                              GNWC,
-                                                              GKXC,
-                                                              Empty_Tuple,
-                                                              GNWK,
-                                                              BF16,
-                                                              BF16,
-                                                              Empty_Tuple,
-                                                              BF16,
-                                                              PassThrough,
-                                                              PassThrough,
-                                                              PassThrough>>>& instances)
+    std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleABD<1,
+                                                                GNWC,
+                                                                GKXC,
+                                                                Empty_Tuple,
+                                                                GNWK,
+                                                                BF16,
+                                                                BF16,
+                                                                Empty_Tuple,
+                                                                BF16,
+                                                                PassThrough,
+                                                                PassThrough,
+                                                                PassThrough>>>& instances)
 {
    add_device_operation_instances(instances,
                                   device_grouped_conv_fwd_xdl_bf16_instances<1,

--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/xdl/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/xdl/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f16_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/xdl/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/grouped_conv1d_fwd/xdl/device_grouped_conv1d_fwd_xdl_gnwc_gkxc_gnwk_f32_instance.cpp