Merge branch 'develop' into ck-flex

6dcc40d4 · Max Podkorytov · GitHub · 2c8e04aa · 800cf897 · 6dcc40d4
Unverified Commit 6dcc40d4 authored Feb 04, 2025 by Max Podkorytov Committed by GitHub Feb 04, 2025
20 changed files
--- a/include/ck/utility/scaled_type_convert.hpp
+++ b/include/ck/utility/scaled_type_convert.hpp
--- a/include/ck/utility/type_convert.hpp
+++ b/include/ck/utility/type_convert.hpp
--- a/include/ck_tile/core.hpp
+++ b/include/ck_tile/core.hpp
@@ -27,6 +27,7 @@
 #include "ck_tile/core/numeric/float8.hpp"
 #include "ck_tile/core/numeric/half.hpp"
 #include "ck_tile/core/numeric/int8.hpp"
+#include "ck_tile/core/numeric/pk_int4.hpp"
 #include "ck_tile/core/numeric/integer.hpp"
 #include "ck_tile/core/numeric/integral_constant.hpp"
 #include "ck_tile/core/numeric/math.hpp"

--- a/include/ck_tile/core/config.hpp
+++ b/include/ck_tile/core/config.hpp
@@ -144,6 +144,10 @@
 #define CK_TILE_USE_AMD_BUFFER_ATOMIC_ADD_INTEGER 1
 #endif
+#ifndef CK_TILE_USE_PK4_LAYOUT_SHUFFLE
+#define CK_TILE_USE_PK4_LAYOUT_SHUFFLE 1
+#endif
 // buffer atomic add: floating point
 #ifndef __HIP_DEVICE_COMPILE__ // for host code
 #define CK_TILE_USE_AMD_BUFFER_ATOMIC_ADD_FLOAT 1

--- a/include/ck_tile/core/numeric/pk_int4.hpp
+++ b/include/ck_tile/core/numeric/pk_int4.hpp
--- a/include/ck_tile/core/numeric/vector_type.hpp
+++ b/include/ck_tile/core/numeric/vector_type.hpp
--- a/include/ck_tile/ops/flatmm/block/uk/flatmm_sn_uk_gfx9_32x128x512_1x4x1_16x16x16.inc
+++ b/include/ck_tile/ops/flatmm/block/uk/flatmm_sn_uk_gfx9_32x128x512_1x4x1_16x16x16.inc
@@ -824,4 +824,4 @@
 #undef _UK_PK_CVT_
 #undef _UK_ATOMIC_ADD_
 #undef CK_TILE_FLATMM_UK_MFMA
-// clang-format on
+    // clang-format on
--- a/include/ck_tile/ops/flatmm/block/uk/flatmm_sn_uk_gfx9_32x128x512_1x4x1_16x16x16_itl.inc
+++ b/include/ck_tile/ops/flatmm/block/uk/flatmm_sn_uk_gfx9_32x128x512_1x4x1_16x16x16_itl.inc
--- a/include/ck_tile/ops/flatmm/block/uk/flatmm_uk_gfx9_32x512x128_1x1x1_16x16x16.inc
+++ b/include/ck_tile/ops/flatmm/block/uk/flatmm_uk_gfx9_32x512x128_1x1x1_16x16x16.inc
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_two_stage_xdl_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_two_stage_xdl_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_comp_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_comp_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_merged_groups_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_merged_groups_instance.hpp
--- a/library/src/tensor_operation_instance/gpu/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/CMakeLists.txt
--- a/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gkm_gkn_gmn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gkm_gkn_gmn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gkm_gnk_gmn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gkm_gnk_gmn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gmk_gkn_gmn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gmk_gkn_gmn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gmk_gnk_gmn_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm/device_batched_gemm_xdl_f16_f16_f16_gmk_gnk_gmn_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/batched_gemm_softmax_gemm/device_batched_gemm_softmax_gemm_xdl_cshuffle_f16_f16_f16_f16_gmk_gnk_gno_gmo_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm_softmax_gemm/device_batched_gemm_softmax_gemm_xdl_cshuffle_f16_f16_f16_f16_gmk_gnk_gno_gmo_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/batched_gemm_softmax_gemm_permute/device_batched_gemm_bias_softmax_gemm_permute_xdl_cshuffle_bf16_bf16_bf16_bf16_gmk_gnk_gno_gmo_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm_softmax_gemm_permute/device_batched_gemm_bias_softmax_gemm_permute_xdl_cshuffle_bf16_bf16_bf16_bf16_gmk_gnk_gno_gmo_instance.cpp
--- a/library/src/tensor_operation_instance/gpu/batched_gemm_softmax_gemm_permute/device_batched_gemm_bias_softmax_gemm_permute_xdl_cshuffle_f16_f16_f16_f16_gmk_gnk_gno_gmo_instance.cpp
+++ b/library/src/tensor_operation_instance/gpu/batched_gemm_softmax_gemm_permute/device_batched_gemm_bias_softmax_gemm_permute_xdl_cshuffle_f16_f16_f16_f16_gmk_gnk_gno_gmo_instance.cpp