Merge branch 'develop' into lwpck-976

1b5af83d · illsilin · aac26d32 · 1fd27d52 · 1b5af83d · 1b5af83d
Commit 1b5af83d authored Oct 20, 2023 by illsilin
20 changed files
--- a/include/ck/utility/inner_product.hpp
+++ b/include/ck/utility/inner_product.hpp
@@ -192,6 +192,8 @@ inner_product<int8x4_t, int8x4_t, int32_t>(const int8x4_t& a, const int8x4_t& b,
 #else
    c = __builtin_amdgcn_sdot4(bit_cast<int32_t>(a), bit_cast<int32_t>(b), c, false);
 #endif
+#elif defined(CK_USE_AMD_V_DOT4_I32_I8_GFX11)
+    c = __builtin_amdgcn_sudot4(true, bit_cast<int32_t>(a), true, bit_cast<int32_t>(b), c, false);
 #else
    const vector_type<int8_t, 4> a_vector{a};
    const vector_type<int8_t, 4> b_vector{b};

--- a/include/ck/utility/math.hpp
+++ b/include/ck/utility/math.hpp
--- a/include/ck/utility/math_v2.hpp
+++ b/include/ck/utility/math_v2.hpp
--- a/include/ck/utility/statically_indexed_array_multi_index.hpp
+++ b/include/ck/utility/statically_indexed_array_multi_index.hpp
--- a/include/ck/utility/type_convert.hpp
+++ b/include/ck/utility/type_convert.hpp
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_groupnorm.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_groupnorm.hpp
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_layernorm.hpp
--- a/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
+++ b/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_data.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm_splitk.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm_splitk.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_dl_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_dl_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_wmma_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_conv_bwd_weight/device_grouped_conv_bwd_weight_wmma_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_backward_weight.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/grouped_convolution_backward_weight.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/normalization.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/normalization.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/normalization_swish.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/normalization_swish.hpp
--- a/library/include/ck/library/utility/check_err.hpp
+++ b/library/include/ck/library/utility/check_err.hpp
--- a/library/include/ck/library/utility/host_common_util.hpp
+++ b/library/include/ck/library/utility/host_common_util.hpp
--- a/library/include/ck/library/utility/host_tensor_generator.hpp
+++ b/library/include/ck/library/utility/host_tensor_generator.hpp