Merge branch 'develop' into hip_tensor_permute

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};

include/ck/utility/statically_indexed_array_multi_index.hpp

@@ -5,6 +5,7 @@
 #define CK_STATICALLY_INDEXED_ARRAY_MULTI_INDEX_HPP
 
 #include "common_header.hpp"
+#include "ck/utility/math_v2.hpp"
 
 namespace ck {
 

library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp

@@ -20,12 +20,8 @@ using F16  = ck::half_t;
 using BF16 = ck::bhalf_t;
 using I8   = int8_t;
 using I32  = int32_t;
-#if defined CK_ENABLE_FP8
 using F8   = ck::f8_t;
-#endif
-#if defined CK_ENABLE_BF8
 using BF8  = ck::bf8_t;
-#endif
 
 using Empty_Tuple = ck::Tuple<>;