changes to make it compile

include/ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer_v6r2.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -258,3 +261,5 @@ struct ThreadwiseTensorSliceTransfer_v6r2
 };
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer_v6r3.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -308,3 +311,5 @@ struct ThreadwiseTensorSliceTransfer_v6r3
 };
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer_v7.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -296,3 +299,5 @@ struct ThreadwiseTensorSliceTransfer_v7
 };
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/gpu/thread/threadwise_welford.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -135,3 +138,5 @@ struct ThreadwiseWelfordMerge
 };
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/gpu/warp/wmma_gemm.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -505,3 +508,5 @@ struct WmmaGemm
 };
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/gpu/warp/xdlops_gemm.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -965,3 +968,5 @@ struct XdlopsGemm
 };
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/operator_transform/transform_contraction_to_gemm.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -286,3 +289,5 @@ struct TransformBatchedContractionContractionToBatchedGemmGemm
 
 } // namespace tensor_operation
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/operator_transform/transform_conv_bwd_data_to_gemm_v1.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -1132,3 +1135,5 @@ struct TransformConvBwdDataToGemm_v1
 
 } // namespace tensor_operation
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/tensor_operation/operator_transform/transform_conv_fwd_to_gemm.hpp

 
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
+
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -878,3 +881,5 @@ struct TransformConvFwdToGemm
 
 } // namespace tensor_operation
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/utility/amd_address_space.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -43,3 +46,5 @@ __host__ __device__ T CK_CONSTANT_ADDRESS_SPACE* cast_pointer_to_constant_addres
 }
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/utility/amd_buffer_addressing.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -1312,3 +1315,5 @@ amd_buffer_atomic_max(const typename vector_type_maker<T, N>::type::type src_thr
 }
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/utility/amd_inline_asm.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -369,3 +372,5 @@ __device__ void amd_assembly_wmma_f32_16x16x16_f16_w32(half16_t a, half16_t b, f
 
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/amd_wave_read_first_lane.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -7,87 +10,75 @@
 #include "ck/utility/functional2.hpp"
 #include "ck/utility/math.hpp"
 
+#ifndef __HIPCC_RTC__
 #include <array>
 #include <cstddef>
 #include <cstdint>
 #include <type_traits>
+#endif
 
 namespace ck {
 namespace detail {
 
-template <unsigned SizeInBytes>
-struct get_carrier;
+template <unsigned SizeInBytes> struct get_carrier;
 
-template <>
-struct get_carrier<1>
-{
-    using type = uint8_t;
+template <> struct get_carrier<1> {
+  using type = uint8_t;
 };
 
-template <>
-struct get_carrier<2>
-{
-    using type = uint16_t;
+template <> struct get_carrier<2> {
+  using type = uint16_t;
 };
 
-template <>
-struct get_carrier<3>
-{
-    using type = class carrier
-    {
-        using value_type = uint32_t;
-
-        std::array<std::byte, 3> bytes;
-        static_assert(sizeof(bytes) <= sizeof(value_type));
-
-        // replacement of host std::copy_n()
-        template <typename InputIterator, typename Size, typename OutputIterator>
-        __device__ static OutputIterator copy_n(InputIterator from, Size size, OutputIterator to)
-        {
-            if(0 < size)
-            {
-                *to = *from;
-                ++to;
-                for(Size count = 1; count < size; ++count)
-                {
-                    *to = *++from;
-                    ++to;
-                }
-            }
-
-            return to;
+template <> struct get_carrier<3> {
+  using type = class carrier {
+    using value_type = uint32_t;
+
+    // std::array<std::byte, 3> bytes;
+    std::byte bytes[3];
+    static_assert(sizeof(bytes) <= sizeof(value_type));
+
+    // replacement of host std::copy_n()
+    template <typename InputIterator, typename Size, typename OutputIterator>
+    __device__ static OutputIterator copy_n(InputIterator from, Size size,
+                                            OutputIterator to) {
+      if (0 < size) {
+        *to = *from;
+        ++to;
+        for (Size count = 1; count < size; ++count) {
+          *to = *++from;
+          ++to;
         }
+      }
 
-        // method to trigger template substitution failure
-        __device__ carrier(const carrier& other) noexcept
-        {
-            copy_n(other.bytes.begin(), bytes.size(), bytes.begin());
-        }
+      return to;
+    }
 
-        public:
-        __device__ carrier& operator=(value_type value) noexcept
-        {
-            copy_n(reinterpret_cast<const std::byte*>(&value), bytes.size(), bytes.begin());
+    // method to trigger template substitution failure
+    __device__ carrier(const carrier &other) noexcept {
+      copy_n(&other.bytes[0], 3, &bytes[0]);
+    }
 
-            return *this;
-        }
+  public:
+    __device__ carrier &operator=(value_type value) noexcept {
+      copy_n(reinterpret_cast<const std::byte *>(&value), 3, &bytes[0]);
 
-        __device__ operator value_type() const noexcept
-        {
-            std::byte result[sizeof(value_type)];
+      return *this;
+    }
 
-            copy_n(bytes.begin(), bytes.size(), result);
+    __device__ operator value_type() const noexcept {
+      std::byte result[sizeof(value_type)];
 
-            return *reinterpret_cast<const value_type*>(result);
-        }
-    };
+      copy_n(&bytes[0], 3, result);
+
+      return *reinterpret_cast<const value_type *>(result);
+    }
+  };
 };
 static_assert(sizeof(get_carrier<3>::type) == 3);
 
-template <>
-struct get_carrier<4>
-{
-    using type = uint32_t;
+template <> struct get_carrier<4> {
+  using type = uint32_t;
 };
 
 template <unsigned SizeInBytes>
@@ -95,44 +86,43 @@ using get_carrier_t = typename get_carrier<SizeInBytes>::type;
 
 } // namespace detail
 
-__device__ inline int32_t amd_wave_read_first_lane(int32_t value)
-{
-    return __builtin_amdgcn_readfirstlane(value);
+__device__ inline int32_t amd_wave_read_first_lane(int32_t value) {
+  return __builtin_amdgcn_readfirstlane(value);
 }
 
-template <
-    typename Object,
-    typename = std::enable_if_t<std::is_class_v<Object> && std::is_trivially_copyable_v<Object>>>
-__device__ auto amd_wave_read_first_lane(const Object& obj)
-{
-    using Size                = unsigned;
-    constexpr Size SgprSize   = 4;
-    constexpr Size ObjectSize = sizeof(Object);
-
-    auto* const from_obj = reinterpret_cast<const std::byte*>(&obj);
-    alignas(Object) std::byte to_obj[ObjectSize];
-
-    constexpr Size RemainedSize             = ObjectSize % SgprSize;
-    constexpr Size CompleteSgprCopyBoundary = ObjectSize - RemainedSize;
-    for(Size offset = 0; offset < CompleteSgprCopyBoundary; offset += SgprSize)
-    {
-        using Sgpr = detail::get_carrier_t<SgprSize>;
-
-        *reinterpret_cast<Sgpr*>(to_obj + offset) =
-            amd_wave_read_first_lane(*reinterpret_cast<const Sgpr*>(from_obj + offset));
-    }
-
-    if constexpr(0 < RemainedSize)
-    {
-        using Carrier = detail::get_carrier_t<RemainedSize>;
-
-        *reinterpret_cast<Carrier*>(to_obj + CompleteSgprCopyBoundary) = amd_wave_read_first_lane(
-            *reinterpret_cast<const Carrier*>(from_obj + CompleteSgprCopyBoundary));
-    }
-
-    /// NOTE: Implicitly start object lifetime. It's better to use std::start_lifetime_at() in this
-    /// scenario
-    return *reinterpret_cast<Object*>(to_obj);
+template <typename Object, typename = std::enable_if_t<
+                               std::is_class<Object>::value &&
+                               std::is_trivially_copyable<Object>::value>>
+__device__ auto amd_wave_read_first_lane(const Object &obj) {
+  using Size = unsigned;
+  constexpr Size SgprSize = 4;
+  constexpr Size ObjectSize = sizeof(Object);
+
+  auto *const from_obj = reinterpret_cast<const std::byte *>(&obj);
+  alignas(Object) std::byte to_obj[ObjectSize];
+
+  constexpr Size RemainedSize = ObjectSize % SgprSize;
+  constexpr Size CompleteSgprCopyBoundary = ObjectSize - RemainedSize;
+  for (Size offset = 0; offset < CompleteSgprCopyBoundary; offset += SgprSize) {
+    using Sgpr = detail::get_carrier_t<SgprSize>;
+
+    *reinterpret_cast<Sgpr *>(to_obj + offset) = amd_wave_read_first_lane(
+        *reinterpret_cast<const Sgpr *>(from_obj + offset));
+  }
+
+  if constexpr (0 < RemainedSize) {
+    using Carrier = detail::get_carrier_t<RemainedSize>;
+
+    *reinterpret_cast<Carrier *>(to_obj + CompleteSgprCopyBoundary) =
+        amd_wave_read_first_lane(*reinterpret_cast<const Carrier *>(
+            from_obj + CompleteSgprCopyBoundary));
+  }
+
+  /// NOTE: Implicitly start object lifetime. It's better to use
+  /// std::start_lifetime_at() in this scenario
+  return *reinterpret_cast<Object *>(to_obj);
 }
 
 } // namespace ck
+
+#pragma clang diagnostic pop

include/ck/utility/amd_wmma.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -256,3 +259,5 @@ struct intrin_wmma_i32_16x16x16_iu8_w64<16, 16, neg_a, neg_b, clamp>
 
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/amd_xdlops.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -419,3 +422,5 @@ struct intrin_mfma_f32_16x16x32f8f8<16, 16>
 };
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/array.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -64,3 +67,5 @@ __host__ __device__ constexpr auto make_array()
 
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/array_multi_index.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -78,3 +81,5 @@ __host__ __device__ constexpr auto operator*(const MultiIndex<NSize>& a, const T
 
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/c_style_pointer_cast.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -23,3 +26,5 @@ __host__ __device__ PY c_style_pointer_cast(PX p_x)
 
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/common_header.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -51,3 +54,5 @@
 #ifdef CK_USE_AMD_MFMA
 #include "ck/utility/amd_xdlops.hpp"
 #endif
+
+#pragma clang diagnostic pop

include/ck/utility/container_element_picker.hpp

+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Weverything"
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
@@ -156,3 +159,5 @@ __host__ __device__ constexpr auto pick_container_element(const Arr& a, Picks)
 
 } // namespace ck
 #endif
+
+#pragma clang diagnostic pop