fixed header guard issues

include/ck/tensor_operation/gpu/device/device_base.hpp

@@ -16,6 +16,7 @@
 
 #include "ck/stream_config.hpp"
 #endif
+#endif
 
 namespace ck {
 namespace tensor_operation {

include/ck/tensor_operation/gpu/grid/block_to_ctile_map.hpp

@@ -17,6 +17,7 @@
 #include <limits>
 #include <stdlib.h>
 #endif
+#endif
 
 namespace ck {
 

include/ck/utility/data_type.hpp

@@ -2624,6 +2624,7 @@ using pk_i4x2_t = typename vector_type<pk_i4_t, 2>::type;
 using pk_i4x4_t = typename vector_type<pk_i4_t, 4>::type;
 using pk_i4x8_t = typename vector_type<pk_i4_t, 8>::type;
 
+#ifdef __HIPCC_RTC__
 #ifdef CK_CODE_GEN_RTC
 template <typename T>
 struct NumericLimits;
@@ -2971,7 +2972,6 @@ struct NumericLimits<bf8_ocp_t>
         return bit_cast<bf8_ocp_t>(binary_qnan);
     }
 };
-#endif
 
 template <>
 struct NumericLimits<f4_t>
@@ -3085,6 +3085,7 @@ struct NumericLimits<e8m0_bexp_t>
     }
 };
 #endif
+#endif
 
 template <typename T>
 struct NumericUtils

include/ck/utility/math_v2.hpp

@@ -465,6 +465,7 @@ inline __host__ double expm1<double>(double x)
     return std::expm1(x);
 }
 #endif
+#endif
 // math functions for the HIP kernel,  some are implemented by calling hip builtin functions
 
 static inline __device__ float abs(float x) { return ::abs(x); };

include/ck/utility/tuple_helper.hpp

@@ -173,6 +173,7 @@ using is_tuple = decltype(ck::declval<T&>().IsTuple());
 template <typename... Ts>
 __host__ __device__ constexpr auto IsNestedTuple(const Tuple<Ts...>&)
 {
+#ifndef __HIPCC_RTC__
 #ifndef CK_CODE_GEN_RTC
     return (is_detected<is_tuple, Ts>::value || ...);
 #endif