AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Enable template max and min for HIP-Clang

Browse Source 
This change is required by AMDMIGraphX.

It was for HCC only. HIP-Clang also needs it for __fp16 since AMDMIGraphX uses it.

Change-Id: Id49322b7b89ef799accdf6b47627a6fce51d1ab5
This commit is contained in:
Yaxun (Sam) Liu
2020-04-19 12:56:50 -04:00
parent 8d6347c6b8
commit 4143d81618
2 changed files with 42 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/hip/hcc_detail/math_functions.h
+7 -6
View File
@@ -1397,12 +1397,18 @@ float func(float x, int y) \
}
__DEF_FLOAT_FUN2I(scalbn)
#if __HCC__
template<class T>
__DEVICE__ inline static T min(T arg1, T arg2) {
return (arg1 < arg2) ? arg1 : arg2;
}
template<class T>
__DEVICE__ inline static T max(T arg1, T arg2) {
return (arg1 > arg2) ? arg1 : arg2;
}
#if __HCC__
__DEVICE__ inline static uint32_t min(uint32_t arg1, int32_t arg2) {
return min(arg1, (uint32_t) arg2);
}
@@ -1424,11 +1430,6 @@ __DEVICE__ inline static unsigned long long min(long long arg1, unsigned long lo
return min((unsigned long long) arg1, arg2);
}*/
template<class T>
__DEVICE__ inline static T max(T arg1, T arg2) {
return (arg1 > arg2) ? arg1 : arg2;
}
__DEVICE__ inline static uint32_t max(uint32_t arg1, int32_t arg2) {
return max(arg1, (uint32_t) arg2);
}