diff options
| author | Jan Vesely <jan.vesely@rutgers.edu> | 2018-05-17 22:55:30 +0000 |
|---|---|---|
| committer | Jan Vesely <jan.vesely@rutgers.edu> | 2018-05-17 22:55:30 +0000 |
| commit | 70a270da5f4a27b670cba1cb8148fcce5c094f4f (patch) | |
| tree | ba5043901fab059f1a5ccfb2578156c62ec4c027 /libclc/amdgpu/lib/math | |
| parent | c762666e8747841ab7dd785415e1354a1682e954 (diff) | |
| download | bcm5719-llvm-70a270da5f4a27b670cba1cb8148fcce5c094f4f.tar.gz bcm5719-llvm-70a270da5f4a27b670cba1cb8148fcce5c094f4f.zip | |
Add initial support for half precision builtins
v2: fix fmax implementation
use consistent checks for __CLC_FP_SIZE
add missing TODOs
fix whitespace in definitions.h
v3: undef ZERO in modf.inc
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
reviewer: Jeroen Ketema <j.ketema@xs4all.nl>
Reviewed-by: Aaron Watry <awatry@gmail.com>
Tested-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 332677
Diffstat (limited to 'libclc/amdgpu/lib/math')
| -rw-r--r-- | libclc/amdgpu/lib/math/nextafter.cl | 5 | ||||
| -rw-r--r-- | libclc/amdgpu/lib/math/sqrt.cl | 7 |
2 files changed, 12 insertions, 0 deletions
diff --git a/libclc/amdgpu/lib/math/nextafter.cl b/libclc/amdgpu/lib/math/nextafter.cl index 5b4521d21c2..b290da0e417 100644 --- a/libclc/amdgpu/lib/math/nextafter.cl +++ b/libclc/amdgpu/lib/math/nextafter.cl @@ -8,3 +8,8 @@ _CLC_DEFINE_BINARY_BUILTIN(float, nextafter, __clc_nextafter, float, float) #pragma OPENCL EXTENSION cl_khr_fp64 : enable _CLC_DEFINE_BINARY_BUILTIN(double, nextafter, __clc_nextafter, double, double) #endif + +#ifdef cl_khr_fp16 +#pragma OPENCL EXTENSION cl_khr_fp16 : enable +_CLC_DEFINE_BINARY_BUILTIN(half, nextafter, __clc_nextafter, half, half) +#endif diff --git a/libclc/amdgpu/lib/math/sqrt.cl b/libclc/amdgpu/lib/math/sqrt.cl index 395a3f9fb21..55626003316 100644 --- a/libclc/amdgpu/lib/math/sqrt.cl +++ b/libclc/amdgpu/lib/math/sqrt.cl @@ -26,6 +26,13 @@ _CLC_DEFINE_UNARY_BUILTIN(float, sqrt, __clc_sqrt, float) +#ifdef cl_khr_fp16 + +#pragma OPENCL EXTENSION cl_khr_fp16 : enable +_CLC_DEFINE_UNARY_BUILTIN(half, sqrt, __clc_sqrt, half) + +#endif + #ifdef cl_khr_fp64 #pragma OPENCL EXTENSION cl_khr_fp64 : enable |

