mirrors/tinygrad
2
0
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-06-24 02:14:17 +00:00
Code Issues Projects Releases Packages Wiki Activity

fixes error when trying to convert float4 -> half4 (#1300)

Browse source
This commit is contained in:
Diogo 2023-07-20 17:20:05 -04:00 • committed by GitHub
commit 8562b5a04f
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
tinygrad/runtime/ops_cuda.py
View file

@ -88,6 +88,7 @@ class CUDACodegen(CStyleCodegen):
#include <cuda_fp16.h>
struct __align__(8) half4 {
half2 x, y;
__device__ __forceinline__ explicit half4(const float4& a): x(make_half2(__float2half(a.x), __float2half(a.y))), y(make_half2(__float2half(a.z),__float2half(a.w))) {}
__device__ __forceinline__ explicit operator float4() const {return make_float4(__half2float(x.x), __half2float(x.y), __half2float(y.x), __half2float(y.y)); }
};
""")