huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-18 19:47:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

More u32 support.

Browse Source
This commit is contained in:
laurent
2023-06-23 14:54:03 +01:00
parent 08394f7924
commit 4c8931d2e4
3 changed files with 20 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
kernels/src/binary.cu
View File

@ -1,4 +1,5 @@
#include "binary_op_macros.cuh"
#include<stdint.h>
#if __CUDA_ARCH__ >= 530
BINARY_OP(__half, badd_f16, x + y)
@ -8,10 +9,14 @@ BINARY_OP(__half, bsub_f16, x - y)
#endif
BINARY_OP(float, badd_f32, x + y)
BINARY_OP(double, badd_fwd_f64, x + y);
BINARY_OP(double, badd_f64, x + y);
BINARY_OP(uint32_t, badd_u32, x + y);
BINARY_OP(float, bdiv_f32, x / y)
BINARY_OP(double, bdiv_f64, x / y);
BINARY_OP(uint32_t, bdiv_u32, x / y);
BINARY_OP(float, bmul_f32, x * y)
BINARY_OP(double, bmul_f64, x * y);
BINARY_OP(uint32_t, bmul_u32, x * y);
BINARY_OP(float, bsub_f32, x - y)
BINARY_OP(double, bsub_f64, x - y);
BINARY_OP(uint32_t, bsub_u32, x - y);