ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

cuda/vulkan: specify fp32-only support for some operations in supports_op (ggml/1129)

Browse source 
ggml-ci
This commit is contained in:
cmdr2 2025-02-28 12:36:46 +02:00 • committed by Georgi Gerganov
parent 8371d44595
commit 0cbee131ad
3 changed files with 8 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml-cuda/ggml-cuda.cu
View file

@ -3155,7 +3155,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
return false;
} break;
case GGML_OP_SILU_BACK:
return ggml_is_contiguous(op->src[0]);
return ggml_is_contiguous(op->src[0]) && op->src[0]->type == GGML_TYPE_F32;
break;
case GGML_OP_NORM:
case GGML_OP_RMS_NORM: