CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-03 09:22:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

CPU/CUDA: fix (GQA) mul mat back, add CUDA support (#11380)

Browse Source
This commit is contained in:
Johannes Gäßler
2025-01-24 12:38:31 +01:00
committed by GitHub
parent 1af6945eb0
commit 8137b4bb2b
7 changed files with 156 additions and 61 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -3002,7 +3002,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
return src0_type != GGML_TYPE_I32 && src0_type != GGML_TYPE_I16;
} break;
case GGML_OP_REPEAT_BACK:
return op->type == GGML_TYPE_F32 && op->src[0]->ne[3] == 1;
return op->type == GGML_TYPE_F32 && (op->src[0]->ne[2]*op->src[0]->ne[3]) <= (1 << 15);
case GGML_OP_CONCAT:
{
ggml_type src0_type = op->src[0]->type;