CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-12 10:47:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: fix GET_ROWS for large tensors (#15882)

Browse Source
This commit is contained in:
Johannes Gäßler
2025-09-09 08:11:01 +02:00
committed by GitHub
parent c252ce67c4
commit 550cf726e1
2 changed files with 41 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/ggml-cuda/ggml-cuda.cu
View File

@@ -3393,10 +3393,6 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
return op->type == GGML_TYPE_F32 && op->src[0]->type == GGML_TYPE_F32 && op->src[1]->type == GGML_TYPE_F32;
case GGML_OP_GET_ROWS:
{
// FIXME: https://github.com/ggml-org/llama.cpp/pull/15868
if (op->src[1]->ne[1]*op->src[1]->ne[2] > 65535) {
return false;
}
switch (op->src[0]->type) {
case GGML_TYPE_F16:
case GGML_TYPE_F32: