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

sched : offload_op also requires supports_op (#7977)

Browse Source
This commit is contained in:
slaren
2024-06-17 16:51:42 +02:00
committed by GitHub
parent c637fcd34d
commit 99052cd227
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-backend.c
View File

@@ -1172,7 +1172,7 @@ static int ggml_backend_sched_backend_id_from_cur(ggml_backend_sched_t sched, st
// check if a backend with higher prio wants to offload the op // check if a backend with higher prio wants to offload the op
if (src_backend_id == sched->n_backends - 1) { if (src_backend_id == sched->n_backends - 1) {
for (int b = 0; b < src_backend_id; b++) { for (int b = 0; b < src_backend_id; b++) {
if (ggml_backend_offload_op(sched->backends[b], tensor)) { if (ggml_backend_supports_op(sched->backends[b], tensor) && ggml_backend_offload_op(sched->backends[b], tensor)) {
SET_CAUSE(tensor, "1.off"); SET_CAUSE(tensor, "1.off");
return b; return b;
} }