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

ggml-backend : only offload from host buffers (#11120)

Browse source
This commit is contained in:
Diego Devesa 2025-01-07 12:38:05 +01:00 • committed by GitHub
parent a4dd490069
commit a3d50bc022
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml-backend.cpp
View file

@ -761,7 +761,7 @@ static int ggml_backend_sched_backend_id_from_cur(ggml_backend_sched_t sched, st
}
// skip ROPE since the rope freqs tensor is too small to choose a backend based on it
// not an ideal solution
if (tensor->op != GGML_OP_ROPE && src->buffer != NULL && src->buffer->usage == GGML_BACKEND_BUFFER_USAGE_WEIGHTS) {
if (tensor->op != GGML_OP_ROPE && src->buffer != NULL && src->buffer->usage == GGML_BACKEND_BUFFER_USAGE_WEIGHTS && ggml_backend_buffer_is_host(src->buffer)) {
int src_backend_id = ggml_backend_sched_backend_from_buffer(sched, src, tensor);
// check if a backend with higher prio wants to offload the op
if (src_backend_id == sched->n_backends - 1) {