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

llama : init metal backend as CPU backend for now

Browse source
This commit is contained in:
Georgi Gerganov 2023-07-18 17:52:13 +03:00
parent 0a3861c47b
commit 90503f150d
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
2 changed files with 48 additions and 27 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-backend.h
View file

@ -78,7 +78,7 @@ extern "C" {
static inline void ggml_backend_graph_compute(struct ggml_backend * backend, struct ggml_cgraph * cgraph) { backend->interface->graph_compute(backend->context, cgraph); }
// buffer and tensor allocation
GGML_API struct ggml_buffer ggml_backend_alloc_buffer(struct ggml_backend * backend, size_t size, size_t max_tensors);
GGML_API struct ggml_buffer ggml_backend_alloc_buffer(struct ggml_backend * backend, size_t size, size_t max_tensors); // GG: probably return ptr
GGML_API void ggml_backend_free_buffer(struct ggml_buffer * buffer);
static inline void ggml_backend_reset_buffer(struct ggml_buffer * buffer) { buffer->backend->interface->reset_buffer(buffer->backend->context, buffer->backend_buffer); }
static inline void ggml_backend_alloc_tensor(struct ggml_buffer * buffer, struct ggml_tensor * tensor) { buffer->backend->interface->alloc_tensor(buffer->backend->context, buffer->backend_buffer, tensor); }