ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

kompute : llama-bench support and ggml_cpu_has_kompute() (#5226)

Browse source
This commit is contained in:
Jared Van Bortel 2024-01-30 19:04:37 -05:00 • committed by GitHub
parent e0085fdf7c
commit e8dc55d006
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 23 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

5
llama.cpp
View file

@ -6878,11 +6878,6 @@ static int llama_decode_internal(
n_threads = std::min(4, n_threads);
}
const bool fully_offloaded = model.n_gpu_layers >= (int) hparams.n_layer + 1;
if ((ggml_cpu_has_cublas() || ggml_cpu_has_vulkan()) && fully_offloaded) {
n_threads = 1;
}
#ifdef GGML_USE_MPI
const int64_t n_layer = hparams.n_layer;
ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);