Skip to content

CUDA: use 1 thread if model is fully offloaded #2915

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Sep 21, 2023
Merged

CUDA: use 1 thread if model is fully offloaded #2915

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 9 additions & 0 deletions llama.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3777,6 +3777,15 @@ static bool llama_eval_internal(
n_threads = std::min(4, n_threads);
}

// If all tensors can be run on the GPU then using more than 1 thread is detrimental.
const bool full_offload_supported = model.arch == LLM_ARCH_LLAMA ||
model.arch == LLM_ARCH_BAICHUAN ||
model.arch == LLM_ARCH_FALCON;
const bool fully_offloaded = model.n_gpu_layers >= (int) hparams.n_layer + 3;
if (ggml_cpu_has_cublas() && full_offload_supported && fully_offloaded) {
n_threads = 1;
}

struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
struct ggml_tensor * embeddings = gf->nodes[gf->n_nodes - 2];

Expand Down