sdgoij/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-05-14 04:54:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

sampling : use pinned memory for backend sampling buffers

Browse Source
This commit is contained in:
Daniel Bevenius
2025-11-21 14:02:16 +01:00
parent c1625620f6
commit 61ffe41dc1
7 changed files with 358 additions and 120 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
common/sampling.cpp
View File

@@ -113,9 +113,9 @@ struct common_sampler {
llama_token_data_array cur_p;
void set_logits(struct llama_context * ctx, int idx) {
const float * sampled_probs = llama_get_backend_sampled_probs_ith (ctx, idx);
const float * sampled_logits = llama_get_backend_sampled_logits_ith (ctx, idx);
const llama_token * sampled_ids = llama_get_backend_sampled_token_ids_ith(ctx, idx);
const float * sampled_probs = llama_get_backend_sampled_probs_ith (ctx, idx);
const float * sampled_logits = llama_get_backend_sampled_logits_ith (ctx, idx);
const llama_token * sampled_ids = llama_get_backend_sampled_candidates_ith(ctx, idx);
const llama_model * model = llama_get_model(ctx);
const llama_vocab * vocab = llama_model_get_vocab(model);