happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

common : fix regression caused by extra memory allocations during sampling

This commit is contained in:
Georgi Gerganov 2025-11-19 13:43:29 +02:00
parent d74eb61aa7
commit 38f408c253
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 6 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
common/sampling.cpp
View File

@ -122,27 +122,24 @@ struct common_sampler {
const int n_vocab = llama_vocab_n_tokens(vocab); const int n_vocab = llama_vocab_n_tokens(vocab);
// Use the member variable instead of allocating locally
cur.clear();
if (sampled_probs) { if (sampled_probs) {
const uint32_t sampled_probs_count = llama_get_backend_sampled_probs_count_ith(ctx, idx); const uint32_t sampled_probs_count = llama_get_backend_sampled_probs_count_ith(ctx, idx);
cur.reserve(sampled_probs_count); cur.resize(sampled_probs_count);
for (uint32_t i = 0; i < sampled_probs_count; ++i) { for (uint32_t i = 0; i < sampled_probs_count; ++i) {
cur.emplace_back(llama_token_data{sampled_ids[i], sampled_logits[i], sampled_probs[i]}); cur[i] = llama_token_data{sampled_ids[i], sampled_logits[i], sampled_probs[i]};
} }
} else if (sampled_logits) { } else if (sampled_logits) {
const uint32_t sampled_logits_count = llama_get_backend_sampled_logits_count_ith(ctx, idx); const uint32_t sampled_logits_count = llama_get_backend_sampled_logits_count_ith(ctx, idx);
cur.reserve(sampled_logits_count); cur.resize(sampled_logits_count);
for (uint32_t i = 0; i < sampled_logits_count; i++) { for (uint32_t i = 0; i < sampled_logits_count; i++) {
cur.emplace_back(llama_token_data{sampled_ids[i], sampled_logits[i], 0.0f}); cur[i] = llama_token_data{sampled_ids[i], sampled_logits[i], 0.0f};
} }
} else { } else {
const auto * logits = llama_get_logits_ith(ctx, idx); const auto * logits = llama_get_logits_ith(ctx, idx);
GGML_ASSERT(logits != nullptr); GGML_ASSERT(logits != nullptr);
cur.reserve(n_vocab); cur.resize(n_vocab);
for (llama_token token_id = 0; token_id < n_vocab; token_id++) { for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
cur.emplace_back(llama_token_data{token_id, logits[token_id], 0.0f}); cur[token_id] = llama_token_data{token_id, logits[token_id], 0.0f};
} }
} }