tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-01 05:05:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : update llama_kv_self API

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-01-14 16:47:34 +02:00
parent fd05ab87aa
commit 17b363afd3
30 changed files with 387 additions and 205 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
examples/batched-bench/batched-bench.cpp
View File

@ -57,8 +57,6 @@ int main(int argc, char ** argv) {
return 1;
}
llama_kv_cache * kv = llama_get_kv_cache(ctx);
const int32_t n_kv_max = llama_n_ctx(ctx);
llama_batch batch = llama_batch_init(n_kv_max, 0, 1);
@ -134,7 +132,7 @@ int main(int argc, char ** argv) {
const auto t_pp_start = ggml_time_us();
llama_kv_cache_clear(kv);
llama_kv_self_clear(ctx);
if (!decode_helper(ctx, batch, ctx_params.n_batch)) {
LOG_ERR("%s: llama_decode() failed\n", __func__);
@ -143,7 +141,7 @@ int main(int argc, char ** argv) {
if (is_pp_shared) {
for (int32_t i = 1; i < pl; ++i) {
llama_kv_cache_seq_cp(kv, 0, i, -1, -1);
llama_kv_self_seq_cp(ctx, 0, i, -1, -1);
}
}