tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-27 20:05:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : remove llama_kv_cache_view API + remove deprecated (#13653)

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-05-20 16:13:16 +03:00
committed by GitHub
parent b69f1647f9
commit a4090d1174
10 changed files with 1 additions and 390 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
examples/parallel/parallel.cpp
View File

@ -178,8 +178,6 @@ int main(int argc, char ** argv) {
// insert new requests as soon as the previous one is done
const bool cont_batching = params.cont_batching;
const bool dump_kv_cache = params.dump_kv_cache;
// is the system prompt shared in the cache
const bool is_sp_shared = params.is_pp_shared;
@ -241,8 +239,6 @@ int main(int argc, char ** argv) {
int32_t n_total_gen = 0;
int32_t n_cache_miss = 0;
struct llama_kv_cache_view kvc_view = llama_kv_cache_view_init(ctx, n_clients);
const auto t_main_start = ggml_time_us();
LOG_INF("%s: Simulating parallel requests from clients:\n", __func__);
@ -272,11 +268,6 @@ int main(int argc, char ** argv) {
LOG_INF("Processing requests ...\n\n");
while (true) {
if (dump_kv_cache) {
llama_kv_cache_view_update(ctx, &kvc_view);
common_kv_cache_dump_view_seqs(kvc_view, 40);
}
common_batch_clear(batch);
// decode any currently ongoing sequences