vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

server : reuse cached context chunks (#9866)

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2024-10-13 18:52:48 +03:00 • committed by GitHub
parent 92be9f1216
commit c7181bd294
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 78 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

7
common/arg.cpp
View file

@ -1788,6 +1788,13 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
params.n_threads_http = value;
}
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_THREADS_HTTP"));
add_opt(common_arg(
{"--cache-reuse"}, "N",
string_format("min chunk size to attempt reusing from the cache via KV shifting (default: %d)", params.n_cache_reuse),
[](common_params & params, int value) {
params.n_cache_reuse = value;
}
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CACHE_REUSE"));
add_opt(common_arg(
{"--metrics"},
string_format("enable prometheus compatible metrics endpoint (default: %s)", params.endpoint_metrics ? "enabled" : "disabled"),

3
common/common.h
View file

@ -277,7 +277,8 @@ struct common_params {
int32_t port = 8080; // server listens on this network port
int32_t timeout_read = 600; // http read timeout in seconds
int32_t timeout_write = timeout_read; // http write timeout in seconds
int n_threads_http = -1; // number of threads to process HTTP requests (TODO: support threadpool)
int32_t n_threads_http = -1; // number of threads to process HTTP requests (TODO: support threadpool)
int32_t n_cache_reuse = 0; // min chunk size to reuse from the cache via KV shifting
std::string hostname = "127.0.0.1";
std::string public_path = ""; // NOLINT