From 2266f48d68ef671c7918d62103ec005dc287d2d5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E8=A8=B1=E5=85=83=E8=B1=AA?= Date: Wed, 11 Mar 2026 09:28:43 +0800 Subject: [PATCH] cli, server: apply --prio process priority setting The --prio flag was parsed but never applied in llama-cli and llama-server. Only llama-completion and llama-bench called set_process_priority(). Add the missing calls after backend initialization so the flag takes effect in all tools. --- tools/cli/cli.cpp | 4 ++++ tools/server/server.cpp | 4 ++++ 2 files changed, 8 insertions(+) diff --git a/tools/cli/cli.cpp b/tools/cli/cli.cpp index d43d105490..97e633fb50 100644 --- a/tools/cli/cli.cpp +++ b/tools/cli/cli.cpp @@ -339,6 +339,10 @@ int main(int argc, char ** argv) { llama_backend_init(); llama_numa_init(params.numa); + if (!set_process_priority(params.cpuparams.priority)) { + LOG_WRN("%s: failed to set process priority\n", __func__); + } + // TODO: avoid using atexit() here by making `console` a singleton console::init(params.simple_io, params.use_color); atexit([]() { console::cleanup(); }); diff --git a/tools/server/server.cpp b/tools/server/server.cpp index 0bd6fda17d..31592c117a 100644 --- a/tools/server/server.cpp +++ b/tools/server/server.cpp @@ -107,6 +107,10 @@ int main(int argc, char ** argv) { llama_backend_init(); llama_numa_init(params.numa); + if (!set_process_priority(params.cpuparams.priority)) { + LOG_WRN("%s: failed to set process priority\n", __func__); + } + LOG_INF("system info: n_threads = %d, n_threads_batch = %d, total_threads = %d\n", params.cpuparams.n_threads, params.cpuparams_batch.n_threads, std::thread::hardware_concurrency()); LOG_INF("\n"); LOG_INF("%s\n", common_params_get_system_info(params).c_str());