nicolas
/
ik_llama_opt
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix compiler warnings (#963) 

* Fix changes meaning warnings

* A couple of more warnings and formatting

---------

Co-authored-by: firecoperana <firecoperana>
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
firecoperana 2025-11-15 05:07:15 +00:00 committed by GitHub
parent bb358223cd
commit 5ec0def0ef
1 changed files with 40 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
examples/server/server.cpp
View File

@ -663,11 +663,11 @@ struct server_prompt_cache {
} }
// next, remove any cached prompts that are fully contained in the current prompt // next, remove any cached prompts that are fully contained in the current prompt
else if(len == it->tokens.size()) { else if(len == it->tokens.size()) {
LLAMA_LOG_INFO(" - removing obsolete cached prompt with length %d\n", len); LLAMA_LOG_INFO(" - removing obsolete cached prompt with length %d\n", (int)len);
it = states.erase(it); it = states.erase(it);
} }
else { else {
++it; ++it;
} }
} }
@ -842,17 +842,17 @@ struct server_slot {
std::string stopping_word; std::string stopping_word;
stop_type stop; stop_type stop;
server_prompt server_prompt; server_prompt server_cached_prompt;
void prompt_save(server_prompt_cache & prompt_cache) const { void prompt_save(server_prompt_cache & prompt_cache) const {
assert(server_prompt.data.size() == 0); assert(server_cached_prompt.data.size() == 0);
const size_t cur_size = llama_state_seq_get_size(ctx, id); const size_t cur_size = llama_state_seq_get_size(ctx, id);
LLAMA_LOG_INFO(" - saving prompt with length %d, total state size = %.3f MiB\n", LLAMA_LOG_INFO(" - saving prompt with length %d, total state size = %.3f MiB\n",
(int)server_prompt.tokens.size(), cur_size / (1024.0 * 1024.0)); (int)server_cached_prompt.tokens.size(), cur_size / (1024.0 * 1024.0));
auto* cur = prompt_cache.alloc(server_prompt, cur_size); auto* cur = prompt_cache.alloc(server_cached_prompt, cur_size);
if (cur == nullptr) { if (cur == nullptr) {
return; return;
} }
@ -861,7 +861,7 @@ struct server_slot {
} }
void prompt_load(server_prompt_cache& prompt_cache, const server_tokens& tokens) { void prompt_load(server_prompt_cache& prompt_cache, const server_tokens& tokens) {
bool res = prompt_cache.load(server_prompt, tokens, ctx, id); bool res = prompt_cache.load(server_cached_prompt, tokens, ctx, id);
if (!res) { if (!res) {
LLAMA_LOG_INFO("failed to load prompt from cache\n"); LLAMA_LOG_INFO("failed to load prompt from cache\n");
} }
@ -1828,21 +1828,21 @@ struct server_context {
update_cache = update_cache && (ret->mctx == nullptr); update_cache = update_cache && (ret->mctx == nullptr);
LLAMA_LOG_INFO("prompt cache: cache size: %d, cache_ram_n_min: %d, f_keep: %.2f, cache_ram_similarity: %.2f\n", LLAMA_LOG_INFO("prompt cache: cache size: %d, cache_ram_n_min: %d, f_keep: %.2f, cache_ram_similarity: %.2f\n",
tokens.size(), cache_ram_n_min, f_keep, cache_ram_similarity); (int)tokens.size(), cache_ram_n_min, f_keep, cache_ram_similarity);
if (update_cache) { if (update_cache) {
const int64_t t_start = ggml_time_us(); const int64_t t_start = ggml_time_us();
LLAMA_LOG_INFO("updating prompt cache\n"); LLAMA_LOG_INFO("updating prompt cache\n");
ret->server_prompt.tokens = server_tokens(tokens.get_text_tokens(), false); // copy cache tokens ret->server_cached_prompt.tokens = server_tokens(tokens.get_text_tokens(), false); // copy cache tokens
ret->prompt_save(*prompt_cache); ret->prompt_save(*prompt_cache);
LLAMA_LOG_INFO("prompt cache save took %.2f ms\n", (ggml_time_us() - t_start) / 1000.0); LLAMA_LOG_INFO("prompt cache save took %.2f ms\n", (ggml_time_us() - t_start) / 1000.0);
} }
// has prompts saved earlier to load // has prompts saved earlier to load
if (!prompt_cache->states.empty()) { if (!prompt_cache->states.empty()) {
const int64_t t_start = ggml_time_us(); const int64_t t_start = ggml_time_us();
ret->server_prompt.tokens = server_tokens(tokens.get_text_tokens(), false); // copy cache tokens ret->server_cached_prompt.tokens = server_tokens(tokens.get_text_tokens(), false); // copy cache tokens
ret->prompt_load(*prompt_cache, task.tokens); ret->prompt_load(*prompt_cache, task.tokens);
prompt_cache->update(); prompt_cache->update();
ret->cache_tokens = server_tokens(ret->server_prompt.tokens.get_text_tokens(), false); // recover cache tokens ret->cache_tokens = server_tokens(ret->server_cached_prompt.tokens.get_text_tokens(), false); // recover cache tokens
LLAMA_LOG_INFO("prompt cache load took %.2f ms\n", (ggml_time_us() - t_start) / 1000.0); LLAMA_LOG_INFO("prompt cache load took %.2f ms\n", (ggml_time_us() - t_start) / 1000.0);
} }
} }