ggml : add NUMA support (#1556)

* detect NUMA systems and pin work threads to nodes (linux)

* disable mmap prefetch/readahead for NUMA systems

* avoid sending finalize op to thread pool if it does nothing

* silence robot

* fix args

* make --numa a param

* recommendation that n_nodes evenly divide n_threads did not warrant such aggressive enforcement

* lower synchronization overhead

* statically allocate

* move numa state to g_state

* add description for --numa

* ggml : minor style changes

* ggml : minor style + try fix sanitizer build

* llama : allow to initialize backend with NUMA support

* llama : avoid ggml include in llama-util.h

* ggml : style / formatting

* ggml : fix handling of ops with n_threads > n_tasks > 1

* server : utilize numa parameter

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

1 files changed, 7 insertions, 3 deletions

diff --git a/llama.cpp b/llama.cpp
index c41c2a8..1a15844 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -774,7 +774,7 @@ struct llama_model_loader {
         }
 
         if (use_mmap) {
-            mapping.reset(new llama_mmap(&file_loaders.at(0)->file, prefetch_size));
+            mapping.reset(new llama_mmap(&file_loaders.at(0)->file, prefetch_size, ggml_is_numa()));
             if (lmlock) {
                 lmlock->init(mapping->addr);
             }
@@ -977,7 +977,7 @@ bool llama_mlock_supported() {
     return llama_mlock::SUPPORTED;
 }
 
-void llama_init_backend() {
+void llama_init_backend(bool numa) {
     ggml_time_init();
 
     // needed to initialize f16 tables
@@ -986,6 +986,10 @@ void llama_init_backend() {
         struct ggml_context * ctx = ggml_init(params);
         ggml_free(ctx);
     }
+
+    if (numa) {
+        ggml_numa_init();
+    }
 }
 
 int64_t llama_time_us() {
@@ -2899,7 +2903,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
 
         // maybe this should in llama_model_loader
         if (model_loader->use_mmap) {
-            model_loader->mapping.reset(new llama_mmap(&model_loader->file_loaders.at(0)->file, /* prefetch */ 0));
+            model_loader->mapping.reset(new llama_mmap(&model_loader->file_loaders.at(0)->file, /* prefetch */ 0, ggml_is_numa()));
         }
     }