ggml-cpu: avoid treating all host RAM as free#22939

fl0rianr wants to merge 1 commit into

ggml-org:masterggml-org/llama.cpp:masterfrom

fl0rianr:fix/cpu_host_memory_estimationfl0rianr/llama.cpp_contribution:fix/cpu_host_memory_estimationCopy head branch name to clipboard