diff options
Diffstat (limited to 'CMakeLists.txt')
-rw-r--r-- | CMakeLists.txt | 14 |
1 files changed, 9 insertions, 5 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index cc7560a..ffda74a 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -75,6 +75,7 @@ set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for option(LLAMA_CLBLAST "llama: use CLBlast" OFF) option(LLAMA_METAL "llama: use Metal" OFF) option(LLAMA_K_QUANTS "llama: use k-quants" ON) +option(LLAMA_QKK_64 "llama: use super-block size of 64 for k-quants" OFF) option(LLAMA_BUILD_TESTS "llama: build tests" ${LLAMA_STANDALONE}) option(LLAMA_BUILD_EXAMPLES "llama: build examples" ${LLAMA_STANDALONE}) @@ -225,6 +226,14 @@ if (LLAMA_BLAS) endif() endif() +if (LLAMA_K_QUANTS) + set(GGML_SOURCES_EXTRA ${GGML_SOURCES_EXTRA} k_quants.c k_quants.h) + add_compile_definitions(GGML_USE_K_QUANTS) + if (LLAMA_QKK_64) + add_compile_definitions(GGML_QKK_64) + endif() +endif() + if (LLAMA_CUBLAS) cmake_minimum_required(VERSION 3.17) @@ -289,11 +298,6 @@ if (LLAMA_METAL) ) endif() -if (LLAMA_K_QUANTS) - set(GGML_SOURCES_EXTRA ${GGML_SOURCES_EXTRA} k_quants.c k_quants.h) - add_compile_definitions(GGML_USE_K_QUANTS) -endif() - if (LLAMA_CLBLAST) find_package(CLBlast) if (CLBlast_FOUND) |