From 7bab17262c51c3072b7622a85685fa65ec24d434 Mon Sep 17 00:00:00 2001 From: JohannesGaessler Date: Sat, 3 Feb 2024 18:05:58 +0100 Subject: [PATCH] make: fix nvcc optimization flags for host code --- Makefile | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/Makefile b/Makefile index bf9e085de..00bfaebbe 100644 --- a/Makefile +++ b/Makefile @@ -109,6 +109,7 @@ MK_NVCCFLAGS += -O3 else MK_CFLAGS += -O3 MK_CXXFLAGS += -O3 +MK_NVCCFLAGS += -O3 endif # clock_gettime came in POSIX.1b (1993) @@ -365,7 +366,7 @@ ifdef LLAMA_CUBLAS MK_CPPFLAGS += -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I$(CUDA_PATH)/targets/x86_64-linux/include -I/usr/local/cuda/targets/aarch64-linux/include MK_LDFLAGS += -lcuda -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L$(CUDA_PATH)/targets/x86_64-linux/lib -L/usr/local/cuda/targets/aarch64-linux/lib -L/usr/lib/wsl/lib OBJS += ggml-cuda.o - MK_NVCCFLAGS = -use_fast_math + MK_NVCCFLAGS += -use_fast_math ifndef JETSON_EOL_MODULE_DETECT MK_NVCCFLAGS += --forward-unknown-to-host-compiler endif # JETSON_EOL_MODULE_DETECT