From 84525e7962bee0abef91108948bbf7f7bfdcf421 Mon Sep 17 00:00:00 2001 From: dylan Date: Fri, 7 Jul 2023 11:25:25 -0700 Subject: docker : add support for CUDA in docker (#1461) Co-authored-by: canardleteer Co-authored-by: Georgi Gerganov --- Makefile | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'Makefile') diff --git a/Makefile b/Makefile index 7141566..6068cbe 100644 --- a/Makefile +++ b/Makefile @@ -163,7 +163,12 @@ ifdef LLAMA_CUBLAS LDFLAGS += -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L$(CUDA_PATH)/targets/x86_64-linux/lib OBJS += ggml-cuda.o NVCC = nvcc - NVCCFLAGS = --forward-unknown-to-host-compiler -arch=native + NVCCFLAGS = --forward-unknown-to-host-compiler +ifdef CUDA_DOCKER_ARCH + NVCCFLAGS += -Wno-deprecated-gpu-targets -arch=$(CUDA_DOCKER_ARCH) +else + NVCCFLAGS += -arch=native +endif # CUDA_DOCKER_ARCH ifdef LLAMA_CUDA_FORCE_DMMV NVCCFLAGS += -DGGML_CUDA_FORCE_DMMV endif # LLAMA_CUDA_FORCE_DMMV @@ -187,6 +192,7 @@ ifdef LLAMA_CUDA_KQUANTS_ITER else NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2 endif + ggml-cuda.o: ggml-cuda.cu ggml-cuda.h $(NVCC) $(NVCCFLAGS) $(CXXFLAGS) -Wno-pedantic -c $< -o $@ endif # LLAMA_CUBLAS -- cgit v1.2.3