aparler/c-llm
1
0
Fork 0
mirror of https://github.com/karpathy/llm.c.git synced 2026-07-28 20:35:09 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Remove arch=native as it only available on recent CUDA versions

Browse source
This commit is contained in:
ademeure 2024-05-05 00:00:58 +01:00
parent 18d7ed9242
commit ec0ab2d2d0
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
Makefile
View file

@ -15,7 +15,7 @@ CUDA_OUTPUT_FILE = -o $@
# -t=0 is short for --threads, 0 = number of CPUs on the machine
# include PTX for both SM52 (Maxwell) and SM80 (Ampere, our main optimisation target at the moment) + native SASS for current GPU
# this increases compile time by ~5%, but we need >=SM70 PTX for some optimisations, and it allows "cuobjdump --dump-sass" to work
NVCC_FLAGS = -O3 -t=0 --use_fast_math -gencode=arch=compute_52,code=compute_52 -gencode=arch=compute_80,code=compute_80 -arch=native
NVCC_FLAGS = -O3 -t=0 --use_fast_math -gencode=arch=compute_52,code=compute_52 -gencode=arch=compute_80,code=sm_80
NVCC_LDFLAGS = -lcublas -lcublasLt
NVCC_INCLUDES =
NVCC_LDLIBS =