A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 17, 04:15 UTC
[Minor Fix] Use cupy-cuda11x in CUDA 11.8 build (#3256)
陈序 committed
Mar 14, 2024 at 00:43 UTC
739c350c1926682f435316294491aa54661849b6
1 file changed
+6
setup.py
+6
@@ -431,6 +431,12 @@ def get_requirements() -> List[str]:
431
else:
432
with open(get_path("requirements.txt")) as f:
433
requirements = f.read().strip().split("\n")
434
+ if nvcc_cuda_version <= Version("11.8"):
435
+ # replace cupy-cuda12x with cupy-cuda11x for cuda 11.x
436
+ for i in range(len(requirements)):
437
+ if requirements[i].startswith("cupy-cuda12x"):
438
+ requirements[i] = "cupy-cuda11x"
439
+ break
440
return requirements
441
442