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 15, 12:15 UTC
Check if bitblas is installed during support check (#17572)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
May 1, 2025 at 22:32 UTC
b4003d11fc7ab82c12c3fb996e853f8d6ea9cd7a
1 file changed
+9
vllm/model_executor/layers/quantization/utils/bitblas_utils.py
+9
@@ -71,6 +71,15 @@ def _check_bitblas_supported(
71
f"Only group_sizes = {BITBLAS_SUPPORTED_GROUP_SIZES} "
72
"are supported.")
73
74
+ # Finally, check if bitblas is installed
75
+ try:
76
+ import bitblas
77
+ if bitblas.__version__ < MINIMUM_BITBLAS_VERSION:
78
+ raise ImportError("bitblas version is wrong. Please "
79
+ f"install bitblas>={MINIMUM_BITBLAS_VERSION}")
80
+ except ImportError:
81
+ return False, "BitBLAS is not installed."
82
+
83
return True, None
84
85