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 6, 23:00 UTC
[ci] Bound nvidia-cudnn-frontend version (#36719)
Signed-off-by: khluu <khluu000@gmail.com>
Kevin H. Luu committed
Mar 10, 2026 at 21:17 UTC
82b110d50ee0516e8cf76ba1388cd571f7811f34
1 file changed
+3
requirements/cuda.txt
+3
@@ -10,6 +10,9 @@ torchaudio==2.10.0
10
torchvision==0.25.0 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
11
# FlashInfer should be updated together with the Dockerfile
12
flashinfer-python==0.6.4
13
+# Cap nvidia-cudnn-frontend (transitive dep of flashinfer) due to
14
+# breaking changes in 1.19.0
15
+nvidia-cudnn-frontend>=1.13.0,<1.19.0
16
17
# QuACK and Cutlass DSL for FA4 (cute-DSL implementation)
18
nvidia-cutlass-dsl>=4.4.0.dev1