Move flashinfer-python to optional extra vllm[flashinfer] (#21959)

Signed-off-by: mgoin <mgoin64@gmail.com>

Author: mgoin

requirements/cuda.txt

@@ -11,6 +11,4 @@ torchaudio==2.7.1
 # These must be updated alongside torch
 torchvision==0.22.1 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
 # https://github.com/facebookresearch/xformers/releases/tag/v0.0.31
-xformers==0.0.31; platform_system == 'Linux' and platform_machine == 'x86_64'  # Requires PyTorch >= 2.7
-# FlashInfer should be updated together with the Dockerfile
-flashinfer_python==0.2.9rc2
+xformers==0.0.31; platform_system == 'Linux' and platform_machine == 'x86_64'  # Requires PyTorch >= 2.7

setup.py

@@ -671,7 +671,9 @@ def _read_requirements(filename: str) -> list[str]:
         ["runai-model-streamer >= 0.13.3", "runai-model-streamer-s3", "boto3"],
         "audio": ["librosa", "soundfile",
                   "mistral_common[audio]"],  # Required for audio processing
-        "video": []  # Kept for backwards compatibility
+        "video": [],  # Kept for backwards compatibility
+        # FlashInfer should be updated together with the Dockerfile
+        "flashinfer": ["flashinfer-python==0.2.9rc2"],
     },
     cmdclass=cmdclass,
     package_data=package_data,