[CI/Build] respect the common environment variable MAX_JOBS (#3600)

docs/source/getting_started/installation.rst

@@ -60,6 +60,15 @@ You can also build and install vLLM from source:
     $ cd vllm
     $ pip install -e .  # This may take 5-10 minutes.
 
+.. tip::
+    To avoid your system being overloaded, you can limit the number of compilation jobs
+    to be run simultaneously, via the environment variable `MAX_JOBS`. For example:
+
+    .. code-block:: console
+
+        $ export MAX_JOBS=6
+        $ pip install -e .
+
 .. tip::
     If you have trouble building vLLM, we recommend using the NVIDIA PyTorch Docker image.
 

setup.py

 import io
 import os
 import re
+import logging
 import subprocess
 import sys
 from typing import List
@@ -13,6 +14,7 @@ import torch
 from torch.utils.cpp_extension import CUDA_HOME
 
 ROOT_DIR = os.path.dirname(__file__)
+logger = logging.getLogger(__name__)
 
 # vLLM only supports Linux platform
 assert sys.platform.startswith(
@@ -54,9 +56,14 @@ class cmake_build_ext(build_ext):
     # Determine number of compilation jobs and optionally nvcc compile threads.
     #
     def compute_num_jobs(self):
+        num_jobs = os.environ.get("MAX_JOBS", None)
+        if num_jobs is not None:
+            num_jobs = int(num_jobs)
+            logger.info(f"Using MAX_JOBS={num_jobs} as the number of jobs.")
+        else:
             try:
-            # os.sched_getaffinity() isn't universally available, so fall back
-            # to os.cpu_count() if we get an error here.
+                # os.sched_getaffinity() isn't universally available, so fall
+                #  back to os.cpu_count() if we get an error here.
                 num_jobs = len(os.sched_getaffinity(0))
             except AttributeError:
                 num_jobs = os.cpu_count()