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 14, 05:30 UTC
[misc] fix multiple arch wheels for the nightly index (#23110)
Signed-off-by: youkaichao <youkaichao@gmail.com>
youkaichao committed
Aug 21, 2025 at 05:15 UTC
1b125004bea9f4cd120d3ce96dc1d3a2962ebace
1 file changed
+16
-2
.buildkite/generate_index.py
+16
-2
@@ -8,7 +8,8 @@ template = """<!DOCTYPE html>
8
<html>
9
<body>
10
<h1>Links for vLLM</h1/>
11
- <a href="../{wheel_html_escaped}">{wheel}</a><br/>
11
+ <a href="../{x86_wheel_html_escaped}">{x86_wheel}</a><br/>
12
+ <a href="../{arm_wheel_html_escaped}">{arm_wheel}</a><br/>
13
</body>
14
</html>
15
"""
@@ -21,7 +22,20 @@ filename = os.path.basename(args.wheel)
22
23
with open("index.html", "w") as f:
24
print(f"Generated index.html for {args.wheel}")
25
+ if "x86_64" in filename:
26
+ x86_wheel = filename
27
+ arm_wheel = filename.replace("x86_64", "aarch64")
28
+ elif "aarch64" in filename:
29
+ x86_wheel = filename.replace("aarch64", "x86_64")
30
+ arm_wheel = filename
31
+ else:
32
+ raise ValueError(f"Unsupported wheel: {filename}")
33
# cloudfront requires escaping the '+' character
34
f.write(
26
- template.format(wheel=filename, wheel_html_escaped=filename.replace("+", "%2B"))
35
+ template.format(
36
+ x86_wheel=x86_wheel,
37
+ x86_wheel_html_escaped=x86_wheel.replace("+", "%2B"),
38
+ arm_wheel=arm_wheel,
39
+ arm_wheel_html_escaped=arm_wheel.replace("+", "%2B"),
40
+ )
41
)