forked from vllm-project/vllm
-
Notifications
You must be signed in to change notification settings - Fork 5
36 lines (32 loc) · 980 Bytes
/
codspeed.yml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
name: codspeed-benchmarks
on:
# Run on pushes to the main branch
push:
branches:
- "master" # or "main"
# Run on pull requests
pull_request:
# `workflow_dispatch` allows CodSpeed to trigger backtest
# performance analysis in order to generate initial data.
workflow_dispatch:
jobs:
benchmarks:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
# ...
# Setup your environment here:
# - Configure your Python/Rust/Node version
# - Install your dependencies
# - Build your benchmarks (if using a compiled language)
# ...
- uses: actions/setup-python@v3
with:
python-version: "3.10"
- name: Install project dependencies
run: pip install -e . && pip install pytest
- name: Run benchmarks
uses: CodSpeedHQ/action@v2
with:
token: ${{ secrets.codspeed_token }}
run: "pytest tests/kernels/test_layernorm.py --codspeed"