Skip to content

Conversation

@fmiao2372
Copy link
Contributor

Motivation

add benchmark scripts for hpu

💡 If this PR is a Cherry Pick, the PR title needs to follow the format by adding the [Cherry-Pick] label at the very beginning and appending the original PR ID at the end. For example, [Cherry-Pick][CI] Add check trigger and logic(#5191)

💡 如若此PR是Cherry Pick,PR标题需遵循格式,在最开始加上[Cherry-Pick]标签,以及最后面加上原PR ID,例如[Cherry-Pick][CI] Add check trigger and logic(#5191)

Modifications

benchmark server and cli scripts for hpu with log-parsing scripts

Usage or Command

please refer to the README in this patch

Accuracy Tests

No accuracy impact with this updates

Checklist

  • [Done] Add at least a tag in the PR title.
    • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
    • You can add new tags based on the PR content, but the semantics must be clear.
  • [Done] Format your code, run pre-commit before commit.
  • [Done] Add unit tests. Please write the reason in this PR if no unit tests.
    conducted by local tests
  • Provide accuracy results.
  • If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

Copilot AI review requested due to automatic review settings December 1, 2025 01:22
@paddle-bot
Copy link

paddle-bot bot commented Dec 1, 2025

Thanks for your contribution!

Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR adds benchmark scripts and utilities for Intel HPU testing in FastDeploy. The changes include server/client scripts for running benchmarks, log parsing tools for analyzing performance data, and configuration files for model deployment on HPU hardware.

Key Changes:

  • Benchmark server and client scripts supporting both fixed-length and ShareGPT datasets
  • Log parsing and visualization utilities for performance analysis
  • Configuration and demo scripts for HPU deployment

Reviewed changes

Copilot reviewed 11 out of 12 changed files in this pull request and generated 13 comments.

Show a summary per file
File Description
benchmark_paddle_hpu_server.sh Server startup script with HPU-specific environment configuration
benchmark_paddle_hpu_server_sharegpt.sh Server script variant supporting ShareGPT dataset benchmarking
benchmark_paddle_hpu_cli.sh Client benchmark script for fixed-length input/output tests
benchmark_paddle_hpu_cli_sharegpt.sh Client benchmark script for ShareGPT dataset tests
parse_benchmark_logs.py Log parser extracting metrics and profiling data into CSV
draw_benchmark_data.py Visualization tool generating graphs from profile logs
offline_demo.py Offline inference demo for HPU
run_offline_demo.sh Shell script to run offline demo with HPU configuration
bench_gsm8k.py GSM8K benchmark evaluation script
intel_hpu_serving_benchmark.md Documentation for running HPU benchmarks
eb45-300b-a47b-32k-bf16.yaml Model configuration for ERNIE-4.5-300B

@codecov-commenter
Copy link

codecov-commenter commented Dec 1, 2025

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (develop@8aec3ac). Learn more about missing BASE report.

Additional details and impacted files
@@            Coverage Diff             @@
##             develop    #5304   +/-   ##
==========================================
  Coverage           ?   59.60%           
==========================================
  Files              ?      324           
  Lines              ?    39702           
  Branches           ?     5973           
==========================================
  Hits               ?    23666           
  Misses             ?    14150           
  Partials           ?     1886           
Flag Coverage Δ
GPU 59.60% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

{"question": "Janet\u2019s ducks lay 16 eggs per day. She eats three for breakfast every morning and bakes muffins for her friends every day with four. She sells the remainder at the farmers' market daily for $2 per fresh duck egg. How much in dollars does she make every day at the farmers' market?", "answer": "Janet sells 16 - 3 - 4 = <<16-3-4=9>>9 duck eggs a day.\nShe makes 9 * 2 = $<<9*2=18>>18 every day at the farmer\u2019s market.\n#### 18"}
{"question": "A robe takes 2 bolts of blue fiber and half that much white fiber. How many bolts in total does it take?", "answer": "It takes 2/2=<<2/2=1>>1 bolt of white fiber\nSo the total amount of fabric is 2+1=<<2+1=3>>3 bolts of fabric\n#### 3"}
{"question": "Josh decides to try flipping a house. He buys a house for $80,000 and then puts in $50,000 in repairs. This increased the value of the house by 150%. How much profit did he make?", "answer": "The cost of the house and repairs came out to 80,000+50,000=$<<80000+50000=130000>>130,000\nHe increased the value of the house by 80,000*1.5=<<80000*1.5=120000>>120,000\nSo the new value of the house is 120,000+80,000=$<<120000+80000=200000>>200,000\nSo he made a profit of 200,000-130,000=$<<200000-130000=70000>>70,000\n#### 70000"}
{"question": "James decides to run 3 sprints 3 times a week. He runs 60 meters each sprint. How many total meters does he run a week?", "answer": "He sprints 3*3=<<3*3=9>>9 times\nSo he runs 9*60=<<9*60=540>>540 meters\n#### 540"}
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这种数据集文件是否可以用一个下载链接的形式给出

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

源文件已经删了,因为repo里面已经有了

```bash
./benchmark_paddle_hpu_cli.sh
```

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这里个example的作用应该也是面向用户实际使用和复现测试结果的。这个文档是否可以再详细一些,介绍下不同脚本的使用场景,包括一些用法、脚本参数、启动环境变量的含义

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

已添加

@zoooo0820 zoooo0820 merged commit 429dd2b into PaddlePaddle:develop Dec 2, 2025
15 of 17 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants