Skip to content

Update README.md #184

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
thesues wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Update README.md #184

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 22 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,28 @@ Currently InfiniStore has been integrated with vLLM. The integration is done via

Integration with SGLang and other inference engines are in progress.


# Performance

Working with LMCache

bench command:

```
python3 vllm/benchmarks/benchmark_serving.py --backend openai-chat --model /data/models/QwQ-32B --base-url http://127.0.0.1:8080 --endpoint /v1/chat/completions --num-prompts 16 --request-rate 1 --metric_percentiles '50,90,95,99' --goodput ttft:5000 tpot:50 --max-concurrency 16 --random-input-len 1000 --random-output-len 250 --dataset-name random --ignore-eos --seed 114
```

TTFT/TPOP results:

| ---- | PD infinistore | PD lmcache | PD redis| origin vllm|
|----------|-------------------|----------------|------|---------|
| TTFT |1872ms | 3700ms |2700ms| 1712ms |
| TPOT | 50ms | 59ms |86ms| 84ms|


compared to origin vllm, infinistore in PD shows very stable and shorter TPOT.


# Installation

## Install from PIP
Expand Down