Issue Station
Sign In
Tools hereToolbox

vllm - 💡(How to fix) Fix [Performance]: Quantized KV Cache Throughput/TTFT/TPOT slower?

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

Code Example

The output of `python collect_env.py`
RAW_BUFFERClick to expand / collapse

Proposal to improve performance

hardware: 4090 vllm: 0.20.0 prefill: 10240 tokens decode: 1024 tokens guidellm mode: input tokens/s and output tokens/s is throughput; TTFT and TPOT is synchronous

  • base server: python3 -m vllm.entrypoints.openai.api_server --tensor-parallel-size=4 --model=/home/jessiew/models/Qwen3-32B --no-enable-prefix-caching
  • no calibration server: python3 -m vllm.entrypoints.openai.api_server --tensor-parallel-size=4 --no-calculate-kv-scales --kv-cache-dtype=fp8 --model=/home/jessiew/models/Qwen3-32B --no-enable-prefix-caching
  • random token calibration server: python3 -m vllm.entrypoints.openai.api_server --tensor-parallel-size=4 --kv-cache-dtype=fp8 --model=/home/jessiew/models/Qwen3-32B --no-enable-prefix-caching --calculate-kv-scales

guidellm results:

ServerKV cache sizeinput tokens/soutput tokens/sTTFT msTPOT ms
base72,28823092.3118.73295.122.6
no calibration115,32820311.7143.93311.821.6
random token calibration109,66420161.6132.33290.921.7

Is it expected that a quantized KV cache would show degradation in Throughput, TTFT, and TPOT? Or is my server configuration incorrect? I would be extremely grateful if someone helps me.

Report of performance regression

No response

Misc discussion on performance

No response

Your current environment (if you think it is necessary)

The output of `python collect_env.py`

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#ssr#memory leak#API versioning#request timeout

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

vllm - 💡(How to fix) Fix [Performance]: Quantized KV Cache Throughput/TTFT/TPOT slower?