[core][misc] simply output processing with shortcut for non-parallel sampling and non-beam search usecase

[youkaichao]

No description provided.

[youkaichao]

it is a tiny step of #7116

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which consists a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of default ones by unblocking the steps in your fast-check build on Buildkite UI.

Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge).

To run full CI, you can do one of these:

• Comment /ready on the PR
• Add ready label to the PR
• Enable auto-merge.

🚀

[youkaichao]

On H100:

python benchmarks/benchmark_throughput.py --input-len 256 --output-len 256 --model meta-llama/Meta-Llama-3-8B

Main branch (3 trials):

Throughput: 24.32 requests/s, 12453.04 tokens/s
Throughput: 24.11 requests/s, 12342.76 tokens/s
Throughput: 24.29 requests/s, 12434.40 tokens/s

This PR (3 trials):

Throughput: 24.52 requests/s, 12555.73 tokens/s
Throughput: 24.59 requests/s, 12591.35 tokens/s
Throughput: 24.74 requests/s, 12667.68 tokens/s

Already 1.5% improvement in throughput.

[zhuohan123]

LGTM!