Add aiter attention support in prefill-attention-backend of gpt-oss

[kkHuang-amd]

Motivation

Improve the attention performance on ROCm platform.

Modifications

aiter_backend.py to support sliding window attention with sink
server_args.py to add "aiter" for supporting list

Accuracy Tests

Server command

SGLANG_USE_AITER=1 python3 -m sglang.launch_server --model-path /dockerx/data/models/openai/gpt-oss-120b/ --tp 8 --trust-remote-code --chunked-prefill-size 130172 --max-running-requests 128 --mem-fraction-static 0.85 --prefill-attention-backend aiter --decode-attention-backend triton --port 8000

Client command

/sglang# python3 benchmark/gsm8k/bench_sglang.py --num-questions 2000 --parallel 2000 --port 8000 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1319/1319 [00:50<00:00, 26.00it/s] Accuracy: 0.829 Invalid: 0.014 Latency: 86.113 s Output throughput: 5072.316 token/s

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review Process

1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
4. After green CI and required approvals, ask Merge Oncalls to merge.

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!