disable prefix on random for qwen3.5 & glm mi355

[functionstackx]

https://github.com/SemiAnalysisAI/InferenceX/blob/main/benchmarks/single_node/qwen3.5_bf16_mi355x.sh
https://github.com/SemiAnalysisAI/InferenceX/blob/main/benchmarks/single_node/qwen3.5_fp8_mi355x.sh
https://github.com/SemiAnalysisAI/InferenceX/blob/main/benchmarks/single_node/glm5_fp8_mi355x.sh

[functionstackx]

waiting till mi355 cluster gets ungg'ed

[functionstackx]

@claude create an PR in the meanwhile

[Klaud-Cold]

Claude finished @functionstackx's task in 1m 5s —— View job • claude/issue-971-20260329-0014 • Create PR ➔

---

Disable prefix caching on random for qwen3.5 & glm5 mi355x

• Read benchmark scripts and understand current prefix/random settings
• Add --no-enable-prefix-caching to sglang launch_server in all 3 scripts
• Commit and push changes

Added --no-enable-prefix-caching to the SGLang server launch command in:

• benchmarks/single_node/qwen3.5_bf16_mi355x.sh
• benchmarks/single_node/qwen3.5_fp8_mi355x.sh
• benchmarks/single_node/glm5_fp8_mi355x.sh

|

[functionstackx]

@claude add perf change log to the end of perfchangelog .yaml

[Klaud-Cold]

Claude finished @functionstackx's task in 1m 11s —— View job • claude/issue-971-20260329-0017 • Create PR ➔

---

Add perf changelog entry

• Read current perf-changelog.yaml and benchmark scripts
• Add changelog entry for disabling prefix caching on random workloads
• Commit and push changes

Added --no-enable-prefix-caching to SGLang server launch in all 3 benchmark scripts and appended a perf changelog entry to perf-changelog.yaml for config keys qwen3.5-bf16-mi355x-sglang, qwen3.5-fp8-mi355x-sglang, and glm5-fp8-mi355x-sglang.

• claude/issue-971-20260329-0017