{
 "id": "h100-qwen3-30b-a3b-prefix-cache-vllm",
 "status": "published",
 "title": "Prefix caching on one H100: Qwen3-30B-A3B-Instruct-2507 FP8, vLLM 0.31.0",
 "published": "2026-10-29",
 "summary": "Quick-mode measurement on TokenWatt's production API node, one H100 SXM on RunPod Secure Cloud (October 29, 2026): Qwen3-30B-A3B-Instruct-2507 FP8 on vLLM 0.31.0 with default automatic prefix caching, chat-like workload (median 1,024 input / 256 output tokens). The same SLO search ran four times with 0%, 50%, 75% and 90% of every prompt drawn from eight shared prefixes; cache hit rates are vLLM's own prefix-cache counters (0%, 48.7%, 74.0%, 89.2%). Capacity within p99 TTFT \u2264 2 s and p99 TPOT \u2264 50 ms rose from 11.5 to 21.0 requests per second. Accuracy was checked on the same node in separate runs (GSM8K 96% quick, 94.5% full); these runs skipped the accuracy gate.",
 "hardware": {
  "vendor": "NVIDIA",
  "accelerator": "NVIDIA H100 80GB HBM3",
  "count": 1,
  "rated_bw_tbs_per_gpu": 3.35,
  "driver": "595.91.07",
  "host": "RunPod Secure H100 SXM (TokenWatt API production node)"
 },
 "software": {
  "engine": "vllm",
  "engine_version": "0.31.0",
  "kernels": "vllm/vllm-openai:v0.31.0 --max-model-len 32768 --enable-auto-tool-choice --tool-call-parser hermes"
 },
 "model": {
  "name": "Qwen3-30B-A3B-Instruct-2507-FP8",
  "shape": "MoE",
  "quantization": "FP8 weights, BF16 KV cache",
  "parallelism": "1 replica \u00d7 TP1"
 },
 "slo": {
  "ttft_p99_ms": 2000,
  "tpot_p99_ms": 50
 },
 "workload": {
  "trace": "chat-mix-v1",
  "input_tokens": "median 1,024 \u00b7 p90 4,096",
  "output_tokens": "median 256 \u00b7 p90 1,024",
  "arrival": "Poisson, rate swept to SLO boundary"
 },
 "power": {
  "source": "GPU board power sum (NVIDIA)",
  "kind": "board",
  "sampling_hz": 1,
  "window": "120 s steady state"
 },
 "accuracy": {
  "suite": "none",
  "threshold_pts": 1.0
 },
 "runs": 2,
 "variance_band_pct": 1.5,
 "fleet": {
  "cards": 64
 },
 "configs": [
  {
   "key": "other",
   "label": "No prompt reuse (0% cache hits)",
   "engine": "vllm 0.31.0 \u00b7 KV BF16",
   "mbu_pct": 60.7,
   "node_power_kw": 0.607,
   "goodput_tok_s": 4740.3,
   "ttft_p99_ms": 380.0,
   "tpot_p99_ms": 40.6,
   "slo_pass_count": 2,
   "repeats": 2,
   "throttle_counts": {
    "sw_power_cap": 181
   },
   "accuracy": null
  },
  {
   "key": "other",
   "label": "50% shared prefix (48.7% cache hits)",
   "engine": "vllm 0.31.0 \u00b7 KV BF16",
   "mbu_pct": 66.7,
   "node_power_kw": 0.647,
   "goodput_tok_s": 6677.7,
   "ttft_p99_ms": 294.0,
   "tpot_p99_ms": 46.1,
   "slo_pass_count": 2,
   "repeats": 2,
   "throttle_counts": {
    "sw_power_cap": 199
   },
   "accuracy": null
  },
  {
   "key": "other",
   "label": "75% shared prefix (74.0% cache hits)",
   "engine": "vllm 0.31.0 \u00b7 KV BF16",
   "mbu_pct": 76.0,
   "node_power_kw": 0.655,
   "goodput_tok_s": 8040.7,
   "ttft_p99_ms": 248.0,
   "tpot_p99_ms": 42.1,
   "slo_pass_count": 2,
   "repeats": 2,
   "throttle_counts": {
    "sw_power_cap": 213
   },
   "accuracy": null
  },
  {
   "key": "other",
   "label": "90% shared prefix (89.2% cache hits)",
   "engine": "vllm 0.31.0 \u00b7 KV BF16",
   "mbu_pct": 81.2,
   "node_power_kw": 0.623,
   "goodput_tok_s": 8508.3,
   "ttft_p99_ms": 257.0,
   "tpot_p99_ms": 39.4,
   "slo_pass_count": 2,
   "repeats": 2,
   "throttle_counts": {
    "sw_power_cap": 112
   },
   "accuracy": null
  }
 ],
 "not_covered": [
  "Whole-node power (no BMC access on this host)",
  "Context lengths beyond the chat-mix-v1 trace",
  "Quick mode: short windows, few repeats",
  "Accuracy: no accuracy gate was run"
 ],
 "reproduce": "tokenwatt run --spec live-qwen3-30b-a3b-prefix-cache.spec.yaml --label 'prefix-0' --role other --out runs/prefix-0.json\ntokenwatt run --spec live-qwen3-30b-a3b-prefix-cache.spec.yaml --label 'prefix-0.5' --role other --out runs/prefix-0-5.json\ntokenwatt run --spec live-qwen3-30b-a3b-prefix-cache.spec.yaml --label 'prefix-0.75' --role other --out runs/prefix-0-75.json\ntokenwatt run --spec live-qwen3-30b-a3b-prefix-cache.spec.yaml --label 'prefix-0.9' --role other --out runs/prefix-0-9.json\n\ntokenwatt report runs/prefix-0.json runs/prefix-0-5.json runs/prefix-0-75.json runs/prefix-0-9.json --title 'Prefix caching on one H100: Qwen3-30B-A3B-Instruct-2507 FP8, vLLM 0.31.0'",
 "spec": {
  "name": "live-qwen3-30b-a3b-prefix-cache",
  "hardware": {
   "adapter": "nvidia",
   "rated_bw_tbs_per_gpu": null,
   "gpus": null,
   "host": "RunPod Secure H100 SXM (TokenWatt API production node)"
  },
  "engine": {
   "endpoint": "http://127.0.0.1:8000/v1",
   "name": "vllm",
   "version": "0.31.0",
   "kernels": "vllm/vllm-openai:v0.31.0 --max-model-len 32768 --enable-auto-tool-choice --tool-call-parser hermes",
   "api": "completions",
   "model": "qwen3-30b-a3b-instruct-2507",
   "ignore_eos": true,
   "api_key_env": null,
   "extra_body": {}
  },
  "model": {
   "name": "Qwen3-30B-A3B-Instruct-2507-FP8",
   "config": "/workspace/hf/hub/models--Qwen--Qwen3-30B-A3B-Instruct-2507-FP8/snapshots/5a5a776300a41aaa681dd7ff0106608ef2bc90db/config.json",
   "weight_dtype": "fp8",
   "kv_dtype": "bf16",
   "quantization": "",
   "replicas": 1,
   "tensor_parallel": 1,
   "vocab_size": null,
   "geometry": {}
  },
  "slo": {
   "ttft_p99_ms": 2000,
   "tpot_p99_ms": 50
  },
  "workload": {
   "trace": "chat-mix-v1",
   "input_tokens": null,
   "output_tokens": null,
   "seed": 0,
   "shared_prefix": 0,
   "prefix_pool": 8
  },
  "measurement": {
   "warmup_s": 45,
   "window_s": 120,
   "search_window_s": 60,
   "repeats": 2,
   "request_timeout_s": 300,
   "max_concurrency": 8192
  },
  "sweep": {
   "rates": null,
   "start_rps": 4,
   "growth": 1.6,
   "max_rps": 2000,
   "bisect_steps": 3,
   "backoff": 0.9,
   "max_backoffs": 4,
   "confirm": "backoff",
   "floor_rps": null,
   "confirm_steps": 5
  },
  "accuracy": {
   "suite": "/workspace/tw/suites/gsm8k-200.jsonl",
   "name": "gsm8k-5shot",
   "scorer": "numeric",
   "max_drop_pts": 1.0,
   "seeds": 1,
   "max_tokens": 384,
   "stop": [
    "\n\nQuestion:"
   ],
   "concurrency": 32,
   "limit": 50
  },
  "power": {
   "source": "board",
   "sampling_hz": 1,
   "url": null,
   "username": null,
   "password_env": null,
   "verify_tls": false,
   "chassis": null,
   "host": null,
   "command": null,
   "field": null
  },
  "not_covered": [
   "Whole-node power (no BMC access on this host)",
   "Context lengths beyond the chat-mix-v1 trace",
   "Quick mode: short windows, few repeats"
  ]
 }
}