ebca0c8921
Closes task #230. Measured PFlash NIAH compression at 16K-260K source contexts on 1× 24 GB / 3090 single-card. Result: PFlash works flawlessly up to 131K source. Compresses 131,068 tokens to 6,524 (5%) in 10.8s with NIAH key + answer both retained. Vanilla llama.cpp pp131072 takes ~257s per Luce's published numbers, so PFlash alone is ~24× faster at this context. End-to-end TTFT (PFlash + target prefill on 6.5K) would be ~12-13s vs ~257s = ~20×. Above 131K, drafter ephemeral forward-pass tensors (K_curr/V_curr/Q_last at full sequence length) exceed 24 GB. K-cache quantization (--pflash-k-type q8_0) doesn't help — the failing allocs are forward-pass not cache, confirmed by separate bench at 200K/260K with identical OOM at the same layer numbers. @weicj's PR #78 claim of 24K → 262K dual-GPU phase split is neither refuted nor reproduced. Their setup was 2× 22 GB Ti with target also loaded co-resident on one card; the "24K" was target+drafter combined. Our 131K is drafter-alone on 24 GB. Reproducing 262K specifically would require investigation of their drafter config (chunk_size, lookahead, BSA window) — drafter activation footprint at 200K+ is the binding constraint regardless of GPU count. Practical recommendation for 24 GB / 3090 single-card users: PFlash is shippable for source contexts ≤ 131K. The ~24× TTFT speedup is genuine and quality holds. Above 131K, fall back to vanilla llama.cpp prefill or wait for upstream drafter optimizations. Adds: - BENCHMARKS.md "PFlash long-context compression on 1× 3090" subsection with full per-context table + drafter ceiling explanation - results/lucebox-pflash-niah-20260504-150321/ (BF16 K cache run) - results/lucebox-pflash-niah-q8k-20260504-150600/ (q8_0 K cache run) This closes our active investigation of the Luce surface — three benches done (DFlash same-card 73.97 mean, K8V4 same-card 74.68 mean, PFlash compression ceiling 131K). Recommendation surface narrows to: PFlash at ≤131K is the one piece of Luce that beats vLLM dual.yml on TTFT for that workload class. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
59 lines
1.7 KiB
JSON
59 lines
1.7 KiB
JSON
{
|
|
"date": "2026-05-04",
|
|
"mode": "dual_gpu_pflash_phase_split",
|
|
"pflash_gpu": 0,
|
|
"pflash_daemon_ready_s": 0.6027854409767315,
|
|
"pflash_drafter": "/mnt/models/gguf/qwen3-draft-0.6b/Qwen3-0.6B-BF16.gguf",
|
|
"tokenizer": "Qwen/Qwen3-0.6B",
|
|
"keep_ratio": 0.05,
|
|
"lookahead": 2,
|
|
"chunk_size": 32,
|
|
"pool_kernel": 13,
|
|
"pflash_k_type": "q8_0",
|
|
"cases": [
|
|
{
|
|
"name": "niah_ctx131068",
|
|
"source_tokens": 131068,
|
|
"compressed_tokens": 6524,
|
|
"compress_wall_s": 10.65377599094063,
|
|
"compress_tok_s": 12302.492572722838,
|
|
"compression_ratio": 0.04977568895535142,
|
|
"retained_key": true,
|
|
"retained_answer": true
|
|
},
|
|
{
|
|
"name": "niah_ctx199996",
|
|
"source_tokens": 199996,
|
|
"compressed_tokens": 0,
|
|
"compress_wall_s": 0.10241466399747878,
|
|
"compress_tok_s": 1952806.2895848923,
|
|
"compression_ratio": 0.0,
|
|
"retained_key": false,
|
|
"retained_answer": false
|
|
},
|
|
{
|
|
"name": "niah_ctx259996",
|
|
"source_tokens": 259996,
|
|
"compressed_tokens": 0,
|
|
"compress_wall_s": 0.10354933398775756,
|
|
"compress_tok_s": 2510841.837290222,
|
|
"compression_ratio": 0.0,
|
|
"retained_key": false,
|
|
"retained_answer": false
|
|
}
|
|
],
|
|
"resource_summary": {
|
|
"samples": 12,
|
|
"mem_max_mib": 21024.0,
|
|
"temp_max_c": 54.0,
|
|
"fan_max_pct": 58.0,
|
|
"power_avg_w": 190.06916666666666,
|
|
"power_max_w": 229.25,
|
|
"util_avg_pct": 82.0,
|
|
"util_max_pct": 100.0
|
|
},
|
|
"logs": {
|
|
"pflash": "/opt/ai/github/club-3090/results/lucebox-pflash-niah-q8k-20260504-150600/pflash_daemon.log",
|
|
"monitor": "/opt/ai/github/club-3090/results/lucebox-pflash-niah-q8k-20260504-150600/gpu_monitor.csv"
|
|
}
|
|
} |