0bdcb69fa3
Closes the "dual variants not yet re-benched on v0.20" caveat from PR #23. Confirms no v0.20 regression on fp8 / FP16 paths — code TPS within bench variance of chart values across all 3. n=5 measured + 3 warmup per prompt: | Variant | Chart (prior) narr / code | v0.20 measured narr / code | Δ | |--------------------|----------------------------|------------------------------|-------------------------| | dual.yml | 69.05 / 88.58 | 68.61 / 90.71 (CV 1.8% both) | narr -0.6% / code +2.4% | | dual-dflash.yml | 81.94 / 124.93 | 77.12 / 125.97 (CV 2-4%) | narr -5.9% / code +0.8% | | dual-dflash-noviz | 78.19 / 126.99 | 78.94 / 123.18 (CV 2-3%) | narr +1.0% / code -3.0% | dual-dflash narrative is the only delta outside CV (-5.9%); could be substrate (different driver / power state at chart capture) or a minor DFlash N=5 spec-decode regression on v0.20. Chart value stays — within ±5pp of measured, within bench noise band. Both dual.yml and dual-dflash* are "Genesis-less by design" (zero Genesis env vars). The migration only bumped the image SHA on these — no env-var changes — which is consistent with the flat result. The +50% TPS jump on TQ k8v4 we reported to Sander in discussion #19 came from enabling his full PROD env-var stack on the TQ KV path; fp8 / FP16 dual paths don't share the same patches and don't exhibit a similar bump. Updated `docs/DUAL_CARD.md` performance summary table with the new measured numbers + a per-variant Δ column. Per-config summaries written to `results/v0.20-migration/dual-{yml,dflash,dflash-noviz}.summary`. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
28 lines
675 B
Plaintext
28 lines
675 B
Plaintext
variant: dual.yml
|
|
date: 2026-05-01 PM
|
|
context: 262144
|
|
gpu_mem_util: 0.92
|
|
kv_dtype: fp8_e5m2
|
|
tp: 2
|
|
spec_decode: mtp_n3
|
|
narr_wall_tps: 68.61
|
|
narr_wall_cv: 1.8
|
|
narr_decode_tps: 69.25
|
|
code_wall_tps: 90.71
|
|
code_wall_cv: 1.8
|
|
code_decode_tps: 92.47
|
|
ttft_ms: 146
|
|
mtp_al: 3.44-3.51
|
|
mtp_avg_accept: 0.813-0.838
|
|
vram_per_card_mib: 22246
|
|
kv_pool_tokens: 169600
|
|
max_concurrency: 2.38
|
|
chart_value_narr: 69.05
|
|
chart_value_code: 88.58
|
|
delta_vs_chart_narr: -0.6%
|
|
delta_vs_chart_code: +2.4%
|
|
notes: |
|
|
No env-var changes vs dev205 (dual.yml is "Genesis-less by design").
|
|
v0.20 image bump only. TPS within bench variance of chart values.
|
|
Confirms no v0.20 regression on the fp8 + MTP path.
|