igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

Class v6 floor lane 4: the first sweep's stock rows (5080, 4070, 3080, 3060) in the tiers table

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Documents-only replay of 5e02a81f5 (9d3bc250cb) for the box mirror master
This commit is contained in:
igneum-labs 2026-10-08 12:29:41 +00:00
parent 59a4fae00c
commit 4e9459b49f
1 changed files with 33 additions and 33 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

66
app/igneum-app/tiers/class-v5-tiers.json
View file

@ -138,7 +138,7 @@
"source": "stock",
"label": "measured",
"uj": 3.54,
"note": "PC 1 (Windows) stock; a rented Linux host read 143.4 W at the same rate on class v3 against 169.7 W here: the two methods differ and neither is corrected"
"note": "PC 1 (Windows) stock; a rented Linux host read 143.4 W at the same rate on class v3 against 169.7 W here: the two methods differ and neither is corrected; the denominator sweep's Vast 5080 (13:16 UK) read class v4 70.95 MH/s at 239.8 W (3.38), v4 over v3 +41 percent on one host, so the rented and the PC stock rows agree on this card within 5 percent"
}
],
"src": "docs/bench-log.md, the RTX 5080 ladder, 8 October 2026 00:45 to 01:54 UTC (PC 1)"
@ -567,7 +567,7 @@
"source": "stock",
"label": "measured",
"uj": 3.69,
"note": "Ember run 6 stock on class v3 (28.72 MH/s at 106.0 W); the bench-log 5 October row 28.7 at 102.7 W (3.58)"
"note": "Ember run 6 stock on class v3 (28.72 MH/s at 106.0 W); the bench-log 5 October row 28.7 at 102.7 W (3.58); the denominator sweep's Vast 4070 (13:21 UK) read class v3 30.83 MH/s at 107.1 W (3.47) stock, class v5 30.88 MATCH"
}
],
"src": "docs/plans/counter-asic-3-status.md item 8 (the 4070 rows, 6 October); docs/bench-log.md Ember run 6"
@ -828,9 +828,9 @@
"memory_gb": 10,
"label": "estimated",
"stock": {
"mhs": 42.0,
"w": 270.0,
"uj": 6.43,
"mhs": 49.83,
"w": 230.0,
"uj": 4.62,
"class": "v4"
},
"tiers": [
@ -840,13 +840,13 @@
"power_pct": 80,
"mem_mhz": 0,
"limit_w": 320,
"mhs": 42.0,
"w": 240.0,
"mhw": 0.175,
"mhs": 49.5,
"w": 208.0,
"mhw": 0.238,
"source": "measured",
"label": "estimated",
"uj": 5.7,
"note": "the Ampere cap lever (10 to 15 percent); the 3080 Ti measured class v3 58.9 MH/s at 293.5 W (4.98) and, at a 220 W host cap with the SM at 749 MHz, class v4 42.6 at 216 W (5.08) with 28 percent of the rate lost: the Ampere cap's known-failed row"
"uj": 4.2,
"note": "the Ampere cap lever (10 percent) on the class v4 stock estimate; the 180 W host-limited class v3 row held the full rate, the 130 W cap did not"
},
{
"id": "balanced",
@ -854,12 +854,12 @@
"power_pct": 90,
"mem_mhz": 0,
"limit_w": 320,
"mhs": 42.0,
"w": 255.0,
"mhw": 0.1647,
"mhs": 49.83,
"w": 218.0,
"mhw": 0.2286,
"source": "measured",
"label": "estimated",
"uj": 6.07,
"uj": 4.37,
"note": "one power rung under stock"
},
{
@ -868,13 +868,13 @@
"power_pct": 100,
"mem_mhz": 0,
"limit_w": 320,
"mhs": 42.0,
"w": 270.0,
"mhw": 0.1556,
"mhs": 49.83,
"w": 230.0,
"mhw": 0.2167,
"source": "stock",
"label": "estimated",
"uj": 6.43,
"note": "the 0.3.12 rented row 40.82 MH/s at 204.9 W was an earlier class; the class v3 and v4 stock rows are owed to the denominator sweep"
"uj": 4.62,
"note": "the denominator sweep (13:28 UK): class v3 49.83 MH/s at 177.8 W (3.57) on a 180 W host limit, class v5 44.31 on the same host; the class v4 pod at a 130 W host cap read 24.0 MH/s at 129.2 W (5.38) with the SM at 501 MHz: the Ampere cap's known-failed shape; the class v4 stock watts by the 3080 Ti's +30 percent: about 230 W (4.62); stock partly measured, lock owed"
}
],
"src": "docs/analysis/prover-tiers-real-cards.md; ~/igneum-fleet/cardbench/rows.jsonl (the 3080 Ti rows); the sweep pending"
@ -951,8 +951,8 @@
"label": "estimated",
"stock": {
"mhs": 26.89,
"w": 145.0,
"uj": 5.39,
"w": 166.1,
"uj": 6.18,
"class": "v4"
},
"tiers": [
@ -963,12 +963,12 @@
"mem_mhz": 0,
"limit_w": 170,
"mhs": 26.89,
"w": 132.0,
"mhw": 0.2037,
"w": 150.0,
"mhw": 0.1793,
"source": "measured",
"label": "estimated",
"uj": 4.9,
"note": "class v3 measured 26.89 MH/s at 111.6 W; the class v4 stock by the Ampere premium; the 3060 Ti at a 130 W host cap held 33.06 MH/s at 128.7 W under class v4 (3.89): a cap row measured on a sibling"
"uj": 5.58,
"note": "the Ampere cap lever (10 percent) on the measured class v4 stock row; the 3060 Ti at a 130 W host cap held 33.06 MH/s at 128.7 W under class v4 (3.89): a cap row measured on a sibling"
},
{
"id": "balanced",
@ -977,11 +977,11 @@
"mem_mhz": 0,
"limit_w": 170,
"mhs": 26.89,
"w": 138.0,
"mhw": 0.1949,
"w": 158.0,
"mhw": 0.1702,
"source": "measured",
"label": "estimated",
"uj": 5.13,
"uj": 5.88,
"note": "one power rung under stock"
},
{
@ -991,12 +991,12 @@
"mem_mhz": 0,
"limit_w": 170,
"mhs": 26.89,
"w": 145.0,
"mhw": 0.1854,
"w": 166.1,
"mhw": 0.1619,
"source": "stock",
"label": "estimated",
"uj": 5.39,
"note": "the class v4 stock row is owed to the denominator sweep"
"label": "measured",
"uj": 6.18,
"note": "the denominator sweep's Vast 3060 (13:19 UK): class v4 26.89 MH/s at 166.1 W (6.18), v4 over v3 +44.6 percent on that host (the 7 October pod read class v3 26.89 at 111.6 W); stock measured, lock owed"
}
],
"src": "~/igneum-fleet/cardbench/rows.jsonl, 7 October (class v3); the sweep pending"