The hardware roster · one machine
the darkroom's box — two RTX 3090s
whole machine·10 figures, from 1 exhibit · named in 2 exhibits
This roster is in beta and still being tested: every figure is proved against the page it came from, and the way they are read and shown is still changing.
a consumer tower built around two renewed RTX 3090s on an ordinary two-slot board, with desktop memory and a desktop power supply; the box the price series prices, which its own page described as never off until a dated correction of 2026-09-18 made it a bench machine, switched on when there is a bench to run
It holds — RTX 3090 (24 GB).
Narrow these 10 figuresevery power cappower cap: 280 W · 5 of 10 figurespower cap: stated in words, not as a number · 1 of 10 figurespower cap: not stated · 4 of 10 figures
The rail is CSS and nothing else — no script runs on this page. It narrows by model and by power cap independently, and the line above says what each one is set to; the size of the two together is the one number it cannot count. A group or a band with nothing left in it goes with its rows, so what you are left looking at is what matched. A narrowed view is not part of the address, so it cannot be linked — each card’s own page is a URL, and that is the thing to share.
Power, temperature, and what it costs
Grouped by model and by what was measured, and inside a group sorted by power cap ascending — alphabetical and numeric, never by value. The heading of a group carries everything that does not change inside it; each row carries the value and the conditions that do. A condition the page does not state reads not stated, which is a third thing and not an uncapped card. A † means the same reading is printed in a second place on its exhibit — one reading, two printings, not two figures.
| value | what moves inside this group |
|---|---|
| not stated draw at the wallWquantisation, runtime, context, batch or steps and concurrency not stated | |
| 233 W average | 280 W caps for six hours, idle the other eighteennot stated |
the page labels this row (derived) in the cell itself: an estimate made in good faith from the caps, a ballpark, not a readingfrom The cost to purchase and run a very capable home AI rig § What the rig costs to runread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the same section · read from home-inference-and-diffusion-rig-cost · section id #what-the-rig-costs-to-run | |
| 91 W | each card at 280 Wmeasured, the day's median, 9 September 2026 |
a later reading of the same thing — see 93 W 2026-09-12 00:00 to 2026-09-15 00:00 UTCfrom The cost to purchase and run a very capable home AI rig § What the rig costs to runread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the same section · read from home-inference-and-diffusion-rig-cost · section id #what-the-rig-costs-to-run | |
| 660 W | both cards at their caps all daynot stated |
the page labels this row (derived) in the cell itself: an estimate made in good faith from the caps, a ballpark, not a readingfrom The cost to purchase and run a very capable home AI rig § What the rig costs to runread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the same section · read from home-inference-and-diffusion-rig-cost · section id #what-the-rig-costs-to-run | |
| not stated draw at the wall, re-read over three daysWeach card at 280 W · 2026-09-12 00:00 to 2026-09-15 00:00 UTC · quantisation, runtime, context, batch or steps and concurrency not stated | |
| 93 W | |
an earlier reading of the same thing — see 91 W measured, the day's median, 9 September 202691 W is 9 September's median; this is the median of each of the three clean days 2026-09-12 to 2026-09-14from The cost to purchase and run a very capable home AI rig § What the rig costs to runtyped in from that page by hand, then checked against it word for word · the card is named in the same section · read from home-inference-and-diffusion-rig-cost · section id #what-the-rig-costs-to-run | |
| not stated price, the two cardsquantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| $2,188.88 | August 2026's average completed sale |
the page prints this column as the one-card reading doubled, and reads the pair against the pair in its own prose: the cheapest new pair of any brand is 26 per cent dearer than this pair cost renewedfrom The cost to purchase and run a very capable home AI rig § The cards, in May and in Septemberread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named elsewhere on that page · read from home-inference-and-diffusion-rig-cost · section id #the-cards-in-may-and-in-september | |
| $2,899.98 | paid by this workshop, May 2026 |
the page prints this column as the one-card reading doubled, and reads the pair against the pair in its own prose: the cheapest new pair of any brand is 26 per cent dearer than this pair cost renewedfrom The cost to purchase and run a very capable home AI rig § The cards, in May and in Septemberread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named elsewhere on that page · read from home-inference-and-diffusion-rig-cost · section id #the-cards-in-may-and-in-september | |
| $3,659.98 | 13 September 2026 |
the page prints this column as the one-card reading doubled, and reads the pair against the pair in its own prose: the cheapest new pair of any brand is 26 per cent dearer than this pair cost renewedfrom The cost to purchase and run a very capable home AI rig § The cards, in May and in Septemberread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named elsewhere on that page · read from home-inference-and-diffusion-rig-cost · section id #the-cards-in-may-and-in-september | |
| $3,699.98 | 13 September 2026 |
the page prints this column as the one-card reading doubled, and reads the pair against the pair in its own prose: the cheapest new pair of any brand is 26 per cent dearer than this pair cost renewedfrom The cost to purchase and run a very capable home AI rig § The cards, in May and in Septemberread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named elsewhere on that page · read from home-inference-and-diffusion-rig-cost · section id #the-cards-in-may-and-in-september | |
| not stated the idle month at 18.34 cents a kilowatt-hour, re-readeach card at 280 W · 2026-09-12 00:00 to 2026-09-15 00:00 UTC · quantisation, runtime, context, batch or steps and concurrency not stated | |
| $12.10 to $12.28 | |
the same month the table prints at $12.10, moved by the three-day re-reading above; the page prints both ends of the move and this row carries its own charactersfrom The cost to purchase and run a very capable home AI rig § What the rig costs to runtyped in from that page by hand, then checked against it word for word · the card is named in the same section · read from home-inference-and-diffusion-rig-cost · section id #what-the-rig-costs-to-run | |
Everything else this shelf measured for it
1 row — same rule as above.
| value | what moves inside this group |
|---|---|
| not stated a day, and a month, at 18.34 ¢/kWheach card at 280 W · measured, the day's median, 9 September 2026 · quantisation, runtime, context, batch or steps and concurrency not stated | |
| 2.18 kWh · $0.40 | |
an addendum in this section re-read the same box over 2026-09-12 to 2026-09-14 and moved the idle month from $12.10 to $12.28; both readings are dated and both standfrom The cost to purchase and run a very capable home AI rig § What the rig costs to runread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the same section · read from home-inference-and-diffusion-rig-cost · section id #what-the-rig-costs-to-run | |
Where it appears
Every released exhibit that names this identity, in the shelf’s own order.
- The cost to purchase and run a very capable home AI rig — 10 figures on this page
- Six graphics cards, the same questions — named, and no figure carried from it — yet
How this one is named here
The canonical name above is the full product name. Everything below is a spelling a published page or kit has actually used for it, kept so a reader who searches an older page’s words lands on this row. The set is closed and historical: no page cut after 2026-09-17 writes a memory class where it can write a card.
- This box is never off
- Two EVGA GeForce RTX 3090 XC3 Ultras, renewed, 24 GB of graphics memory each, 48 GB between them, on a typical consumer board with two slots, desktop memory and a desktop power supply
- the darkroom's box