Igneum 2.0 D4: the die's ticket reconciled (two machines: zero shadow at USD 0.6, the placed shadow core at 2.94; the adversary's at about USD 1.0 per MH/s, 0.5 to 1.6, set by the power train and the core); a sunk USD 10 M fleet about 10 TH/s; the die's economics are project economics

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

docs/design/class-v6-rotating-family.md

@@ -617,6 +617,8 @@ The placed 18-family core: 9.36 pJ per lane-op on the class v4 draw at ASAP7 (8.
 
 The five-year rows at USD 2.94 are the hybrid's within a few points (a USD 10 M fleet is 3.4 TH/s, not 16): private supply growing 16 / 14 / 8 percent of the chain with 13 to 16 classes above water, flat 25 / 33 / 56 percent, shrinking 20 / 38 / 100; hardware sales and multiple suppliers take flat and shrinking chains by year 5 as before. **What moved and why: the die's whole failure in the model's sections 2, 8 and 12 was its ticket; the joules never decided it (both lanes read 2.3x to 3.3x at the knee, inside (e)). Lane B priced the die's silicon (USD 0.25 to 0.4 per MH/s, 0.6 with the board) and the adversary lane priced a complete machine (2.94): a 5x gap carried as two rows with their sources, not resolved; it is the pin that decides the die's verdict, and the served reading until it is settled is the range: a sunk USD 10 M die fleet holds 16 to 70 percent of the chain on landing depending on whether its machine costs USD 3 or 0.6 per MH/s, and takes a flat or shrinking chain by year 5 at either. The conditions that fail the die at either ticket are (c) and (f).** **The pin, made (the coordinator, 17:38 UK, under the plan's evidence standard): the die's ticket is the adversary lane's complete machine, USD 2.94 per MH/s; the USD 0.6 row is kept as the labelled silicon floor.** Lane 3's document is re-read on it (class-v6-floor-sram 5b47fe65: section 0 carries the pin, section 12's condition table has one die column per ticket, the result paragraph and 12b read on the pinned ticket, the workbook row labelled), so the die's headline verdict is the hybrid's: coexistence under private supply in a growing chain (16 / 14 / 8 percent of the chain, 13 to 16 GPU classes above water), failure on flat or shrinking paths by year 5 and under hardware sales or multiple suppliers on those paths; the conditions that fail it are (c) (a third of the chain USD 9.4 M against USD 40 M a year at IGN 0.10, 4.3x short) and (f); (a), (b) and (e) pass at the pinned ticket and fail at the floor. The served form until the adversary lane's one-line reconciliation of what USD 2.94 includes (18:30 UK) is the range (a sunk USD 10 M die fleet holds 16 to 70 percent of the chain on landing depending on whether its machine costs USD 3 or 0.6 per MH/s, and takes a flat or shrinking chain by year 5 at either); lane 3 folds the reconciliation in within the hour and says whether the range narrows. The board's (c) tip at the final USD 4.84 (six of seven) stands.
 
+**The reconciliation (the adversary lane with lane B, 17:2x UK; its section 15 at d3ad15493; with lane 3 for the re-run):** the gap is two machines. Lane B's USD 0.6 per MH/s is the 2 GiB die at zero shadow (2,100 MH/s at 300 W, no shadow core); the USD 2.94 is the same die carrying the class v4 shadow on the placed core (the shadow 13x the die's energy per hash) held at 270 W, 383 MH/s. The adversary picks neither: it raises the power budget until the power train and the core silicon bind. The components, labelled: the die USD 500 (400 to 600, lane B, claimed); the shadow core silicon USD 0.55 per MH/s on the lane's 5-phase slot (placed, a measured floorplan) or 0.11 to 0.2 pipelined (approximate); the package USD 60 organic to 200 interposer (approximate); the board USD 150 plus PSU and cooling USD 0.15 per W (approximate); the host USD 15. The rate on the placed core 852 / 1,420 / 2,130 / 2,840 MH/s at 600 W / 1 kW / 1.5 kW / 2 kW; USD per MH/s 1.63 / 1.20 / 0.98 / 0.88 on the lane's core, 1.07 / 0.73 / 0.56 / 0.47 pipelined with the organic package. **RECONCILED: about USD 1.0 per MH/s (0.5 to 1.6) at 1 to 1.5 kW per machine, set by the power train and the core, not the die; per joule unchanged (2.3x same-node, 3.1x a node ahead).** For the coexistence verdict a sunk USD 10 M fleet at USD 1.0 is about 10 TH/s (5 to 20), between the two readings; and a maker's first batch of 1,000 dies (1 to 2 TH/s, several times the chain) pays the project, USD 50 to 250 per MH/s of NRE, not the unit ticket, plus first-yield losses of USD 100 to 400 per die (approximate): the die's economics are project economics, and the unit ticket of a fleet that exists is USD 1.0 (0.5 to 1.6). Lane 3's re-run at USD 1.0 (owed within the hour) sets the die's headline verdict; until then the range narrows to a sunk USD 10 M die fleet holding about a third of the chain on landing (a fifth to a half), and the conditions that fail it stay (c) and (f).
+
 
 #### 10.0d The rotation schedule the close adopts (the rotation lane, `docs/design/class-rotation-four-layers.md` on class-v6-rotation at bd43f808, build-3, gate green, 14:4x UK; one line per layer; both of this document's constraints held: the 180-day family epoch not shorter, W = 4 not drawn)