93e55f9d43
docs/analysis/class-v6/hardware-future.md: one table over PIM and PNM (HBM-PIM, AiM, LPDDR5X-PIM, UPMEM), HBM3E, HBM4 and the custom HBM4E base die, fine-grained DRAM on logic, LPDDR6, 3D DRAM, RLDRAM 3 and Folded Banks, FPGA with HBM2e, wafer-scale, CXL, optical I/O, chiplets and UCIe, and SRAM at N2: energy per dependent read, latency, cost per GB, availability to a non-hyperscaler, the k_read and shadow k bands, the edge per joule at zero shadow against the 5090 and the M5 Max, and the one layer of the four that blunts each with a number. The three findings: a 2 GiB SRAM full store on one N2 reticle is the strongest five-year chip (13x to 17x per joule, about USD 0.3 per MH/s) and no layer reaches it (layer 2 moves its capex, not its joules; the shadow at k = 0.5 holds 4.8x); per-bank PIM is blind (1.6 percent of reads in-bank at 2 GiB) and the custom HBM4E base die is the f = 1 chip with its controller in the stack (6.5x to 14x), untouched by layers 1 to 4; the denominator is the wrong card (the M5 Max at 0.78 microjoules halves to thirds every chip edge). Section 5 carries the k bands for the research lane's chip rows; section 7 the four decisions with defaults and deadlines; section 9 the sources with URLs and dates. tools/ci/export-exclude.txt: docs/analysis/class-v6 is research, not public export. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
24 lines
2 KiB
Text
24 lines
2 KiB
Text
# Paths under the public export list that are NOT exported: research and operations documents written for the
|
|
# team, not for the public spec mirror (one path per line, relative to the repository root; a directory excludes its
|
|
# tree; # comments ignored). Read by tools/ci/identity-check.sh (pruned from the export copy before the grep) and by
|
|
# igneum-public/tools/sync.sh (pruned from the mirror after the copy), so the two can never disagree.
|
|
#
|
|
# Rule (CLAUDE.md, "CI red is stop-the-line", 6 October 2026): a research document that reviews internal operations or
|
|
# quotes the identity patterns as text lives here, outside the export list. A document that IS meant for the mirror is
|
|
# not listed here and must pass the identity grep like everything else the public reads.
|
|
#
|
|
# 6 October 2026, 21:2x UK: the Horizon lane's polish review (lane 6) quotes the overlay network's product name, the
|
|
# intake key variable name and the identity guard's own regexes as text; it blocked every master run from 20:23Z.
|
|
docs/analysis/horizon/polish.md
|
|
# the CI failure classification of 6 October 2026 (an operations document: run ids, step names, the fixes)
|
|
docs/analysis/ci-failures-2026-10-06.md
|
|
# 7 October 2026: the last research round (the mission lanes and the closed list): internal research written for the
|
|
# owner, quoting his words and the operations record; the public spec mirror carries none of it
|
|
docs/analysis/mission
|
|
# 7 October 2026: the in-house adversarial cryptanalysis pass (crypto-engage and the adv-* lanes): rule set, plans, reports and
|
|
# copied logs are research and operations documents, not public export; the public text is the served sentence main landed.
|
|
docs/analysis/cryptanalysis
|
|
docs/plans/cryptanalysis
|
|
# 8 October 2026: the class v6 research lanes (the hardware future, the rotating-family research): internal research written
|
|
# for the founder on the chip model, quoting the operations record; the public text is whatever main lands from the synthesis.
|
|
docs/analysis/class-v6
|