The Visual6502 project reverse-engineered the MOS 6502 by photographing its silicon and simulating every transistor. Their website has been quietly falling apart. This is a mirror, with attribution, so that it does not go.
| Part of the site | State | Recovered here |
|---|---|---|
| The wiki — 128+ pages of research | HTTP 500 on every page | Rebuilt from wikitext recovered out of the Internet Archive |
| Die photography — 516 scans, 2.3 GB | Served, but nothing links to it | Mirrored and made browsable for the first time in years |
| The JavaScript simulator and docs | Still working | Mirrored as it stands |
The wiki returns 500, not 404 — MediaWiki is failing rather than the content being deleted, which suggests the pages are still on disk behind a broken database. The die photographs are stranger still: all 2.3 GB are served correctly right now, but every page that linked to them is dead and the directory listings return 403, so nothing on the open web points at them. Intact and unreachable at once — which is usually how this material disappears. Not deleted, just orphaned.
A decade of research on the 6502 — decimal mode, interrupt behaviour, timing states, the decode ROM — plus the project's index of chips in their collection. Rebuilt as static pages from the original wikitext, so it cannot rot the way the original did.
Read the wikiTwenty-times microscope scans of the silicon: the 6502 and 6800, the 68000, Atari's POKEY and GTIA, the NES PPU, and more. For several of these parts these are the only public images that exist.
Browse the collectionvisual6502.org as it still stands, including the JavaScript simulator that started all of this and the SIGGRAPH 2010 talk describing how the chip was photographed and vectorised.
Open the mirrorNone of this is our work. The Visual6502 project is Greg James, Barry Silverman and Brian Silverman, with the Visual6502 team and its contributors. The die images, polygon data and wiki content are theirs, licensed CC BY-NC-SA 3.0; that licence permits this mirror and carries the same NonCommercial and ShareAlike terms onward to anyone who takes it further.
Every page here links back to its original URL and to the Internet Archive snapshot it came from. Where something could not be recovered, it is marked as missing rather than quietly omitted — an archive that hides its gaps is worth less than one that shows them.
visual6502.org · the project blog · source data on GitHub
Nothing in the wiki is known to be lost. Two filenames looked at first like the one real gap — description pages for what appeared to be scans of hand-drawn 6502 schematic sheets, with no image behind them at any resolution. They turned out to be red links: the archived pages read “No file by this name exists”, and the Internet Archive holds no record of either name at any date. Nothing was ever uploaded under them.
The hand-drawn schematics themselves — dated November 1974 and August 1975, from before the 6502 shipped — survive under different filenames and are in this archive at 2593×873 and 2597×877, each with a lower-resolution variant.
Six wiki pages exist only as renderings rather than as source, and 39
File: description pages could not be recovered as source because the
wiki refused anonymous edits, so the Internet Archive captured a
permission-denied form instead of the editor. Those are served as archived HTML.
Everything else is rebuilt from the original wikitext.