PCem

This must have been some sort of regression since the October commits between the 16th and Nov 1st.

It'll show all the resolutions it shouldn't do and crash the emulator when you choose one of them.

My BIOS is dumped from a 1mb card, so I doubt it's that...

this nasty surprise came after happening to run The Labyrinth which was set to primary display for a menu-testing reason, nearly reported this as a fatal V2 bug but turns out it's doing illegal accesses of memory it can't use for a double-buffered 640x480x16

It is the one included in Windows XP's Plus?

Pheraps you might take a look at the s3_init function to know that the culprit is there, especially the vram_max part.

Seems to work okay here. Which Trio64 driver are you using?

The stock S3 Trio64 driver with Windows 98SE and also the DirectX 5-provided S3 Trio64 driver (on Windows 95).

These drivers reported 1mb back in mid-October.


EDIT: Scitech Display Doctor also erroneously reports it as a 2MB card.

I have a suspicion as to what the problem may be. leilei, could you confirm if the board you dumped the BIOS from has FPM or EDO memory? A quick google of the RAM part number should say one way or another.

I _think_ that there's a difference in how memory is mapped between the FPM and EDO versions of the card. I confirmed in October that my FPM card wraps memory at 1MB if only 1MB is present, but the Phoenix BIOS that I have (which is definitely from an EDO card) expects it not to wrap.

It's a Phoenix-based BIOS and i'm fairly certain it's an EDO card IIRC.

EDIT: two chips of SM81C256KA1-40

Yep, that looks like EDO. I've committed my fix at rev 941.

Technically I should probably also limit this card to 2 MB, as the Trio64 only supports 4 MB with FPM. I'd probably get shouted at though.