ATI Radeon 9250 PCI cards, converting from 64-bit to 128-bit and other details

[65a]

This is the 9200LE. I was pretty sure it was a dead card so was just hacking, will clean up the wire-wrap wire "straps" and the M25P10 soldering later. Pin 3&11 was done with a scalpel.

Flashed the retail 9200 v127 full rom from macelite, but ran the OS X updater which I think downgraded it to 124 (?). Both seem to work.

 

[synclavier]

I removed the M25P10VP, put it in the T48 programmer and it successfully read the flashed 128k ROM. I re-wrote the same ROM back to the chip successfully, so I know the chip itself is viable. I moved R219 to R220, so both R229 and R219 have been moved. The ROM still isn't loading on startup, although there are better signs this time around. I couldn't reset PRAM or shut the Mac off with the physical power button last time. This time, both of those functions are working. The drive activity light on the BlueSCSI v2 drive is blinking after startup and stops about 1-2 mins, suggesting that the system is loading. It just seems there is something keeping the Radeon's ROM from loading, but I don't know what it is.

@joevt I don't have another Mac with a serial connection right now. I so wish I did, though. I'd love to see what your T/S steps would reveal. I will look for a cheap one. I think I would be work the cost to learn how to work like that.

I found a reference 9250 on Ebay and pulled the trigger. This video card upgrade/mod is a fun aside, but I need to keep this project on track. If I can't get this particular card to work, I plan to shelve it, mod the reference 9250 card, and carry on.

 

[65a]

Small update: I spoke too soon, I am getting minor issues with the LE. Tiger goes into a loop trying to set the resolution during loginwindow (but OS 9 and 10.5 work...). OS 9 works, but exiting quake does the same thing as tiger's login window. I reflashed v127 (I had used the OS X ATI updater from tiger, so this worked at some point). Weirdly VNC screen is also black. Leopard and OS9 much better behaved. It might be a cap, and this is a "farm-fresh" MDD PSU, so many variables here.