Does anyone have a 667MHz or 800MHz Powerbook G4?

[obsolete]

Specifically, I'm looking for the "DVI" model, A1001 (EMC 1913), M8591LL/A or M8592LL/A. These have the ATI Mobility Radeon 7500 GPU.

I happened across a PCI card with this chip on it, and I'd like to see whether I can flash it from PC to Mac, so I'm looking for the video ROM from the Powerbook. It doesn't seem to have been archived anywhere. If you have one of these TiBooks and would be willing to help, please let me know.

Here's what it looks like in System Profiler (stolen from a random eBay auction):

 

[treellama]

I’ve got one (667 dvi), but I’m traveling this weekend. I can help next week though!

Would be neat if you could flash the 7500. Were there any 7500 Mobility AGP cards? Could be a low heat upgrade to the Cube.

 

[obsolete]

Thank you for offering to help! Unfortunately, it's looking trickier than I hoped...

I just checked the readme for the ATI ROM dumping and flashing tools from Graphiccelerator, and it says: "Note that on-motherboard graphics cards (iMacs, eMacs, PowerBooks, iBooks) are not supported by the dumper tool."

That's probably why no one's dumped this ROM yet...well, that and the fact that a mobile chipset isn't really supposed to be on a PCI card, so what use would it have been? As far as I've gathered from a couple of Vogons threads, these 7500 Mobility PCI cards are some low-budget Chinese boards, probably someone trying to use up leftover chips long after they were relevant. I have no idea whether any others are out there; it certainly wasn't advertised as "Mobility" when I bought it!

@joevt, do you have a method for extracting the ATI ROM from the Powerbook?

 

[joevt]

Thank you for offering to help! Unfortunately, it's looking trickier than I hoped...

I just checked the readme for the ATI ROM dumping and flashing tools from Graphiccelerator, and it says: "Note that on-motherboard graphics cards (iMacs, eMacs, PowerBooks, iBooks) are not supported by the dumper tool."

That's probably why no one's dumped this ROM yet...well, that and the fact that a mobile chipset isn't really supposed to be on a PCI card, so what use would it have been? As far as I've gathered from a couple of Vogons threads, these 7500 Mobility PCI cards are some low-budget Chinese boards, probably someone trying to use up leftover chips long after they were relevant. I have no idea whether any others are out there; it certainly wasn't advertised as "Mobility" when I bought it!

@joevt, do you have a method for extracting the ATI ROM from the Powerbook?

If the device's Open Firmware code is not in a PCI option ROM, then it is in the ROM of the Mac. What version of Open Firmware does the Mac have? I have a few ROMs with the ATY,RageM7 fcode, but not the one that has revision 113-XXXXX-119.

Use "get-new-world-rom.command" to capture a copy of the PowerBook's ROM, zip the result, post it here, and I'll have a look.

Question: How powerful of a graphics card will work in a beige Power Macintosh G3?

I have made a note. You can use the attached "get-new-world-rom.command" in Mac OS X to dump the firmware into a file named "macrom#.bin" (where # is a unique number), zip the result, and post it here so I can examine it for more information.

forums.macrumors.com

The first post has a list of Open Firmware versions that I know about. I would like to get a ROM dump for any Mac/Open Firmware combo that is not in that list or that is in that list marked as "don't got it".

Creating a PCI Option ROM from the fcode image in the captured Mac ROM is easy, but it might not work.
I'll grab all the ATY,RageM7 revisions then we can compare and choose one to make a PCI Option ROM.

Do you have a copy of the ROM from the PC PCI card? Is there a utility to extract the timings from that ROM?

 

[obsolete]

Thanks @joevt, I should've figured you would have a couple Powerbook ROMs stashed already
I don't have any particular affinity for version 113-XXXXX-119 of the ATY,RageM7 fcode, it's just the one I happened to find a System Profiler screenshot of. It would be interesting to know whether @treellama has it or another unknown version, though.

I've attached the PC ROM from the 7500 Mobility PCI card as well as log output from RaBiT v1.6 and v2.2.1 that shows all the known timing strings and their offsets in the PC ROM. The two versions disagree a bit on the parsing of the timings, but agree on their contents and locations. I remember using the output from RaBiT v1.6 to make Radeon 9500 and 9800SE ROMs from the 9800 Pro v130 ROM back in the Strangedogs days.

 

[joevt]

I created a 32K ROM in the Test1 folder in the attached zip. The ROM is using version 113-XXXXX-210 of the Rage Mobility Mac firmware from Open Firmware 4.6.4f1 iBook G3. It doesn't contain a Mac OS or Mac OS X driver. Those should be included in the OS.

I guess the first step is to see if it can be flashed.
Then after that, see if the card works in Open Firmware.

I give it a 10% chance of working.

 

[obsolete]

I tried the PCI card with your Test1 firmware in my G4. It would not boot to OF or Mac OS with the card installed; just a bong and a black screen forever. I had the monitor connected to a different, known working video card, and the system works fine with the PCI card pulled. I put the PC ROM back on, and here's how it shows up in OF:


Here's how it shows up with the EEPROM removed:

 

[joevt]

This is going to be difficult to debug without a built-in serial port.

I modified the Forth code to include bytes 0x60 to 0x7F from the PC ROM. These bytes include the subsystem vendor id and device id (0082:0515). I don't know what the other bytes are for.

I also added a property "start" that gets set at the beginning of the code. Not very helpful if you can't get into Open Firmware though.

 

[obsolete]

Thanks. I wish I had a serial port adapter for the G4. I also have a couple of PCI old world machines, a 7600 and a 5400, along with a PC that could be pressed into service. I'll give your new ROM a try later today.

 

[obsolete]

Well, "later today" came a few days late, but the results of test2 are the same as test1. Thank you for your time and effort nonetheless, @joevt.

Would an nvramrc script to start telnet in open firmware be sufficient, or does it need to be a serial connection to debug an issue early in boot like this?

 

[joevt]

I don't know if telnet can be used before probe-all? probe-all is when the fcode of the graphics cards are loaded.

Do an nvram script to see what's available before probe-all? But where would the output go? Maybe it's buffered and it will appear when the console is installed and opens the working graphics card.

dev / ls
unselect-dev
." [ probe-all" cr
probe-all
." ] probe-all" cr
install-console
banner

Is telnet listed? Is an ethernet device listed?

For debugging, you could add to the beginning of the fcode of the card a function to read an nvram variable (from the options package). The value of the nvram variable can then be used to decide when the fcode should leave early (using end0).
Keep increase the value of the nvram variable until OF stops working. This will narrow down where the problem occurs in the fcode.