• Hello, Guest! Welcome back, and be sure to check out this post for more info about the recent service interruption and migration.

The Great Gazelle PCI Hack Thread, Part 2

cheesestraws

Well-known member
I'd be careful selling one of these machines with a combo card unless you know the person buying

Just leave the patcher application on the HD with a big obvious note on the desktop saying what to do, or something. So long as you're not booting off the combo card, that's fine: that's how my TAM came, for example.
 

macuserman

Well-known member
Just leave the patcher application on the HD with a big obvious note on the desktop saying what to do, or something. So long as you're not booting off the combo card, that's fine: that's how my TAM came, for example.
Yeah I mean that's kinda what I figured although I need to figure out how it works myself. But I'd personally rather have a 2.0 combo card in a machine with just one pci slot but that's just me even if it has to be patched if something like that changed. Can't please everyone of course, but that's just me.
 

Melkhior

Well-known member
For those tinkering with OpenFirmware, an entire OpenFirmware implementation is available on GitHub: https://github.com/MitchBradley/openfirmware/. Although probably not directly the one used by Apple, it has PPC code for the PREP specifications, I believe. Perhaps it could be useful if someone wants to figure out how to enumerate devices behind PCI bridges.
It can also be useful to 'detokenize' Forth code; Another command 'detok' is also available in e.g. the fcode-utils package in some Linux distribution. Those can help poke around a dump of the problematic Mac ROM, if that is available (such as figuring out if some words defined in Apple's ROM appear also in Firmworks' implementation).
 

cheesestraws

Well-known member
That could be useful. One of the major obstacles I'm running into, frankly, is my own competence: my Forth is not good at all, and especially reading other people's Forth is painful for me. So I'll have a look but it may give me a headache :-D.
 

Phipli

Well-known member
So first issue. I've found that open firmware on the 6500 and similar age machines doesn't read HFS or HFS+.

:/

Also, having issues with only the 255 characters being written into NVRAMRC.
 

pc297

Member
I confirm that this works with the Tango 2.0 PCI (latest version) on a PM 5500, both USB and firewire enabled! Many thanks @cheesestraws! Also just to confirm that with the OrangeLink PCI USB+FW, only USB shows up in System Profiler, outside of the device tree, just as with the Tango 2.0.

On a different note, I am still having issues with 3D acceleration when the card is enabled (onboard Rage IIc 2Mb), as with the OrangeLink card, I thought my OrangeLink card had some issues (a capacitor had come off and I soldered it back on) but then again it doesn't seem to be specific to it. The symptoms are random crashes whenever QuickDraw3D or OpenGL are invoked. This does not happen when the nvram patch isn't present (either the OrangeLink one or the current one). I have tried various revisions of the ATI extensions including the latest ones (as in as of 2002 or so) but to no avail. Any ideas? Anyone else experiencing this?

I am using USB 1.5.5 and FireWire 2.8.5, however not the Firewire Enabler that came with it (only 2.5 worked with the OrangeLink card). I will try the one from 2.8.5 with the Tango 2.0, and remove the OrangeLink Firewire extension.

Cheers and well done @cheesestraws !
 

Phipli

Well-known member
@pc297 If I'm not running a card like the 9200 that needs them, I find older ATi or the stock Apple drivers work more reliably in beige G3s with a Rage II. I've not got enough hours on my 6500 to say the same is true, but I'm just running it with the Apple supplied driver so far. Only 3D I've done was Strata 3D with opengl enabled, and it behaved.

Only other thing is, are you running out of VRAM? 2MB on these machines is useless. Most of it is used up by the framebuffer leaving little for 3D. I remember not being able to play (effectively) any 3D games on a G3 with only 2MB.
 

cheesestraws

Well-known member
Very happy that this worked in your 5500, thankyou for the report :).

Any ideas? Anyone else experiencing this?

That's bizarre. The patch doesn't do anything that should affect any other card at all, it just adds devices to the tree.

I'm afraid someone who knows more about 3D acceleration will have to assist you here, this is a long way beyond my knowledge. Might be worth posting a separate thread in the PCI subforum to see if anyone's had issues with those cards in general?
 

Phipli

Well-known member
I stepped through the pitcher program in debug and it freezes at line 98? nvpack();

Do I have something set up wrong regarding libraries perhaps? I anticipate operator error.

If this is inline with what I've seen before, it will have loaded the first 255 characters then stopped after writing "$c" from "$call..."
 

Attachments

  • IMG_20220115_110452_003.jpg
    IMG_20220115_110452_003.jpg
    3.4 MB · Views: 4

Phipli

Well-known member
Regarding my last post, I suspect operator error and that the patch I tried to load was too long. Please ignore.

Still not clear of the woods, but have some digging to do before I decide what is wrong.
 

pc297

Member
Might be worth posting a separate thread in the PCI subforum to see if anyone's had issues with those cards in general?

Yes you're absolutely right, sorry, I'll go ahead and do that to see if anyone else's ever experienced this. It might very well be the mobo itself.

Thanks again for the great work in any case! You should let Sonnet know about this, who knows :)

Cheers!
 
Top