• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

The Great Gazelle PCI Hack Thread, Part 2

ried

Well-known member
@ried

That program in my last post should get you running!
How does one execute this on Mac OS 9? It expands as a document / text file, with an XtraStuf.mac folder beside it.

Edit: Trying MacZip.
Edit 2: Holy cow! It works! :cool:
 
Last edited:

Phipli

Well-known member
How does one execute this on Mac OS 9? It expands as a document / text file, with an XtraStuf.mac folder beside it.

Edit: Trying MacZip.
Yeah, sorry, I didn't have an appropriate version of Drop Stuff on the machine to .sit it and preserve the resource fork.
 

ried

Well-known member
Amazing work, @Phipli and @cheesestraws, @dosdude1, @joevt and everyone else who contributed. Thank you. Holy smokes 🔥

Here's the TAM's Apple System Profiler with the Sonnet Tempo Trio and devices attached to USB, FireWire and ATA-133. FireWire is still not represented in the System Profiler, as before:

Picture 1.PNG

I've personally been wondering how the internal ATA controller with its SSD would perform vs. an ATA-133 card with an SSD. Now we know:

Picture 2.PNG

Are you not entertained? :D

IMG_6135.jpeg
 
Last edited:

Byrd

Well-known member
A nice early Christmas gift - thanks all for putting your minds to this. Does this mean I can try my luck with other FW/USB combo cards (eg Keyspan, SiiG) as opposed to the oversized Tango 2.0 in my TAM that blocks airflow venting out the top? Great stuff.
 

dosdude1

Well-known member
A nice early Christmas gift - thanks all for putting your minds to this. Does this mean I can try my luck with other FW/USB combo cards (eg Keyspan, SiiG) as opposed to the oversized Tango 2.0 in my TAM that blocks airflow venting out the top? Great stuff.
Yes, this should rectify this issue with any PCI card that utilizes a PCI-PCI bridge.
 

joevt

Well-known member
I guess the next thing I'll do is update my list of nvramrc patches for supporting 512MB GPUs to include a patch for OF 2.0.3.
 

jeremywork

Well-known member
Many thanks to all who have their time in this! This is really brilliant work; I've been switching between trying to understand all of these leaps and actually testing it on my TAM throughout the day.

this should rectify this issue with any PCI card that utilizes a PCI-PCI bridge.
It's wonderful that this is the case. It was encouraging enough when cheesestraws was able to prove additional cases could be added via the old patch method, but obviously this was the dream solution, now realized :) Probably worth a "final" necro on this thread: https://68kmla.org/bb/index.php?threads/usb-firewire-cards-or-chipsets-that-work-in-the-tam.5760/ since it's still a top hit for people searching for why their card isn't working.

In my own case, a rough and dirty test of this works without a single flaw. I can boot from the 30GB ATA drive and press cancel on the dialogue to format my 500GB osx Firewire drive with my USB mouse. Excellent.
IMG_6683.jpg
If I switch them in place though, I get weird delays, conditional bus errors on boot, and no mountable volumes on the CF card. Assuming it's a simple issue between the CF/adapter and the Trio (removing the CF card from the ATA adapter resolves everyting, regardless of the 30GB being present on the onboard,) I presume a different ATA-CF solution (cable select mode?) might be the fix, but a different SSD will also work. Too bad, it was about to look pretty.

IMG_6690.jpg
I'll make it perfect laterish...
 
Last edited:

dosdude1

Well-known member
Many thanks to all who have their time in this! This is really brilliant work; I've been switching between trying to understand all of these leaps and actually testing it on my TAM throughout the day.


It's wonderful that this is the case. It was encouraging enough when cheesestraws was able to prove additional cases could be added via the old patch method, but obviously this was the dream solution, now realized :) Probably worth a "final" necro on this thread: https://68kmla.org/bb/index.php?threads/usb-firewire-cards-or-chipsets-that-work-in-the-tam.5760/ since it's still a top hit for people searching for why their card isn't working.

In my own case, a rough and dirty test of this works without a single flaw. I can boot from the 30GB ATA drive and press cancel on the dialogue to format my 500GB osx Firewire drive with my USB mouse. Excellent.
View attachment 50450
If I switch them in place though, I get weird delays, conditional bus errors on boot, and no mountable volumes on the CF card. Assuming it's a simple issue between the CF/adapter and the Trio (removing the CF card from the ATA adapter resolves everyting, regardless of the 30GB being present on the onboard,) I presume a different ATA-CF solution (cable select mode?) might be the fix, but a different SSD will also work. Too bad, it was about to look pretty.

View attachment 50451
I'll make it perfect laterish...
I’ve actually been working on my own custom 2.5” IDE SSD, of which the design and PCBs are already complete. All I need to do now is assemble the drives, which I’ll do in about a week or so after the holidays, and then I can start selling them. PCB design and schematic will also be available on my GitHub, if you want to make one yourself.

2E682A8E-29FE-477A-B822-A5B04796A139.jpeg

Alternatively, you may want to try re-programming that CF card, as I’ve demonstrated in this video of mine:


I don’t know if it will help with the issue you’re having, but is still probably worth a try.
 

joevt

Well-known member
Finally got to grab my PowerMac 6500, I have now gotten the information you have requested, using your LSPCI script. Output is attached.
I think I posted an old version of lspci for OpenFirmware. Can you try this newer version? It will probe the BARs to detect their sizes. I would like to know if the `assigned-addresses` and `reg` properties match the BAR sizes.
 

Attachments

  • lspci for Open Firmware Dec 25 2022.zip
    10.9 KB · Views: 2

dosdude1

Well-known member
I think I posted an old version of lspci for OpenFirmware. Can you try this newer version? It will probe the BARs to detect their sizes. I would like to know if the `assigned-addresses` and `reg` properties match the BAR sizes.
Done, got the output both with your NVRAMRC patch and without.
 

Attachments

  • lspci-nopatch.txt
    4.9 KB · Views: 2
  • lspci-patch.txt
    8.6 KB · Views: 2

cheesestraws

Well-known member
Can confirm your patch works with all the combo cards I have here. Good stuff!

Ethernet on the three-way card doesn't work, but it's detected, and I'm apt to put that down to Realtek weirdness rather than Gazelle weirdness.
 

Phipli

Well-known member
Can confirm your patch works with all the combo cards I have here. Good stuff!

Ethernet on the three-way card doesn't work, but it's detected, and I'm apt to put that down to Realtek weirdness rather than Gazelle weirdness.
The product / manufacturer id changes between different realtek cards with the same chipset (depending on the board manufacturer) - compare the two driver versions I sent you a while back and make changes at the points I did to match the ids of you card... might be all thats needed.
 

cheesestraws

Well-known member
The product / manufacturer id changes between different realtek cards with the same chipset (depending on the board manufacturer) - compare the two driver versions I sent you a while back and make changes at the points I did to match the ids of you card... might be all thats needed.

No, I did that: the interface was detected and came up but according to the switch was just sending gibberish: byte count kept going up, but 0 valid frames.
 

Phipli

Well-known member
No, I did that: the interface was detected and came up but according to the switch was just sending gibberish: byte count kept going up, but 0 valid frames.
Ah, curses.

Have you tried giving it a good talking to and explaining how it needs to make some serious life decisions?
 

ried

Well-known member
Because I like doing pointless things, I ordered two inexpensive 8GB disk-on-modules to slot into the Sonnet Tempo Trio's two ATA-133 connectors.

IMG_6354.jpeg

DOM1 and DOM2 each performed reasonably well at ~160% of the baseline Power Macintosh G3/300 in MacBench. Then I used SoftRAID to create a striped array and checked the resulting performance. Pretty decent performance increase for a software-only solution, totaling 230% of baseline.

SoftRAID.png

So this TAM has FireWire, USB and a RAID0 SSD (on the same PCI card) thanks to The Great Gazelle PCI Hack. :cool:
 
Top