Sonnet 7200 PCI programming

[Hedeon]

Jbenam: Hold on... You said it didn't work with the old 17.13 version.

[Hedeon]

Grelbfarlk: also the same question: can you place the cards that the voodoo is first regarding addresses?

[jbenam]

Originally Posted by Hedeon

Jbenam: Hold on... You said it didn't work with the old 17.13 version.

Yep, indeed. That's the case. I just get freezing and/or ramlib issues after the RAM from the M1 gets added with 17.13.

I was referring to the fact that grelbfarlk's config (which I assume was working with 17.13) gets the same error as mine does with 17.14 beta. Which apparently means that something broke when it comes to M1/K1 support. At least that's my assumption.

Should I keep trying with 17.14 at this point, or should I just go back to 17.13 while I test other power supplies?

[Hedeon]

17.13 for now. I have to find the error in 17.14.

[Hedeon]

Please give me the output using the following attached library.

Thanks.

[grelbfarlk]

Originally Posted by Hedeon

Please give me the output using the following attached library.

Thanks.

Seems to work chief!

Sashimi installed ([Ctrl]+C or "Break 8" to remove)
Started Library Init routine
Version: $VER: powerpc.library 17.14 beta (23.03.21)

Detected supported Elbox Mediator PCI bridge
Opening of support libraries completed
3DFX card detected, Gfx address at 64000000, config address at 62000000
Size of Gfx card in PCI memory is 03FFFFF0
Calculated from FE000008
Detected Killer M1/K1 PPC PCI card
Setting up 64MB of PPC memory at 68000000
Setting up PPC card config base at 60400000
Setting up VGA PCI window at C4000000 located from 64000000 with size 800FE000
Setting up VGA config block for 3DFX Voodoo located at 62000000
Waiting on PPC card to respond
Accessible PPC memory set up with header at 68410000
Inserted PPC memory into Exec memory list
Library base set up at 6BFF6CB0
Copied PPC kernel code into place
Gort 68K interrupt now running at 093A1B8C
MasterControl now running at 09436A08
Completed adding various patches to Exec library
Zen support interrupt now running at 68410024
Started Kryten PPC cleanup task
Opened ppc.library emulator for PowerUP support
Sashimi removed.

[Hedeon]

But does VoxelSpace and Warp3D work?

[grelbfarlk]

Warp3D yes, I ran a GLQuake timedemo, 37FPS@640x480. Voxelspace works. Tested a couple other things like ScummVM2, Dgen, Prboom@640x480 is only getting 15.1FPS which maybe that's right for K1. Ragnarok gets about 59FPS. I guess it's right, Adoom@640x480 gets 19.22FPS.

[Hedeon]

OK thanks. Good to hear.

[grelbfarlk]

The 600MHz Rapture is working now too with this version. Prboom@640x480 30.1FPS. GLQuake 640x480 43FPS. Quake2 demo1 640x480 58.4FPS. Adoom 640x480 33.42 FPS.
Asm version not working, tested back to build 381. Just says could not allocate DMA memory space machines.

[BULI]

Originally Posted by jbenam

..
VoodooMem 13
..

BTW
If you aren't using Spider Buffer, set this to:
VoodooMem 15


More VRAM more fun

[jbenam]

Originally Posted by Hedeon

OK thanks. Good to hear.

Thanks for the fix. I've finally been able to test it.

The new library allocates memory then it burns down in a sea of errors. Sadly I can't save anything since the system completely freezes.

I get a very large dump in MuForce around the "Accessible PPC memory set up with header at 68410000". Then another one at the end (around "Zen support...").

I think I am gonna get another voltage regulator - I trust my A4000's power supply (5V is literally a perfect 5.0V, even when under load) but I am not sure I can say the same of the power regulator I am using (AIC 1084), which was salvaged from an Ethernet PCI card I had laying around.

I have also tried with an ATX power supply, things didn't seem to improve. I still want to test with a new power regulator (this time a LM1085 as in the wiki) just to be sure.

[grelbfarlk]

Originally Posted by jbenam

Thanks for the fix. I've finally been able to test it.

The new library allocates memory then it burns down in a sea of errors. Sadly I can't save anything since the system completely freezes.

I get a very large dump in MuForce around the "Accessible PPC memory set up with header at 68410000". Then another one at the end (around "Zen support...").

I think I am gonna get another voltage regulator - I trust my A4000's power supply (5V is literally a perfect 5.0V, even when under load) but I am not sure I can say the same of the power regulator I am using (AIC 1084), which was salvaged from an Ethernet PCI card I had laying around.

I have also tried with an ATX power supply, things didn't seem to improve. I still want to test with a new power regulator (this time a LM1085 as in the wiki) just to be sure.

Eh that LM1085 was just the first one I found on an old SCSI card or something. An LT1085 should be fine too or whatever.

[jbenam]

Originally Posted by grelbfarlk

Eh that LM1085 was just the first one I found on an old SCSI card or something. An LT1085 should be fine too or whatever.

In the meanwhile I have purchased some of those cheap adjustable buck converters. Should be here tomorrow.

The AIC1084 literally gets super hot which definitely doesn't make me comfortable (it even partially melted one of my hook probes!), so I thought I could test one of these nifty thingmabobs. I could also crank it up a bit and see if it helps... Will report back tomorrow

[grelbfarlk]

Oh good, we haven't had a good fire around here in ages!

[Hedeon]

Originally Posted by jbenam

Thanks for the fix. I've finally been able to test it.

The new library allocates memory then it burns down in a sea of errors. Sadly I can't save anything since the system completely freezes.

I get a very large dump in MuForce around the "Accessible PPC memory set up with header at 68410000". Then another one at the end (around "Zen support...").

Scrolled a bit back and in September last year you had the jumper on the Mediator differently. With Memory space at 0x40000000-0x60000000 and now it seems it is at 0x60000000-0x80000000. I hope you updated your MMU config accordingly.

EDIT: Nevermind. I see an updated MuScan from last week... Still, MuForce tells me something is wrong M68K side.

[jbenam]

Originally Posted by Hedeon

EDIT: Nevermind. I see an updated MuScan from last week... Still, MuForce tells me something is wrong M68K side.

What does that imply? OS install issue? Faulty 040?

I have an A3640 right now. I've also got a 040-to-060 adapter I could build, if needed, and test everything with a 060.

EDIT: The el cheapo buck regulators I bought were a massive failure as it wasn't humanly possible to set them to 3.3V. I ordered another kind. Meanwhile I am testing the AIC1084 and it actually produces a very stable 3.3V. I dunno, maybe it isn't a power issue after all...

EDIT2: I have tested with another 1084 I had around (it came with the 040-to-060 kit) and same issues. I think we can safely exclude the matter of power. I will now build the adapter and try with a 060...

[grelbfarlk]

PCI termination or slot order then? Have you tried an 8029 or 8139 NIC in the top slot?

[jbenam]

Forgot to say I have tested with three different 040s, so it's not a CPU issue.

Originally Posted by grelbfarlk

PCI termination or slot order then? Have you tried an 8029 or 8139 NIC in the top slot?

Pretty sure I did in the past, but right now I am trying with just the M1 and the Voodoo 3. I'll try adding it to the mix.

[trixster]

Ive used an A3640 accelerator in both mediator and Prometheus and it works fine with Ragnarok and apocalypse, so the 040 might not be the issue