Sonnet 7200 PCI programming

[grelbfarlk]

Originally Posted by BSzili

Never mind, I had some issue with the youtube video where the video and audio got desynchronized. I watched it again, and now it looks fine.
Too bad about the PSX controller support, back in the day more WarpOS titles supported it than 68k. My port had no vorbis support, if you wish to add it S_OpenAudio is the function where you could put it:
https://github.com/BSzili/NBlood-Ami...sound.cpp#L104

Is MP3 built in? I see it looking for .mp3s but it doesn't seem to want to load them.

[BSzili]

Version 1.1 and 1.2 had MP3 support. If it doesn't work the CLI output of the game and SnoopDOS would help a lot. Have you tried the MHI playback in AmigaAMP? What type of card are you using?
Before we turn this thread into an off-topic discussion, let's continue here
http://eab.abime.net/showthread.php?t=108291

[Hedeon]

Attached you can find a first beta for nBlood for WOS.

1) No MHI / MP3
2) Network is still sluggish on WOS native due to context switches (WIP).
3) Window mode is sluggish on WOS native due to context switches from CGX.

For music either use CAMD or the .WAV files as found on the site of BSzili

[grelbfarlk]

Seems to work fine. Tested joystick which is a PSX joystick with adapter in CD32 mode-works fine. 60-80 fps @ 800x600 with Voodoo3/Ragnarok.

[Hedeon]

Improved speed a bit on multiplayer on WOS.

Should be 60-100% faster on native WOS while in multiplayer.

[BSzili]

Nice! You managed to track down the context switch issues?

[Hedeon]

Originally Posted by BSzili

Nice! You managed to track down the context switch issues?

Actually no. I put some code in place to get network data in a 68k loop and while testing it with a single loop to see if the code was fine I found this acceleration. Then I saw that I also needed more code if I wanted more loops (regarding buffer copying) that for now I'm happy with the progress.

Not sure if it is playable on BPPC or CSPPC.

[BULI]

Originally Posted by Hedeon

Actually no. I put some code in place to get network data in a 68k loop and while testing it with a single loop to see if the code was fine I found this acceleration. Then I saw that I also needed more code if I wanted more loops (regarding buffer copying) that for now I'm happy with the progress.

Not sure if it is playable on BPPC or CSPPC.

Great Thanks
I tested yesterday the first beta and was very FAST on Ragnarok

[Hedeon]

Ok, I finally after some time (3 years?) have a working OS4 machine again.

nBlood WOS didn't work on it. So I updated ReWarp, and fixed the startup.o from WOS newlib and removed direct access to framebuffer when ReWarp is detected and now it seems to run on my OS4 machine.

CD music is disabled. You need the .WAVs as found on BSzili's site.
updated powerpc.library for OS4 (ReWarp) is included.

Let me know how it runs.

[dayowalker]

Does this mean OS4 does run on SonnetAmiga?

[Hedeon]

Originally Posted by dayowalker

Does this mean OS4 does run on SonnetAmiga?

No. This was an update to nBlood for WOS which should also work now on OS4. At least it did on my OS4 machine.

[dayowalker]

I see, sorry. :-)

[lukzer]

I'm having trouble starting a sound when using PPC in Prometheus / Firebird .. trying to run anything with sound by Paul freezes Amiga.
Is this normal behavior? do i need some sound card in pci?

[Hedeon]

Read the requirements please on Aminet and read the wiki regarding sound. Or wait a few days for me to get the time to reply in full.

[fazior]

Originally Posted by Hedeon

Actually no. I put some code in place to get network data in a 68k loop and while testing it with a single loop to see if the code was fine I found this acceleration. Then I saw that I also needed more code if I wanted more loops (regarding buffer copying) that for now I'm happy with the progress.

Not sure if it is playable on BPPC or CSPPC.

It works in 320X240 on BPPC 603e 330 Mhz. It shows 35 -40 FPS. On 640x480 is unplayable.

[jbenam]

Sorry if this isn't exactly on-topic, but I am not sure where else I could be getting the help of a group of Mediator users with known-working PPC cards.

I am still looking into my weird 8000 0004 issue, which now I know are caused by not being able to execute anything in a Zorro III address space.

I was looking at DiagROM and I noticed the following thing:

If I try to look at the Zorro III range address the Mediator normally allocates memory using the manual memoryedit function of DiagROM, I see the following pattern:

$6000000
60 FE FF FF 60 FE FF FF 60 FE FF FF 60 FE FF FF

$6100000
61 FE FF FF 61 FE FF FF 61 FE FF FF 61 FE FF FF

And so on. ($62 would have 62 FE FF FF, etc)

This is with my Buster 11 and no busboard inserted. I am kinda worried about that FE - I suspect it fills the first byte with the first two numbers of the address range as a kind of diagnostic test, and the rest should be filled with FFs - or at least so I suppose. If I insert my Mediator, everything gets initialised to FFs, and I am not sure that's correct either. I can't write at all to those memory addresses, but that might be the correct behaviour.

Any of you kind souls might check on your known-working A4000s if that's the case and see what normally gets allocated around $62D0000 with both the Mediator inserted and without?

Thanks in advance!

[Hedeon]

There should be nothing there except if inserted cards set memory or config up. And that is after pci library has been opened for the first time.

What you get back depends on how your mmu is set up (if it is).

What can be read is the PCI config space if cards are inserted. It's in the second mediator address space.

[neocdtv2]

Is there maybe a list of PMC-to-PCI Adapters/Carriers, which would work with a PRPMC800?
On the technobox page you can find 3 types:
https://www.technobox.com/adapters-carriers/4954
https://www.technobox.com/adapters-carriers/4933
https://www.technobox.com/adapters-carriers/4930

I guess the technobox cards differ mainly in speed?

Or the unknown.jpg, which im not really sure about the name and manufacturer.

[grelbfarlk]

For us the speed is irrelevant, the 64-bit cards will work as 32-bit.
The difference is mainly whether the card does 5V->3.3V conversion or not. Of which the first in your picture looks like it does, the other two do not.
I have tested the 2nd and 3rd but added a 3.3V VRM to them, they seemed to work ok.
Overall the Ramix PMC239/F seems to be the most stable, worry free adapter. However I could see situations where the Technobox 4930 would be preferable with the drive power molex connector on it.
That PRPMC800 looks strange can we get a picture of the other side?
It does not appear to be an Apocalypse as there is no L2 that I see.

[jbenam]

Originally Posted by Hedeon

There should be nothing there except if inserted cards set memory or config up. And that is after pci library has been opened for the first time.

What you get back depends on how your mmu is set up (if it is).

What can be read is the PCI config space if cards are inserted. It's in the second mediator address space.

Sorry for the late reply - apparently something sets up these addresses that way. Might be DiagROM itself, or maybe something in the Buster, like an internal self-diagnostic mode?

This is what I can see in DiagROM, so I don't think the MMU is playing any role in this.

Anyway, it seems like the second pair of bytes (FF XX FF FF) is "floating". If I keep refreshing, it sometimes spontaneously change the value to something else. When changing the value to something else (writing at these addresses using manual memory-edit) the other bytes seem "stuck" to FF (changing any value in the third or fourth pair) and won't change to anything, BUT the second pair of bytes changes very quickly to FF then sets back to FE.

That doesn't look right and might explain my ramlib issues when accessing Zorro III-mapped memory (which the Sonnet Library init apparently uses).

I haven't got a known working Amiga 4000D to compare it though - I was wondering if any of you could check on their usual PPC-PCI equipped machines how these values show up