DOOM - Unofficial CD32 Release - FastRAM Required!

[SKOLMAN_MWS]

@turrican3

Akiko performs only graphics conversions from Chunky to Planar, he is not a 3D graphics accelerator. 3D graphics render CPU.

I'm not an coder either.

[Marchie]

Pardon the poor fool lost in all these Doom versions but...

Which version will run best on a TF328 and where can I get it?

[trixster]

Probably DoomAttack which is available on Aminet. You can freely find online the shareware Doom WAD or you can get other commercial wads from places like gog.

[emiespo]

Originally Posted by Marchie

Pardon the poor fool lost in all these Doom versions but...

Which version will run best on a TF328 and where can I get it?

As already said, DoomAttack with the AKIKO2 c2p routine selected and in single buffer mode.

On a 020@14Mhz Akiko makes quite a difference, as the C2P takes a considerable amount of the processing power available. On the TF330 (v1 firmware) Akiko seems to still be just slightly better than the most optimized C2P.

[emiespo]

Originally Posted by turrican3

Hi guys,
could it be possible to make some parts of the 3d calculate by the main cpu and others parts by akiko ???
I'm not a coder don't kick me too hard !

It would be nice if the 68030 of the TF330 could work in parallel with the 68020 on the CD32 board so that the 68020 could take care of the screen refresh + Akiko C2P, while the 68030 does the "heavy math". But I think this would require another OS, to start with (and is possibly not even supported by the CPU bus).

[wairnair]

Here are my findings with TF330 using different c2p routines running DoomAttack.
This is after installing the 680x0 libs http://aminet.net/package/util/sys/Mu680x0Libs.

c2p: can't initialize c2p, starts in RTG mode, giving 6-8 fps and very little graphics corruption
c2p_020_optimized: flickers badly, otherwise 10-12 fps. This is the speed I get under WinUAE with a very similar setup. Too bad it flickers.
cp2_020_blitter: starts and immediately crashes with graphic corruption
c2p_akiko: first "software failure" then black screen
c2p_akiko2: this one is actually identical to c2p byte-for-byte so naturally it does the same thing.

looks like I can't play Doom for now

I also tried c2p_020_optimized and turning off data cache. I wrote a startup script that contains this:

CPU INSTCACHE INSTBURST NODATACACHE DATABURST
stack 20000

cd da08beta4
assign doomhome: ""
doomattack -displayid "0x21000" -nodoublebuffer -joypad -noinputhack -fps -nomusic

Is this how the data cache is supposed to be turned off?
Any help appreciated.

[Marchie]

Originally Posted by trixster

Probably DoomAttack which is available on Aminet. You can freely find online the shareware Doom WAD or you can get other commercial wads from places like gog.

Thanks!

[wairnair]

wads:

https://github.com/dgeng/iwad

[wairnair]

Originally Posted by wairnair

CPU INSTCACHE INSTBURST NODATACACHE DATABURST

Is this how the data cache is supposed to be turned off?

Can anyone please advise?

[SKOLMAN_MWS]

cpu nodatacache

or check:

cpu nocache + cpu instcache burst

http://www.jaruzel.com/amiga/amiga-o...-help/cpu.html

[wairnair]

thanks. no luck with ALL caches disabled either. I tried with and without 680x0 libs too. I double checked with sysinfo and AIBB that all caches were off.
I'll just let akiko c2p go, I know there's almost nothing in it.

what's interesting is I get graphics artifacts no matter which c2p routine I use.. ehh.

Originally Posted by SKOLMAN_MWS

cpu nodatacache

or check:

cpu nocache + cpu instcache burst

http://www.jaruzel.com/amiga/amiga-o...-help/cpu.html

[emiespo]

Originally Posted by wairnair

thanks. no luck with ALL caches disabled either. I tried with and without 680x0 libs too. I double checked with sysinfo and AIBB that all caches were off.
I'll just let akiko c2p go, I know there's almost nothing in it.

what's interesting is I get graphics artifacts no matter which c2p routine I use.. ehh.

Let me recap: I had the same issues with the TF330 and the latest V2 firmware (you see a V2 if you go to the Amiga early boot menu). For some reason Akiko stops working or is not detected.

With the V1 firmware (the slightly slower one) Akiko works, including all softwares that make use of it.

You only need to install the MMU libraries (that simply make Akiko accesses non-cacheable) or "cpu nodatacache", but, again this only works with the TF330 V1 firmware.

I am waiting for TF to do some tests with these games, it will take some time, but I am pretty sure he'll find a fix for this weird issue. Akiko *does work* as the CD drive works with both firmwares for me, it's only the C2P bit that is not working.

Anyway, the Akiko C2P is either slightly faster (DoomAttack) or noticeably slower (Wing Commander) than the CPU only C2P...

As for your flickering issue, you should run DoomAttack in double-buffer, and che C2P_020_optimized will give good results with the TF330 V2.

EDIT: just remove the -nodoublebuffer option from DoomAttack command line. That option only really works with Akiko, and gives a small speed increase (~0.5 FPS for me).

[emiespo]

Hi guys, I'm attaching an archive with two files:

1) akikofix: this is a very simple executable that will force the Akiko pointer in the graphics.library. It's intended for TF330 boards with the V2 firmware, on my setup it makes Akiko function again

2) c2p_akiko_fixed: this is the OS friendly C2P routine (c2p_akiko) for DoomAttack. The original has a pointer bug that makes it crash. The only difference with the other routine (c2p_akiko2) is that the former gets exclusive access to Akiko before using it (it will actually lock access to the Blitter, but Commodore was running out of time and money and they couldn't think of anything better than that).

Let me know if it works for you. Akikofix is kinda beta, it will check for the Akiko ID code before doing anything, so it should be safe to run, but it also won't output any message.

PS: bear in mind that 1 this will only make Akiko detectable again by the software, to see a working C2P conversion on a 030, you still need the mmu.library *or* to turn off the data cache.

`cpu nodatacache`

[wairnair]

very interesting, I'll check it out when I get home next weekend

[wairnair]

Here are some results, tested on TF330 with v2 firmware.

akikofix DOES enable detection of akiko c2p, so that's great.
akikofix was in my user-startup throughout testing.

Gloom Deluxe: c2p set to akiko.
CPU NODATACACHE: works, but it's noticeably slower than software c2p with DATACACHE ON.
CPU DATACACHE enabled: totally garbled screen, seems pretty fast though

DOOM: c2p set to c2p_akiko_fixed
CPU NODATACACHE: works, but it starts in "RTG" mode and it's slow (~7fps vs. ~10-11fps expected)
CPU DATACACHE enabled: same thing. it seems akiko c2p doesn't work at all regardless of DATACACHE (I guess RTG mode already tells that akiko c2p doesn't work)

Wing Commander: garbled screen regardless of DATACACHE when using akiko c2p.

I don't have mmu.library + 68030.library working yet. Copying them to LIBS: is not enough, SetPatch needs to be patched with Mu680x0Libs's installer but I was just afraid to run it without a backup. And too lazy to make a backup yet.

Originally Posted by emiespo

Hi guys, I'm attaching an archive with two files:

1) akikofix: this is a very simple executable that will force the Akiko pointer in the graphics.library. It's intended for TF330 boards with the V2 firmware, on my setup it makes Akiko function again

2) c2p_akiko_fixed: this is the OS friendly C2P routine (c2p_akiko) for DoomAttack. The original has a pointer bug that makes it crash. The only difference with the other routine (c2p_akiko2) is that the former gets exclusive access to Akiko before using it (it will actually lock access to the Blitter, but Commodore was running out of time and money and they couldn't think of anything better than that).

Let me know if it works for you. Akikofix is kinda beta, it will check for the Akiko ID code before doing anything, so it should be safe to run, but it also won't output any message.

PS: bear in mind that 1 this will only make Akiko detectable again by the software, to see a working C2P conversion on a 030, you still need the mmu.library *or* to turn off the data cache.

`cpu nodatacache`

[trixster]

Akikofix kinda seems to work for me too, using v2 firmware and testing on DoomAttack.

I’m not using cpu nodatacache, I’ve installed mmulibs.

So even with akikofix loaded akiko_fixed and akiko don’t appear to enable c2p, they both still say operating in RTG single buffered mode.

Akiko2 however does seem to work with akikofix loaded. c2p in double buffered mode now works (before akikofix it reported RTG single buffered) and it gives an extra couple of fps over pre-akikofix RTG single buffered.

In fact, akiko2 with akikofix now seems as fast (maybe a tiny bit faster) than 020 blitter c2p which was the fastest c2p beforehand.

[wairnair]

great
btw, what type of 030 do you have? full or EC030?

Originally Posted by trixster

Akikofix kinda seems to work for me too, using v2 firmware and testing on DoomAttack.

I’m not using cpu nodatacache, I’ve installed mmulibs.

So even with akikofix loaded akiko_fixed and akiko don’t appear to enable c2p, they both still say operating in RTG single buffered mode.

Akiko2 however does seem to work with akikofix loaded. c2p in double buffered mode now works (before akikofix it reported RTG single buffered) and it gives an extra couple of fps over pre-akikofix RTG single buffered.

In fact, akiko2 with akikofix now seems as fast (maybe a tiny bit faster) than 020 blitter c2p which was the fastest c2p beforehand.

[trixster]

EC, so doubtful that mmu.library is doing anything. Should maybe have just installed the 68030.library and 680x0.library

[wairnair]

akiko2 c2p + akikofix + CPU NODATACACHE works indeed (still have no mmu libs, but I might try)!

however, I found a very strange error: once I save the game, it stutters for ~2 secs and freezes completely.
Then after pressing the reset button the TF330 is no longer there together, gone completely. The CD32 chime and boot screen comes up!
Very strange.

[emiespo]

Originally Posted by trixster

EC, so doubtful that mmu.library is doing anything. Should maybe have just installed the 68030.library and 680x0.library

I'm glad to have helped with this. I also have a 68EC030, but the MMU library (and table) still works for me. Without the mmu.library, I'd have to completely disable the cpu datacache, thus losing a bit of performance.

The archive is quite up to date (a few days ago) and it comes with an installer that *will do a backup copy of everything* including setpatch.

Interestingly enough, with highly optimized C2P routines (or maybe simpler games ), Akiko seems slower. Given the silly way it works, I guess that anything above a 68030 would be faster as well.

The original "c2p_akiko" found in DoomAttack has a (bad) bug, so it's pretty normal if the CD32 crashes with it. I've added a fixed version in the Aminet archive, you should copy that in the DoomAttack/C2P folder and delete the old one.

The "c2p_akiko2" doesn't have the bug, and also works in single-buffer (thus it is slightly faster than 020+Blitter, that only works in double-buffer mode).

All in all, I'd say that getting Akiko back is also nice if you use the Workbench (anything writing to the screen with WriteChunkyPixel() will be much faster, unless you also patch that with one of the many patches on Aminet ).

I would be interested to see if a full 68030 gives better results with the MMU driver on Fusion, but I haven't seen anyone with a TF330 with a full 030