Hello,

Happy new year to all!

I have found a graphic bug with a Radeon 9800 in 16bit resolution on my G5.

Any idea ?

https://www.meta-morphos.org/download/images/bug16bit.jpg

Thanks

[ Edité par Papiosaur 05.01.2020 - 11:15 ]

Had the same effect with MorphOS 3.11 and a flashed Radeon 9800, but it seems gone with MorphOS 3.12.

MorphOS 3.12 here :(

same as problem here with ATI 9800 flashed in PowerMac G5 and MorphOS 3.12 :( (if I remember correctly there was not this problem only in MorphOS 3.9 or 3.8)

boot with ISO, change resolution to 16bit in it, and find the first distro version that has this bug, probably the problem is from 3.9 or 3.10

[ Edited by Divinity 05.01.2020 - 22:09 ]

any news to fix this problem ?

Hello Papiosaur,

In 207 I wrote an article for the french fanzine amigaPower and published one year later on Obligement here, http://obligement.free.fr/articles/powermacg5_decouverte.php . At the chapter 8 I speak about the problems I have encoutered with the graphic card ATI Radeon 9800. I do not any more know whether it was a pc card reflashed to work in powermac g5 or a genuine mac edition of this card. The problems I encoutered suggested that this grapic card did not like displaying ambient screen or nay other program in 16 bits mode. Display was corrupted, as in negative colors, with or without 'improved display' checked. And I do not tell you about games non sdl or sdl opening in full screen in 16 bits mode... Absolutly corrupted.

Is this card from the non corectly working powermac g5 that I have sold to you in 2018 at the µalchimie or an other one ?

Brumiga

I got similar issues on Pegasos 1 with MorpOS 3.06 and Voodoo 3 3000 PCI 16 bit 8 bit too and usually it fails to change screen resolution differently from 1024x768, ever over that res. Tried it on 3 different LCD moniors.

Then I got black screen on a new ATI 9250 128 MB PCI! But this video card can run fine on a PC Windows.

My Mini G4 1,5 GHz looks working at 16 bit.

@ Brumiga

Did you submit a bug report about it?

@ Goodman

Quote:
Due to performance reasons, it is not advisable to use resolutions higher than 1024x768 with an old Voodoo 3 card. Apart from the severe lack of video memory, the lack of proper digital display ports is another issue that comes into play as you move towards higher resolutions.

Quote:
As jPV has already explained, this is because of a bug in MorphOS 3.12, which is not present in older versions. In general, the Radeon will be a much better graphics card and is a decent choice especially due to the included 128MB of video memory.

@Asiegel,

I have not thought to submit a bug report about this problem. And now I do not have any other powermac g5 a1047 or a1177.

When I have sent my article to Obligement Daff told me that the second or third powermac g5 he bought had the same graphic card and that he encoutered the same display problems with it. He finally changed it for an ATI Radeon 9600 or 9650 mac edition. Then all the problems were gone.

Brumiga

Quote:

According to AmbientMui Voodoo 3 could handle FullHD but then it rarely switched below 1024x768 res.

I opened a topic cause I'm not able to install 3.12 from 3.06. Now I know there's a bug on 3.12 but I guess even before.

Quote:
Was it completely black from the start? I mean, did you even see the open firmware texts at boot or was it just MorphOS (logo and so on) which was black? Did the monitor give any signs of life (leds indicating it gets some kind of signal etc)?

Quote:
Although you have to remember that Voodoo 3 doesn't support higher color modes than 16bit, so it doesn't consume that much video memory. I've been running Voodoo 3 in 1600x1200x16bit resolution on my A1200/Mediator just fine... that is of course 2D accelerated and I haven't tried how enhanced modes on MorphOS would act or consume mem, but if running non-enhanced modes, it should be fine I think.

> I know there's a bug on 3.12 but I guess even before.

This bug was not in 3.11 according to that thread.

Quote:

I had similar problems with a Powerbook in the past (before MOS3.10). I replaced the Radeon-Driver (Mossys:devs/Monitors/Radeon) with a version from a earlier distribution (from where I know it works). In my case it works, but I'm not sure if it does it in your special case (R9800 flashed) and under MOS_3.12, because it seems that there are some deeper changes in driver-system since MOS_3.9 (-> e.g. see new drawer Mossys:Devs/Firmware/Radeon).
I did a short test on my G5 (installed MOS_3.12/Radeon 9600)) with Radeon-driver 50.30 (from MOS_3.7) and for the first look it runs properly. Just try and test it. If it fails you can replace it back by booting from 3.12-CD.

> it seems that there are some deeper changes in driver-system since
> MOS_3.9 (-> e.g. see new drawer Mossys:Devs/Firmware/Radeon).

I think this drawer was introduced in MorphOS 3.8 for AtomBIOS (Radeon R5xx) compatibility, so shouldn't affect any earlier Radeon GPUs.

Quote:

So black I thought I had a faulty video card: no Open Firmware and no signal from 3 LCD monitors.

tested with Morphos 3.13 and same as problem with ATI 9800 PC flashed, in all 16bit modes (no problem in 24bit modes)

if someone need help to debug, i have a PMAC G5 + ATI 9800 ready
It is only a problem in 16bit modes (24bit modes are fine)

Quote:

We've tried but could not repro this on a 9800 in a Peg2 (so obviously with original PC firmware). I'm afraid there isn't much to be done here without physical access to the card.

Quote:

if I can be useful with ROM 9800 file flashed used or other, please tell me
At the moment I'll use the Mac original 9600 card

was this bug fixed in MorphOS 3.14 ?

sorry, was this bug fixed? I no longer tried to replace the video card in MorphOS 3.15 (and 3.14)

hello, any news to solve this annoying problem?

@Divinity

You were given a definitive answer two years ago.

Repeating the same question after every OS release or two does not change that this particular issue cannot be fixed without physical access to the problematic hardware component.

What else is there to say? It is nice to offer to do remote testing but for some issues this is just not good enough as it is impractical/too time-consuming to figure out the problem this way.

As has been mentioned both in this thread and also elsewhere, we can't possibly support reflashed Radeons. If they work for you, then great. If not, well, then it's really too bad.

However there is one thing you can do. I of course can't promise that it'll lead to anything at all, but if you or anyone else who has a card affected by this problem can open up the GraphicsBoards utility and take a screenshot of it (while using the affected card, of course), then maybe that could give me a hint.