Features Still Being Sought By Open-Source AMD Users

[duby229]

No. If the hardware part didn't change between r100 and r200 there's no need to write new code for it (that would just be duplicated code and as such a maintenance burden). AFAIK the r200 driver alteady maxes out the hardware (more than the latest catalyst version supporting that old hardware did), so there's nothing to improve anymore.

I still remember when r200 was the best open source driver around... and r300 was lucky to have modesetting...

EDIT: That was back when I had my heavily modified 9500Pro... It died long before the r300 drivers matured....

[smitty3268]

In fact when browsing through the radeon-code in the kernel, much of the r200 code simply calls r100 code. That's why it was - and still is - my impression that there might be some room for imrovement. As for the suggestion to buy new hardware, i wasn't aware that the spirit of Steve Jobs haunts the Phoronix forums. I'll give the latest kernel a try and see what happens.

The r100 and r200 drivers are hardly tested at all, so i wouldn't be surprised if you run into bugs with them. But they have pretty much maxed out what the hardware is capable of. You're talking about hardware that isn't even capable of shaders, so nothing that has been done in the last 10 years or so really even remotely applies to them.

[klapauzius]

The r100 and r200 drivers are hardly tested at all, so i wouldn't be surprised if you run into bugs with them. But they have pretty much maxed out what the hardware is capable of. You're talking about hardware that isn't even capable of shaders, so nothing that has been done in the last 10 years or so really even remotely applies to them.

A driver that makes the whole system freeze hard when using AGP mode isn't exactly what I'd call "maxed out"

[vitalogy]

Phoronix: Features Still Being Sought By Open-Source AMD Users

UVD with RS880

[smitty3268]

A driver that makes the whole system freeze hard when using AGP mode isn't exactly what I'd call "maxed out"

Most people think features and performance when you say "maxed out", not bugs. Unfortunately, that hardware is so old none of the developers even have it anymore, so whenever they have to port Mesa changes, they just have to try to do their best and wait to see if anyone reports problems.

[klapauzius]

The r100 and r200 drivers are hardly tested at all, so i wouldn't be surprised if you run into bugs with them. But they have pretty much maxed out what the hardware is capable of. You're talking about hardware that isn't even capable of shaders, so nothing that has been done in the last 10 years or so really even remotely applies to them.

By the way, the hardware does support Shaders (and some other nice features). See

http://en.wikipedia.org/wiki/R200#Radeon_9200

But that's okay. I'm a QA guy by profession, so i'm more than used to developers telling me they're done, finished and their code "maxes out".

[GreatEmerald]

I'm pretty sure those are assembly shaders, not OpenGL shaders.

[curaga]

Just in case you didn't know, AGP is a horrible buggy mess, with a ton of boards requiring quirks. As a QA person you should hate to test such a system, with such a number of combinations: practically all AGP cards in all AGP chipsets, otherwise you'll miss some quirky combination.

[klapauzius]

Just in case you didn't know, AGP is a horrible buggy mess, with a ton of boards requiring quirks. As a QA person you should hate to test such a system, with such a number of combinations: practically all AGP cards in all AGP chipsets, otherwise you'll miss some quirky combination.

Yeah, i know. But as i pointed out in an above post, the card worked fine in AGP mode with UMS drivers. So the only "advantage" that KMS has for such old systems is that the driver running at ring0 can freeze the whole system. Ah well.

[agd5f]

Yeah, i know. But as i pointed out in an above post, the card worked fine in AGP mode with UMS drivers. So the only "advantage" that KMS has for such old systems is that the driver running at ring0 can freeze the whole system. Ah well.

The driver ran in ring0 with UMS as well. Also, the UMS driver had no memory manager and didn't really utilize AGP memory for much at all so it's not surprise that with certain AGP/GPU combinations you have hangs when to start to full utilize AGP memory. Have you tried adjusting the AGP mode, etc.? A lot of AGP/GPU combinations only worked in specific AGP modes (even with UMS).