The Dirty List Of GPUs With Open-Source Drivers Gone Wildly Wrong

[oibaf]

Don't know what this article was recommended in 2024...

It's its 10 years birthday! 🎂

[QwertyChouskie]

Don't know what this article was recommended in 2024, but admittedly it is cool to se just how far Linux GPU drivers have come in so many ways. Well, except for reclocking on pre-20series Nvidia GPUs, that's still an issue haha

[aht0]

I'd be interested to know if any of the failed AMD cards work better with FreeBSD 10's radeonkms.

Latest '11.2' has amdgpu driver, which should now light up those R9 cards as well. Might as well test it all the way

[Aeder]

Will this list ever get revisited in a new article?

[smitty3268]

So why doing it on his hardware then ? Not all people run Linux software on less-than-a-year hardware. The ultrabook may be very cool, but I'm not a rich guy that can afford change my hardware each year. Doing the same tests (multiple kernels with power monitoring) on an older hardware makes sense.

Hardware support for older machines was already mentionned as becoming sometimes problematic. It IS interesting to see the kernel comparison then.

I'm not saying whether it's interesting or not. That's a personal opinion. What i'm saying is that it's useless to you if you are trying to figure out what will happen on your machine. The only way to see that is to do it yourself and find out.

Laptops all have pretty specialized hardware, a comparison with more standard desktop hardware would be a lot more comparable.

[Morpheus]

So why doing it on his hardware then ? Not all people run Linux software on less-than-a-year hardware. The ultrabook may be very cool, but I'm not a rich guy that can afford change my hardware each year. Doing the same tests (multiple kernels with power monitoring) on an older hardware makes sense.

Hardware support for older machines was already mentionned as becoming sometimes problematic. It IS interesting to see the kernel comparison then.

[smitty3268]

When I ask Michael on Twitter about testing older hardware when comparing kernel releases (regarding power consuption), he responds that it's not interesting. I have a six years old laptop, I'm not sufficiently skilled to run the tests myself, but I want to know if newer kernel can help gain battery life. I'm not interesting, according to Michael. But I can't afford buy a 1000€+ laptop with top-notch components.

Power usage will depend heavily on what drivers you are using - what hardware your laptop has. I doubt any comparison michael did would mean much anyway, since it would be completely different hardware from what you are running.

[Morpheus]

When I ask Michael on Twitter about testing older hardware when comparing kernel releases (regarding power consuption), he responds that it's not interesting. I have a six years old laptop, I'm not sufficiently skilled to run the tests myself, but I want to know if newer kernel can help gain battery life. I'm not interesting, according to Michael. But I can't afford buy a 1000?+ laptop with top-notch components.

[stqn]

2005 called. They want their hardware back.

My father?s PC*contains a Radeon 9200 SE from 2004, and it is working fine (under Xubuntu 12.04 with a 1680?1050 screen). It can be useful to have recent information about old hardware (for example I?m trying to compare an Athlon 64 3000+ to recent processors, and it?s almost impossible to find any comparison).

[agd5f]

The firmware should fix it, but AFAIK, even with current kernel Git the R7 260X re-clocking support is still disabled by default.

It's enabled by default when the new firmware is loaded.