Cant get working Kaveri APU - A10-7850k

[Dekoski]

It was a HW fault

Hi again,

For the records, I have received a new A10-7850K CPU and now I'm working fine with the RadeonSI drivers. So it was a HW fault.

Thanks to all for your help.
?scar.

[nanonyme]

Typical Linux arrogance: the "if it doesn't happen on my machine it doesn't exist and everybody else are liars" attitude.

http://www.reddit.com/r/linux/comments/26vtpv/im_about_done_with_amd_gpus_but_am_i_really/chvmqm9

https://ask.fedoraproject.org/en/question/37783/solved-installed-fedora-20-on-new-computer-and-it-is-very-slow-need-help/

Toodles.

That's not a bug report platform either. Just so this will be clear: Fedora bugs are tracked on the Fedora section of Red Hat Bugzilla.

[dungeon]

If you ask me that A4-5000 is pretty interesting - looks to me like hardware problem , because it is only one that has revision A4-5100 (and also one E1-2200) just 50MHz more according to the specs, but GPU stay the same 497 MHz...

And later as a desktop part called Athlon 5150 it is transformed to 25W device, +50MHz but now full 600MHz for the GPU... so if you ask me, that really smells like some sort of hardware design problem on first revision mobile part A4-5000, so some kind of performance suffering seems expected there

[mmstick]

Not all hardware is equal, and from what I've read, AMD hardware is particularly diverse...
https://bugs.freedesktop.org/show_bug.cgi?id=60389#c64

That doesn't apply to the A4-5000 which is a quad core 1.5 GHz mobile SoC using Jaguar cores. Everyone gets the same system-on-a-chip. Implying that the A4-5000 is broken on Linux is false -- it is one of the best supported models on Linux.

[nanonyme]

Hi again,

For the records, I have received a new A10-7850K CPU and now I'm working fine with the RadeonSI drivers. So it was a HW fault.

Thanks to all for your help.
?scar.

Honestly that makes me somewhat terrified. That's the very chip I was planning to buy and not fun to hear such heterogenic variants have been shipped

[bridgman]

Honestly that makes me somewhat terrified. That's the very chip I was planning to buy and not fun to hear such heterogenic variants have been shipped

Sounds like a defect, not a variant.

[nanonyme]

Sounds like a defect, not a variant.

I was meaning it can possibly mean a batch of defective chips are out there. I do trust AMD but many retailers these days do their very best to try to nullify warranty and I personally aren't capable of proving something is a defect

[bridgman]

I was meaning it can possibly mean a batch of defective chips are out there. I do trust AMD but many retailers these days do their very best to try to nullify warranty and I personally aren't capable of proving something is a defect

OK, I guess.... (although the expression on my face is saying "sure, but aliens could have drilled a teeny tiny hole in the APU lid and sucked out all the graphics life force too").

That said, "heterogenic" seemed to suggest design variants not defects. For what it's worth "heterogenic" probably applies better to the way we *design* products (mixing IP blocks from different per-block generations based on what is available & stable at the time) than to the manufacturing process (where "obsessively monoculture" is probably a better fit).

[asdfblah]

That doesn't apply to the A4-5000 which is a quad core 1.5 GHz mobile SoC using Jaguar cores. Everyone gets the same system-on-a-chip. Implying that the A4-5000 is broken on Linux is false -- it is one of the best supported models on Linux.

I was giving a generic example on how things aren't always equal, and that you may have a different experience with hardware than that of other people.