[fritsch]

Running one thread auf burnmmx in the background also "fixes" it? :-)

Btw. any upstream bugs we can reference here?

[AndyFurniss]

Ahh, I'll have to try burnmmx, not that I need to, but I am curious why for me doing eg 4x dd if=/dev/urandom of=/dev/null doesn't work but certainly spins well and gets my cores to full freq and hot.

I may later file a new bug against tonga and try and get some input as to what difference the devs can see between si and older (and I know rv770/rs880 don't have the issue in my box)

I am sure there are more kernel bugs around but can't see them currently - I guess you already know these -

https://bugzilla.kernel.org/show_bug.cgi?id=86351
https://bugs.freedesktop.org/show_bug.cgi?id=77677

[AndyFurniss]

Interesting burnMMX and burnK7 from cpuburn-1.4a don't help for me (phenom II x4 965be) even if I run 4.

This prompted me to retry p95(mprime) - I knew this worked with the default torture test settings (4 threads test 3) but this time I tried 1 thread and the other tests - and only 3 works (2 is better than 1) so it seems I need ram/memory controller load.

1 = Small FFTs (maximum heat and FPU stress, data fits in L2 cache, RAM
not tested much).
2 = In-place large FFTs (maximum power consumption, some RAM tested).
3 = Blend (tests some of everything, lots of RAM tested).

[AndyFurniss]

I filed another bug -

https://bugs.freedesktop.org/show_bug.cgi?id=91386