Linux 3.18 Should Let Radeon GPUs Clock Higher For OC'ed Cards

[CrystalGamma]

Hint: if things gone dire and you can't see anything, logs still could contain valuable data. So if you can reboot and boot at least somehow (maybe resorting to VGA mode temporarily, etc) - you can examine logs to see if there're something useful. Most notably, kernel log.

I would try that, but until 19th December, all my AMD hardware unavailable to me (living somewhere else until then).
So if the next kernel doesn't fix the card, I can look into it.
BTW, do issues in the DRM part of the driver still belong on freedesktop.org or only the userspace parts?

[Melcar]

Hmm I never noticed this - I have my GPUs manually overclocked via BIOS changes, but I never bothered to confirm in linux if the clock settings are the same. It'll be interesting to see if this affects me.

Same here. The HD7850 I'm running now had it's BIOS tweaked to non reference 1GHz core and 1.4GHz memory. I did notice a while back that radeon started to report the clocks as 1GHz core and 1.2GHz memory.

[edoantonioco]

so, radeon.dpm=1 in /etc/environment I guess is not needed anymore.

[GraysonPeddie]

re-arrange some audio code to properly reflect connected status in the audio driver

What could cause Radeon HDMI audio to not reflect connection status properly? With my AMD A10-5700 APU, I am using a built-in HDMI audio with JACK as a frontend with a buffer size of 1024 samples, sample rate of 48kHz, and periods per buffer of 2 in KXStudio's Cadence. I do get 8 channels of playback, but I'm only using 5.1 surround sound.

[Azpegath]

Have you heard any news regarding the dynamic overclocking of the 290? We've seen in your previous benchmark that there seems to be some issues with it as it is.