More likely its delayed by legal.
Originally Posted by pingufunkybeat
I don't think there is any unused PM hardware in pre-r600 chips. AFAIK everything was driver-controlled in those generations.
How do you determine that it always runs in "performance" mode and what do you mean by that exactly? Reading /sys/class/drm/card0/device/power_dpm_state will always report performance, and that is perfectly normal. Have you done actual power (and/or temperature) measurements? Even if DPM doesn't change frequencies and voltages in your case for some reason, clock gating should make a big difference.
Originally Posted by Vim_User
I don't think he cares that much about driver code, especially since this is adding new hardware support. As long as it comes in during the merge window, and not for rc6.
Originally Posted by madbiologist
The older cards had much less pm capabilities in their hardware, so while i'm sure some things can be done it is a lot less of an impact than with the newer cards.
Originally Posted by arianon
My first thought was about how Linux said 3.10 was the kernel with the most changes in any single release ever, and then how this ONE git pull alone will touch 52k lines. 3.10 may become the kernel with the SECOND most changes ever.
Originally Posted by smitty3268
Just a note, 52k lines of code change would only get you into 3rd place in the 3.9 kernel in terms of a companies contribution. Intel had over 75k. Amazing pace of developement...
AMD added 52k new lines of code, not changed 52k lines. It's more.
Originally Posted by set135
However you are right, intel changed a LOT on 3.9, especially when you look at 3rd place, Wind River, is an Intel subsidiary.
53,269 total lines in the original patchset, which would clearly be #3 in that 3.9 list.
Originally Posted by n3wu53r
Tags for this Thread