New P-State Patches Could Boost Intel Graphics Performance Under Some Conditions

[phoronix]

Phoronix: New P-State Patches Could Boost Intel Graphics Performance Under Some Conditions

Francisco Jerez of Intel's open-source Linux graphics driver team has posted not some new DRM driver patches today but rather CPUFreq/P-State driver patches that could really help Intel integrated graphics performance under some conditions and especially on the lower-power Intel platforms...

http://www.phoronix.com/scan.php?page=news_item&px=P-State-Better-Graphics

[overwatch]

BTW, is CPUFreq still recommended over P-State (Clear Linux, etc) or is it the other way around?

[Michael]

BTW, is CPUFreq still recommended over P-State (Clear Linux, etc) or is it the other way around?

Clear switched to P-State months ago; defaulting to P-State performance governor. Generally with recent kernels, P-State should be at least as good as CPUFreq.

[aufkrawall]

intel_pstate powersave leads to severe stuttering on www.vsynctester.com here with Skylake. Quite pathetic.

[schmidtbag]

I have a feeling this isn't going to have much impact in benchmarks. Maybe a few graphically intensive games, but those games don't play well on Intel hardware anyway.

[Vistaus]

intel_pstate powersave leads to severe stuttering on www.vsynctester.com here with Skylake. Quite pathetic.

You mean the VSYNC logo on the right? Yes, that one stutters as well for me intel_pstate performance even. Celeron 7310 here.

[aufkrawall]

You mean the VSYNC logo on the right? Yes, that one stutters as well for me intel_pstate performance even. Celeron 7310 here.

Yes, and accordingly, also the animation of the background shows heavy stuttering. It's smooth for me with schedutil.
That happens also with maxed out GPU clocks (Polaris), it's entirely related to CPU clock governor.

[Linuxxx]

Clear switched to P-State months ago; defaulting to P-State performance governor. Generally with recent kernels, P-State should be at least as good as CPUFreq.

I'd like to disagree.
As others have already noted, Intel's very own P-State driver simply sucks when it comes to good latency-performance, no matter the setting!

But You know what, it would be simple for You to test this with a few latency-oriented PTS cases (like frame-latency tests)...

[linuxgeex]

My C720-2640 Chromebook is absolutely flawless in this test, even when I drag the window around (other than the unavoidable fringe at the difference areas of the VSYNC text.)

My IVB laptop with xfwm4, stutters with desktop compositing and VSync enabled. Disable compositing or disable VSync and the stutters are much more rare.

$ sudo apt-get install  i7z msr-tools
$ echo turbo disable | i7z_rw_registers

That makes the VSYNC test perfectly stable in Google Chrome with xfwm4's compositing enabled. But no turbo boost.

Interestingly this does reduce framedrops in KODI with 1080p h264 videos using VAAPI, and my fan runs a little quieter, but not quiet enough to want to use KODI as a full-time media player.