Liquorix 3.11 Kernel Benchmarks
Phoronix: Liquorix 3.11 Kernel Benchmarks
It's been several kernel releases since last benchmarking the Liquorix kernel, an optimized version of the Linux kernel that's advertised as "built using the best configuration and kernel sources for desktop, multimedia, and gaming workloads." In Liquorix having out their version of the Linux 3.11 kernel since late September, here are some benchmarks comparing Liquorix to recent mainline versions of the vanilla Linux kernel.
I thought that we already went through this? That having some kid compiling a kernel and applying his interwEbz haxx0r name to it, does not make it anything special.... Its the linux kernel, compiled by... some douchebag, that nobody knows or cares about.
We saw many 3.11-3.12git-related tests this week. None of them are done with power usage metrics comparison in parallel with these modes.
We definitely need it.
Facepalm- please measure LATENCY for Liquorix
I thought we already went through this. Having Liquorix benchmarks without measuring LATENCY is quite meaningless.
Regarding Liquorix being useful or not, it has Debian packages and usually I get new version of Linux kernel from Liquorix sooner than it gets into Debian Sid as "official" Debian kernel. So for me it's useful if for no other reason than getting last Linux kernel into my Debian Sid system sooner. I haven't seen it break anything on my dev machine yet, so it is stable enough for desktop usage at least.
If it gives more latency than stock kernel or not I cannot tell, as I haven't done any measurements and I wouldn't know how- measuring latency is HARD.
Yea, really, since PTS can now show frame latency, why aren't there any graphs of it? That's supposed to be the entire point of such kernels.
Because all the distros have exactly the same kernel with the exactly same patches and tunables, right? Oh wait, what is a distro exactly?
Originally Posted by droidhacker
Anyway, it's good that you picked up 3.11-3.dmz.2 that fixes some power states issues on Intel CPUs: http://techpatterns.com/forums/about2325.html (
I hope Valve does a better job with SteamOS...
different use case - interesting!
Some professionals really should!
Originally Posted by tmpdir
Liquorix had significantly performed better on tests not related to graphics!
This shows once more there are different use cases which cannot be pleased by one kernel only !!!
Yes, what I tell everyone is that Liquorix should perform worse where throughput is concerned. Sometimes Liquorix wins in throughput because a patch was backported from the LKML to improve a corner case. It's interesting that Liquorix performs worse in games. But also note that many of those games are CPU bound because the frame rate is so insanely high. You'll get more realistic and visual results when your framerate drops below your refresh rate which is 60 in most desktop configurations.
Originally Posted by ulenrich