Arch Linux User Repository

MuQSS 0.150 reverted the Hz back to 100Hz and I think that's the recommended value from ck. @QuartzDragon, r08 do you change both the below: CONFIG_HZ_100_MUQSS=y CONFIG_HZ=100

@Quartz and r08 - Post to ck's blog?

In the past, I too had noticed weird stuttering using 100Hz, so now I always set it to 1000Hz. System is smooth. :)

Interesting how I noticed slow downs using the default MuQSS@100Hz while using Firefox and running Steam games like Team Fortress 2. Setting it to 1000Hz fixed the latency issues I was having. Should 1000Hz tick be the recommended setting for a "normal desktop user?" Running old Intel Core 2 T9500@2.60Ghz

Anyone else getting mce errors at boot, occasionally, on Haswell? I only get them when I hard reboot if the computer stalls after displaying time till next fsck.

Have you read it? https://wiki.archlinux.org/index.php/Makepkg#Creating_optimized_packages

@j1simon: I think SuperIce97 meant that if you compare output for 'native' for both Skylake and Kaby Lake then you don't see any difference. I have two hosts one with Skylake and other with Kaby Lake. I compared they 'native' output and the difference: $ diff -y --suppress-common-lines native-sl.out native-kl.out -march= skylake | -march= broadwell -mtune= skylake | -mtune= generic For example, this is difference between Sandy Bridge and Skylake: $ diff -y --suppress-common-lines native-sb.out native-sl.out -mabm [disabled] | -mabm [enabled] -madx [disabled] | -madx [enabled] -march= sandybridge | -march= skylake -mavx2 [disabled] | -mavx2 [enabled] -mbmi [disabled] | -mbmi [enabled] -mbmi2 [disabled] | -mbmi2 [enabled] -mclflushopt [disabled] | -mclflushopt [enabled] -mf16c [disabled] | -mf16c [enabled] -mfma [disabled] | -mfma [enabled] -mfsgsbase [disabled] | -mfsgsbase [enabled] -mlzcnt [disabled] | -mlzcnt [enabled] -mmovbe [disabled] | -mmovbe [enabled] -mprfchw [disabled] | -mprfchw [enabled] -mrdrnd [disabled] | -mrdrnd [enabled] -mrdseed [disabled] | -mrdseed [enabled] -msgx [disabled] | -msgx [enabled] -mtune= sandybridge | -mtune= skylake -mxsavec [disabled] | -mxsavec [enabled] -mxsaves [disabled] | -mxsaves [enabled] Sandy Bridge: Intel(R) Core(TM) i5-2410M CPU @ 2.30GHz Skylake: Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz Kaby Lake: Intel(R) Core(TM) i5-7400 CPU @ 3.00GHz @SuperIce97: at first I was confused why 'native' reports that arch is 'broadwell' and even thought that GCC don't support this new processor. But if you are right about behavior of GCC when one passes 'native' vs 'NAME' then it is related only to 'generic' build for Kaby Lake and that processor is fully supported by GCC.

@j1simon: I believe this is because gcc optimizes in a way to make sure that all CPUs from that architecture will run correctly, meaning that if 1 CPU is missing an instruction, it will not be included in the optimization. This leads to differences as some CPUs in the same architecture supports newer instructions. Here is the diff between the broadwell arch optimizations and native for an Intel 3205u Celeron Broadwell CPU: diff -y --suppress-common-lines broadwell.out native.out -mabm [disabled] | -mabm [enabled] -mcx16 [disabled] | -mcx16 [enabled] -mfsgsbase [disabled] | -mfsgsbase [enabled] -mfxsr [disabled] | -mfxsr [enabled] -mlzcnt [disabled] | -mlzcnt [enabled] -mmmx [disabled] | -mmmx [enabled] -mmovbe [disabled] | -mmovbe [enabled] -mno-sse4 [enabled] | -mno-sse4 [disabled] -mpclmul [disabled] | -mpclmul [enabled] -mpopcnt [disabled] | -mpopcnt [enabled] -mprfchw [disabled] | -mprfchw [enabled] -mrdrnd [disabled] | -mrdrnd [enabled] -mrdseed [disabled] | -mrdseed [enabled] -msahf [disabled] | -msahf [enabled] -msse [disabled] | -msse [enabled] -msse2 [disabled] | -msse2 [enabled] -msse3 [disabled] | -msse3 [enabled] -msse4 [disabled] | -msse4 [enabled] -msse4.1 [disabled] | -msse4.1 [enabled] -msse4.2 [disabled] | -msse4.2 [enabled] -mssse3 [disabled] | -mssse3 [enabled] As you can see, on my Broadwell CPU, broadwell optimization != native optimizations. Also notice how "native" only enables optimizations instead of disabling. This is why I usually recommend -march=native, as you will be able to optimize your CPU as much as possible. For Kaby Lake vs Skylake, here is the Wikipedia page listing the differences (which are all clock speed/process, socket/chipset, and GPU changes, not microarchitecture changes): https://en.wikipedia.org/wiki/Kaby_Lake#Architecture_changes_compared_to_Skylake

@SuperIce97 I don't sure of that. My CPU: Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz I execute: $ gcc -march=skylake -mtune=skylake -Q --help=target > skylake.out and: $ gcc -march=native -mtune=native -Q --help=target > native.out and these are the differences: $ diff -y --suppress-common-lines skylake.out native.out -mabm [disabled] | -mabm [enabled] -madx [disabled] | -madx [enabled] -maes [disabled] | -maes [enabled] -march= skylake | -march= broadwell -mavx [disabled] | -mavx [enabled] -mavx2 [disabled] | -mavx2 [enabled] -mbmi [disabled] | -mbmi [enabled] -mbmi2 [disabled] | -mbmi2 [enabled] -mclflushopt [disabled] | -mclflushopt [enabled] -mcx16 [disabled] | -mcx16 [enabled] -mf16c [disabled] | -mf16c [enabled] -mfma [disabled] | -mfma [enabled] -mfsgsbase [disabled] | -mfsgsbase [enabled] -mfxsr [disabled] | -mfxsr [enabled] -mhle [disabled] | -mhle [enabled] -mlzcnt [disabled] | -mlzcnt [enabled] -mmmx [disabled] | -mmmx [enabled] -mmovbe [disabled] | -mmovbe [enabled] -mno-sse4 [enabled] | -mno-sse4 [disabled] -mpclmul [disabled] | -mpclmul [enabled] -mpopcnt [disabled] | -mpopcnt [enabled] -mprfchw [disabled] | -mprfchw [enabled] -mrdrnd [disabled] | -mrdrnd [enabled] -mrdseed [disabled] | -mrdseed [enabled] -mrtm [disabled] | -mrtm [enabled] -msahf [disabled] | -msahf [enabled] -msgx [disabled] | -msgx [enabled] -msse [disabled] | -msse [enabled] -msse2 [disabled] | -msse2 [enabled] -msse3 [disabled] | -msse3 [enabled] -msse4 [disabled] | -msse4 [enabled] -msse4.1 [disabled] | -msse4.1 [enabled] -msse4.2 [disabled] | -msse4.2 [enabled] -mssse3 [disabled] | -mssse3 [enabled] -mtune= skylake | -mtune= generic -mxsave [disabled] | -mxsave [enabled] -mxsavec [disabled] | -mxsavec [enabled] -mxsaveopt [disabled] | -mxsaveopt [enabled] -mxsaves [disabled] | -mxsaves [enabled] I think it isn't the same.

@zhenyu: 1. gpg --receive-keys 79BE3E4300411886 38DBBDC86092693E 2. repo-ck.com, not ck-repo.com.