Fresh Linux 4.16 Kernel Benchmarks With KPTI & Retpolines

[phoronix]

Phoronix: Fresh Linux 4.16 Kernel Benchmarks With KPTI & Retpolines

For those wondering how the Linux 4.16 kernel is performing with regards to the ongoing work around Spectre and Meltdown mitigation in the kernel, here are some fresh benchmark results.

http://www.phoronix.com/vr.php?view=26071

[PLum]

So, if I see this correctly - 4.14 was much better kernel then recent ones ? why ... and who can fix it ?

[ryszardzonk]

I wonder if kernel 4.14.0 performance equals that of 4.14.24. After all those repotoline/ktpi patches have been backported to the 4.14 branch.

[cybertraveler]

So, if I see this correctly - 4.14 was much better kernel then recent ones ? why ... and who can fix it ?

Or no?

(I'm not very familiar with kernel development... I just assume it looks something like this)

[dungeon]

Yeah, it would be nice to compare let say all on january 5. - kernels 4.4.110, 4.9.75 and 4.14.12.against current 4.4.120, 4.9.86 and 4.14.24.

That would draw more correct picture what happened in last 2 months with all these mitigations

[darkbasic]

So, if I see this correctly - 4.14 was much better kernel then recent ones ? why ... and who can fix it ?

4.14 didn't have any spectre/meltdown fix at all.

[dungeon]

4.14 didn't have any spectre/meltdown fix at all.

All current kernels on kernel.org have meltdown mitigation, spectres only 4.4+.

But yeah, 4.14.0 does not have it - that zero is from november

[darkbasic]

All current kernels on kernel.org have meltdown mitigation, spectres only 4.4+.

But yeah, 4.14.0 does not have it - that zero is from november

The benchmark makes use of 4.14.0 indeed.

[eddy]

I think what this shows is there is a massive regression from 4.14.0 to 4.15.0.
Could there have been a mistake? I mean maybe the disable repoline and KPTI flag does fully disable it?

[dungeon]

The benchmark makes use of 4.14.0 indeed.

Yep, only benchmark uses that and no one else, so it represents - nothing