You are not logged in.

#1 2018-01-07 16:21:14

ugjka
Member
From: Latvia
Registered: 2014-04-01
Posts: 1,358

Performance degradation after Meltdown mitigation?

What is your experience?


ENTER

~ Arch is not Arch unless it is Arch ~

Offline

#2 2018-01-07 16:32:24

Slithery
Forum Moderator
From: Norfolk, UK
Registered: 2013-12-01
Posts: 3,399

Re: Performance degradation after Meltdown mitigation?

I haven't noticed any, but I don't do anything that stresses the CPU. Phoronix has run a lot of detailed benchmarks in the last few days...

https://cse.google.co.uk/cse?cx=partner … gsc.page=1

Last edited by Slithery (2018-01-07 16:32:57)


No, it didn't "fix" anything. It just shifted the brokeness one space to the right. - jasonwryan
Closing -- for deletion; Banning -- for muppetry. - jasonwryan

aur - dotfiles

Online

#3 2018-01-07 16:56:30

ugjka
Member
From: Latvia
Registered: 2014-04-01
Posts: 1,358

Re: Performance degradation after Meltdown mitigation?

https://www.epicgames.com/fortnite/foru … ity-update

The following chart shows the significant impact on CPU usage of one of our back-end services after a host was patched to address the Meltdown vulnerability.

https://i.imgur.com/h2KcUlE.png

----------

Metldown Patch on Linux - Interrupt Activity

https://imgur.com/a/Bs8Ao

Last edited by ugjka (2018-01-07 17:03:31)


ENTER

~ Arch is not Arch unless it is Arch ~

Offline

#4 2018-01-07 17:50:25

thatnewyorker
Member
From: Brooklyn
Registered: 2009-01-24
Posts: 77

Re: Performance degradation after Meltdown mitigation?

I'm afraid to update the kernel because I run Windows 10 in a VM with pci passthrough so I can play Windows games without having to dual-boot. I wish I had some solid numbers to know just how bad it'll be.


R.I.P In Pieces

Offline

#5 2018-01-07 18:02:28

globalmatador
Member
Registered: 2017-11-21
Posts: 2

Re: Performance degradation after Meltdown mitigation?

thatnewyorker wrote:

I'm afraid to update the kernel because I run Windows 10 in a VM with pci passthrough so I can play Windows games without having to dual-boot. I wish I had some solid numbers to know just how bad it'll be.

you can turn it off via kernel paramters nopti or pti=off

Offline

#6 2018-01-07 18:13:17

Thorsten Reinbold
Member
From: Germany
Registered: 2011-12-06
Posts: 325

Re: Performance degradation after Meltdown mitigation?

Sorry, but not to closing a big security hole because of a few FPS (it really shouldn't be much) is a bit stupid.

Last edited by Thorsten Reinbold (2018-01-07 18:15:23)

Offline

#7 2018-01-07 18:21:58

ewaller
Administrator
From: Pasadena, CA
Registered: 2009-07-13
Posts: 16,756

Re: Performance degradation after Meltdown mitigation?

thatnewyorker wrote:

I'm afraid to update the kernel because I run Windows 10 in a VM with pci passthrough so I can play Windows games without having to dual-boot. I wish I had some solid numbers to know just how bad it'll be.

Ah, but the next logical question is, what are you going to do about updates to the Windows 10 kernel?


Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Sometimes it is the people no one can imagine anything of who do the things no one can imagine. -- Alan Turing
---
How to Ask Questions the Smart Way

Offline

#8 2018-01-07 18:29:27

thatnewyorker
Member
From: Brooklyn
Registered: 2009-01-24
Posts: 77

Re: Performance degradation after Meltdown mitigation?

globalmatador wrote:

you can turn it off via kernel paramters nopti or pti=off

That's good to know, guess I should update then.

ewaller wrote:

Ah, but the next logical question is, what are you going to do about updates to the Windows 10 kernel?

Cry sad

Last edited by thatnewyorker (2018-01-07 18:32:45)


R.I.P In Pieces

Offline

#9 2018-01-07 19:07:58

Head_on_a_Stick
Member
From: London
Registered: 2014-02-20
Posts: 4,940
Website

Re: Performance degradation after Meltdown mitigation?

OK, good news folks — I've just had a quick blast of CS:GO with the page table isolation in effect and the framerate seems undiminished (as measured by cl_showfps 2), AFAICT.

EDIT: Intel Haswell (HD4600) laptop.

Last edited by Head_on_a_Stick (2018-01-07 19:09:15)

Offline

#10 2018-01-07 20:30:23

seth
Member
Registered: 2012-09-03
Posts: 12,635

Re: Performance degradation after Meltdown mitigation?

Thorsten Reinbold wrote:

Sorry, but not to closing a big security hole because of a few FPS (it really shouldn't be much) is a bit stupid.

Is it runtime switchable?
(If you're sure not to run external code, eg. no browser process etc., it'd be justifiable to temporarily switch it off when gaming - a few FPS can make the difference between playable and stuttering. Though afaiu gaming performance shouldn't be notably impacted anyway by the nature of games and this bug)

Offline

#11 2018-01-07 20:42:29

Thorsten Reinbold
Member
From: Germany
Registered: 2011-12-06
Posts: 325

Re: Performance degradation after Meltdown mitigation?

Regarding to some benchmarks on phoronix, the impact in gaming performance is going to zero. Seriously: There's no reason to panic.

Offline

#12 2018-01-07 21:12:43

loqs
Member
Registered: 2014-03-06
Posts: 8,715

Re: Performance degradation after Meltdown mitigation?

seth wrote:

Is it runtime switchable?

@seth not currently https://www.spinics.net/lists/kernel/msg2691093.html

Offline

#13 2018-01-11 23:36:27

possessedbysatan
Member
Registered: 2018-01-11
Posts: 7

Re: Performance degradation after Meltdown mitigation?

I ran a quick test which involved doing tons of syscalls inside a for loop and the performance drop was pretty much noticeable even with pti=off.
Linux 4.14 unpatched: 4.100s (avg)
Linux 4.15 w/nopti: 4.800s (avg)
macOS with meltdown patch: 12-14s (same hardware)

You might want to check this out. Usage: time ./test

#include <unistd.h>
#include <sys/syscall.h>
#include <stdio.h>
int main(void)
{
	long value;
	for (int i=0; i<100000000;i++)
		value=syscall(SYS_getpid);
	printf("%d\n",(int)value);
return 0;
}

Offline

#14 2018-01-11 23:56:58

EmilyShepherd
Member
From: Bucks, UK
Registered: 2016-03-20
Posts: 45
Website

Re: Performance degradation after Meltdown mitigation?

possessedbysatan wrote:

I ran a quick test which involved doing tons of syscalls inside a for loop and the performance drop was pretty much noticeable


In fairness due to what KPTI is it was always going to add some overhead to syscalls, so a test that just does them over and over was always going to be slower. While this is a useful benchmark, it doesn't tell us much about how the patches affect day to day / real usage.

Offline

#15 2018-01-12 00:52:41

exad
Member
Registered: 2016-11-15
Posts: 33

Re: Performance degradation after Meltdown mitigation?

I'm using an apollo lake n4200 cpu and I don't really notice any significant change in speed as per my day to day activities. I remote into a windows machine when I want to do some light gaming but to be fair it's an AMD machine that wasn't all that powerful to begin with. An fx8320@4.2Ghz + 750ti.

All in all, I'm just glad the patches are out.

Of course, since patch source is visible, I wonder how long it will be before the patch code itself is attacked...

Last edited by exad (2018-01-12 04:05:22)

Offline

#16 2018-01-12 01:17:52

Batou
Member
Registered: 2017-01-03
Posts: 221

Re: Performance degradation after Meltdown mitigation?

ugjka wrote:

What is your experience?

Before, my fans would rarely turn on when I wasn't compiling, recoding media,  or running some programs that require lots of CPU. After all the patches, my fans turn on when I open my Youtube subscriptions page. They also turn on when I'm playing a movie. Yeah, it's not fun.


Please vote for all the AUR packages you're using. You can mass-vote for all of them by doing: "pacman -Qqm | xargs aurvote -v" (make sure to run "aurvote --configure"  first)

Offline

Board footer

Powered by FluxBB