Multi-Core, Multi-OS Scaling Performance

[phoronix]

Phoronix: Multi-Core, Multi-OS Scaling Performance

In this article we are looking at how Linux, OpenSolaris, and FreeBSD scale across multiple cores. Benchmarked are CentOS 5.5, Fedora 14, PC-BSD/FreeBSD 8.1, and OpenIndiana b148 as we see how the performance differs when running on one, two, three, four, and six cores, plus when Intel Hyper Threading is enabled.

http://www.phoronix.com/vr.php?view=15706

[Shining Arcanine]

Michael, if you want to talk about muticore scaling with compilation benchmarks, then you should specify important things like what MAKEOPTS variable was being used. Without that information, your compilation results are meaningless.

[kraftman]

It will be great to see Kubuntu (unlike Fedora it has sel disabled), 2.6.38 kernel (scalability improvements) and os x (does it scale?).

[allquixotic]

The green "spider" like shape of the BSD graph on LU.A was amusing. It's so out there that it seems quite implausible. Are OS schedulers really that finnicky?

[V!NCENT]

I must say that I'm realy pleased with the Fedora 14 KDE spin. So far it hasn't crashed and feels significantly faster than Windows 7 om my new laptop, although I haven't done any poweruser things on it; Rekonq, Koffice, Okular, AmaroK, VLC, PIM, Virtualbox with AMD-v and jMonkeySDK.

Powerdevil on powersave doesn't feels like an impact on performance although file indexing is then put on hold.

It's pretty surprising to me as I'm used to preload on my desktop.

Core i3 370m 4 cores (Fusion where were you?), 4GB RAM, 5400rpm hdd, HD5470 with 512MB dedicated RAM. Fglrx feels like it doesn't outperform the stock driver (is it gallium?) on the desktop (only when I run glxgears), so I feel like switching back.

[jakubo]

no ubuntu test results, huh? did you have an argument with mark shuttleworth lately?
just kidding.
i just wonder why not using the .38 kernel with the big-lock-less option an those other patches.
how can it be that in the LU.A test starts at the performance of 4 cores? how did you normalize the graphs? everyone by itself or is there a common reference between the graphs?
considering the greater steepness of bsd's graph after 4 cores im curious to see whether the steepness remains this way for 12 cores (no HT) maybe on your dual amd system or whatever it was.

[jakubo]

ok, sorry no common reference... but it cannot start at 4 then...

[ChrisXY]

When I saw the title I immediately knew there would be no information about what CPU scheduler is used in these benchmarks. I just have to assume it is CFQ in all Linuxes? Why don't you benchmark other schedulers too?

In other benchmarks it was shown that the filesystem actually can have a rather big impact on compiling. Why not also try it within a ramdisk (maybe with some filesystem formatted all benchmarked operating systems support)?

Would you PLEASE stop using colored subpixels?

That it looks good on YOUR screen doesn't mean it will on every screen.
Yes, I can see the colors with 100% zoom and yes, my eyes are very confused by it.

[crazycheese]

ChrisXY
Maybe vector result table is rasterized first and then scaled down + saved into PNG, instead of scale down first and then rasterize. Bicubic scaler artifacts..?

[devius]

Can anybody explain what those "harmonic mean" numbers mean? Why are smaller values better?