Linux 2.6.32 Kernel Benchmarks

[phoronix]

Phoronix: Linux 2.6.32 Kernel Benchmarks

With the Linux 2.6.32 kernel being released in a few days, we found it time to benchmark this newest kernel release that brings new drivers, kernel mode-setting improvements, virtualization enhancements, and more.

http://www.phoronix.com/vr.php?view=14403

[hax0r]

I'm starting to think that those big boys should really start testing and profiling their crap. I wonder how 2.6.24 would compare. Good article.

[Apopas]

This x264 boost is rather interesting, but this postmark and iozone regressions are due to the ext4 "problem"?

[blackshard]

I'm starting to think that those big boys should really start testing and profiling their crap. I wonder how 2.6.24 would compare. Good article.

I asbolutely quote here.

[Ant P.]

This x264 boost is rather interesting, but this postmark and iozone regressions are due to the ext4 "problem"?

No idea what's wrong with ext4, but Con Kolivas needs three cheers for getting them to finally fix their damn scheduler.

[Michael]

Read: http://www.phoronix.com/vr.php?view=14285 where PTS shows the specific commit that causes the performance drop.

[Darkfire Fox]

I was really psyched about Ext4 when it first came out, but perhaps I should just stick with good old Ext3 for another year...

[Jimmy]

Meh, just get an ups and mount with nobarrier.

[Ranguvar]

No idea what's wrong with ext4, but Con Kolivas needs three cheers for getting them to finally fix their damn scheduler.

Well, CK and Dark Shikari. http://x264dev.multimedia.cx/?p=185

@Phoronix team:

...the Linux 2.6.32 kernel had the lowest overall CPU usage when using X-Video with MPlayer.

Not lowest. Either 'worst', or 'highest'.

[kraftman]

No idea what's wrong with ext4, but Con Kolivas needs three cheers for getting them to finally fix their damn scheduler.

Apache benchmark is meaningless. Not sure about others, but it's probably because of change in Ext4 like Michael said. Con's scheduler is slower in Apache. Btw. notice performance in not disk related benchmars is as it should be.