Fedora 12 vs. Ubuntu 9.10 Benchmarks

[phoronix]

Phoronix: Fedora 12 vs. Ubuntu 9.10 Benchmarks

Canonical released Ubuntu 9.10 last week, which introduced the Ubuntu Software Center and brought a wide variety of other improvements, while Red Hat is scheduled to release Fedora 12 in two weeks. With the impending release and the current development freeze, we took the compose release candidate for Fedora 12 x86_64 and have looked at how its performance compares to Ubuntu 9.10. In this article are our results, which actually show some rather large differences between Fedora and Ubuntu when it comes to the speed of the Linux desktop.

http://www.phoronix.com/vr.php?view=14355

[droidhacker]

More useless benchmarks....

<<--Yawn-->>

[TeoLinuX]

Maybe benchmarks are not the only target of developers (I hope so). But two systems so similar yet delivering that performance differences is a symptom that something must be broken.

So benchmarking is not to prove "mine is longer than yours" but should give a warning that something is going wrong or at least warn that there's room for improvement

[Deagleson]

Could we perhaps get a benchmark comparing the Open-Source ATI drivers under Fedora 12 contra Ubuntu 9.10 once F12 gets officially released?

[frantaylor]

Regarding disk benchmarks:

These are dependent on the disk controller and the disk.

People who are interested in performance are usually using hopped-up machines and they don't use the motherboard disk controller with a single drive. They are more likely to be using multiple drives with software RAID or a RAID controller, or a SSD.

I don't think it's possible to use these results to project what kind of performance one would get on a high-performance system.

If you are interested in single-disk performance with the motherboard controller, then it's probably because you are using a laptop, in which case the benchmarks would be much more indicative if they were done on a laptop.

[andreskru]

The releases of fedora that arent final release, dont use debug kernel mode or something?

[deanjo]

Regarding disk benchmarks:

These are dependent on the disk controller and the disk.

People who are interested in performance are usually using hopped-up machines and they don't use the motherboard disk controller with a single drive. They are more likely to be using multiple drives with software RAID or a RAID controller, or a SSD.

I don't think it's possible to use these results to project what kind of performance one would get on a high-performance system.

If you are interested in single-disk performance with the motherboard controller, then it's probably because you are using a laptop, in which case the benchmarks would be much more indicative if they were done on a laptop.

This is true of any benchmarking, however in the case linux as they are both using the drivers in the linux kernel on the same hardware. Performance deltas can be accounted for by the tweaking of the kernel module for that hardware (a less likely scenario especially when both system use the same kernel revision) between kernel revisions or system configuration that a particular distro uses and it's selection of supporting libraries (a more likely scenario). Now unless there is a complete incompetence in supporting the hardware at the module level for a piece of utilized hardware one can extrapolate, more or less, which OS will give the user better performance.

Having said all of that, I still think that if PTS is going to be used to bench distro vs distro it should use the precompiled distro packages for it's tests as that is what a majority of linux users will be using instead of downloading and compiling the libraries which is a action that a user is unlikely to do when wanting to run that software. They will usually pull it off the distro's repo if available. After all you are benching a distro's packaging and config which is all a distro actually is. If this capability was put into PTS you could even bench pre-packaged vs built from scratch on the same distro to determine if it is even worth recompiling to see any gains. The way it stands now PTS is really only suited to compare 2 pieces of competing hardware on the same OS such as graphics cards, i/o setups, cpu's etc etc where recompiling of the libs may take advantage of extra device capabilities not included in the pre compiled binaries.

[bugmenot]

How annoying I find those fake hyperlink word ads.

can't move the mouse around without an explosion of ajax ads. Reminds me to only view this site in lynx or the like.

[damentz]

I like how everyone ignores the gaming benchmarks.

They're only more important and interesting than all the other ones.

[DanL]

The releases of fedora that arent final release, dont use debug kernel mode or something?

Beta releases (and earlier), yes; Fedora release candidate does not use kernel debugging symbols.

[Michael]

How annoying I find those fake hyperlink word ads.

can't move the mouse around without an explosion of ajax ads. Reminds me to only view this site in lynx or the like.

Premium can solve that for you: http://www.phoronix.com/scan.php?page=phoronix_premium

[perpetualrabbit]

Since in the first benchmark Fedora is almost exactly twice as fast as Ubuntu, could it be some kind of threading issue? I.e. fedora excuting two threads and ubuntu only one?

It seems a strange coincidence that they differ by a factor of almost exactly 2. How does this benchmark work?

[fxkuehl]

It's hard to tell from the screenshot, but I don't see any shadows around the windows on Fedora. If there is no composite manager running on Fedora, that would give OpenGL applications an advantage on Fedora compared to Ubuntu, which has Compiz running by default.

[ldesnogu]

It's hard to tell from the screenshot, but I don't see any shadows around the windows on Fedora. If there is no composite manager running on Fedora, that would give OpenGL applications an advantage on Fedora compared to Ubuntu, which has Compiz running by default.
Aren't the gaming benchmarks run in fullscreen mode? In which case I hope Compiz gets stopped or at least doesn't interact.

[natewiebe13]

Compiz still runs while in fullscreen mode. When you do the final benchmark with the final release of Fedora 12, make sure you have compiz disabled in Karmic by turning desktop effects to none. I think that is why Fedora performed so much better than Ubuntu when it came to OpenGL.

[ldesnogu]

Compiz still runs while in fullscreen mode.
That seems stupid! After all it's only use is when you're at the desktop right?

BTW I did a quick test: running WoW with and without desktop effects, on my Fedora11. That didn't change the frame rate at all.

[natewiebe13]

That seems stupid! After all it's only use is when you're at the desktop right?

BTW I did a quick test: running WoW with and without desktop effects, on my Fedora11. That didn't change the frame rate at all.

Hmm.. That's strange, but my guess is because it has to do with Wine. I just ran the standard (glxgears -fullscreen) on an old computer of mine. I'm running Ubuntu Karmic with an nVidia 7600GS and the nVidia 185 Drivers from the repositories. With effects set to standard (Compiz), I average at around 68 FPS, if I turn off Comiz by setting effects to none, I average 228 FPS. So there is definitely a chance that the reason Fedora benched higher was because Comiz was still enabled.

[ldesnogu]

I definitely see the same as you for glxgears (5000 vs 2500 fps).
One more reason to not use Compiz :)

I wonder why it doesn't impact Wine...

[Dragoran]

I definitely see the same as you for glxgears (5000 vs 2500 fps).
One more reason to not use Compiz :)

I wonder why it doesn't impact Wine...

glxgears is not a benchmark, it draws to many frames so that any overhead has a great impact on it.

Real apps should not be affected, also compiz has an option to unredirect fullscreen windows i.e leave them alone so it does not have any effect on them at all.

[ldesnogu]

The question here was not to measure exactly the impact of compositing by using glxgears, but to see there is an impact.

All we need to know is whether Michael left compiz on when running the gaming benchmarks.

[Dragoran]

The question here was not to measure exactly the impact of compositing by using glxgears, but to see there is an impact.

All we need to know is whether Michael left compiz on when running the gaming benchmarks.

Still glxgears is the wrong way to do this, because just running a DRI2 driver has an negative impact on glxgears, while real apps tend to be faster/not impacted.

glxgears just renders a very simply frame as often as possible so _any_ additional work could have a great impact on it.

So if you want to test the impact just use any real app.
It would be close to zero if measurable at all.

[ldesnogu]

OK, you know better than I do about this :)

So would you have any guess about why Michael's results were so bad?

[BlackStar]

Hmm.. That's strange, but my guess is because it has to do with Wine. I just ran the standard (glxgears -fullscreen) on an old computer of mine. I'm running Ubuntu Karmic with an nVidia 7600GS and the nVidia 185 Drivers from the repositories. With effects set to standard (Compiz), I average at around 68 FPS, if I turn off Comiz by setting effects to none, I average 228 FPS. So there is definitely a chance that the reason Fedora benched higher was because Comiz was still enabled.

I have a gut feeling that Karmic disables the "unredirect fullscreen windows" option by default. I'm saying that because popup notifications no longer cause my fullscreen applications to flicker and flash like crazy, like they used in Jaunty.

While such a change would make for a better desktop experience, it would impact fullscreen 3d performance quite a bit, especially on lower-end / bandwidth-limited hardware (redirected windows cannot page-flip, so they have to copy memory from back- to front-buffers every frame - which translates into a few hundred MB/sec of "lost" texture bandwidth).

[damentz]

It's not compiz.

It would make sense for Fedora to backport the fixes in 2.6.32 of CFS due to the performance regressions exposed during benchmarks against BFS. They tend to grab a lot of new features and mix stable with new when there's no obvious stability penalty.