Ubuntu, Fedora, Mandriva Performance Compared

[ivanovic]

I think what is most interesting to me in these results is the difference in compilation times, especially for the kernel. I guess, in that case, one could argue that there is simply more source to compile for 2.6.25 than for 2.6.24, or that Fedora 9's gcc is slower than Ubuntu 8.04's.

Uhm, the kernel that is compiled is identical for all tests. The only "real" difference is the kernel of the underlying system as well as the compiler used. And it is *very* likely that gcc 4.3 is slower than 4.2.x but probably it also produces more/better optimized binary code.

[Del_]

Thanks for a fantastic job Michael. Just a little nitpick, please make clear whether OS and/or binaries are 64-bit.

[Extreme Coder]

Thanks for the benchmarks Michael..
As you said, what matters more is what features you require from a distro
I use Mandriva, and I am glad with it.

[jeffro-tull]

wow. Fedora, Ubuntu, and something that isn't OpenSolaris? I'm impressed. And happy. I cut my Linux teeth on Mandrake, and it's still my Plan B for desktop usage.

[Luis]

That's a pretty old post. Here are a couple of benchmarks with barrier enabled. Top one is a couple of old Maxtor 6L250S0 250 Gig drives with barriers enabled, the one below are Seagate 7200.11 500 Gig drives. Both are running Raid 0 on a dmraid setup.

http://global.phoronix-test-suite.co...97-27642-20171

Are you certain that barriers are supported in that setup? I'm inclined to think they aren't. You should check your dmesg for any messages about it.

Besides, there is nothing to compare against, so it's hard to say how much barriers are hurting performance.

[apaige]

The sqlite benchmark with barrier enabled by default on my XFS partitions took an insane amount of time on my old SATA drive, like 400+ seconds. When I re-mounted the partition with nobarrier, it went down to… 2.7 seconds.

As for compilers: GCC 4.3.x produces much faster binaries on modern CPUs in certain (not all) cases. It's a pity the recently released Ubuntu still uses GCC 4.2.x, although it's understandable (GCC 4.3.x is brand new and that ubuntu release boasts a two-year support IIRC).

[yoshi314]

In many of these tests, the results were very similar while in others there was a much greater delta. As to which distribution is the fastest, from all of these tests and their varying results, it really depends what areas of the Linux desktop are important to you.

i thought the desktop was about low latency, and responsiveness.

this article makes no sense. some tests are done only on 2 distros, and in the end there is no definite answer.

there should be more information on kernel used in each distribution, and distinct patches it used. also what services were running at the time of benchmark, etc.

[shahgols]

Can you please do a comparison between Vector Linux (i686 optimized) and some i386 and i586 optimized distribution? There are many posts on the internet that discuss how i686 users experience faster response time from their distributions, but then there are others on those same threads that say that this cannot be true. Can you please put this to the test for us?

Thanks!
Vector Linux SOHO user

[deanjo]

Are you certain that barriers are supported in that setup? I'm inclined to think they aren't. You should check your dmesg for any messages about it.

Besides, there is nothing to compare against, so it's hard to say how much barriers are hurting performance.

I am very sure that the hardware supports barriers. When I run the exact same distro on older hardware, when the hardware does not support barriers dmesg puts out a line something like "barriers not supported, disabling barriers".

[Luis]

I am very sure that the hardware supports barriers. When I run the exact same distro on older hardware, when the hardware does not support barriers dmesg puts out a line something like "barriers not supported, disabling barriers".

Ok, thanks for clarifying. I was unsure if barriers would work on a raid0 setup.