Ubuntu 7.04 to 8.10 Benchmarks: Is Ubuntu Getting Slower?

[tzu23]

i just uploaded my audio-encoding results.
-> http://global.phoronix-test-suite.co...443-32506-2318

they are twice as fast as the ones presented on the web site. There is something terribly wrong with the intrepid installation/configuration of the (phoronix) test machine. Or the test has quite a few ugly bugs.

[obscurite]

Flawed approach?

Michael Larabel - listen up! How can you claim that any platform that is hardware dependent is slowing down after testing it on one system! Several of us have added our two cents about this, please respond with a correction!

The information you provided is interesting and useful, but the title of the article should have been "Is 32 bit ubuntu on a T60 getting slower".

Also, concerns about destructive background processes should be addressed - I know on at least one laptop that the indexer was slowing down my system on 7.10 and 8.04. This might best be disabled on every test since it's potentially broken depending on your system.

Try a few different hardware systems, and try 64 bit linux as well! It really annoys me to see such a grandiose claim being made without a proper investigation on a site I really respect. Thanks for the hard work, but you can do better!

PS. I registered just so I could post this comment!

[Michael]

Michael Larabel - listen up! How can you claim that any platform that is hardware dependent is slowing down after testing it on one system! Several of us have added our two cents about this, please respond with a correction!

The information you provided is interesting and useful, but the title of the article should have been "Is 32 bit ubuntu on a T60 getting slower".

Also, concerns about destructive background processes should be addressed - I know on at least one laptop that the indexer was slowing down my system on 7.10 and 8.04. This might best be disabled on every test since it's potentially broken depending on your system.

Try a few different hardware systems, and try 64 bit linux as well! It really annoys me to see such a grandiose claim being made without a proper investigation on a site I really respect. Thanks for the hard work, but you can do better!

PS. I registered just so I could post this comment!

Note that I didn't make any definitive claims about Ubuntu getting slower but more along the lines of provided some thoughts on what I had experienced, posed some questions, didn't attempt to make any blanket claims, and I didn't start pointing any fingers around.

Though reading the feedback though from here and on the Ubuntu Forums in particular, others definitely agree that there are areas within Ubuntu that could be better optimized for performance. Going out of this hopefully the open-source community will pay better attention to performance testing.

As mentioned several times now, other systems and distributions are in the process of being tested, but alas with limited staff (essentially just me) and very limited resources (I can assure you Phoronix Media is no cash cow), it's not feisable to test everything out there.

[deanjo]

but alas with limited staff (essentially just me) and very limited resources (I can assure you Phoronix Media is no cash cow), it's not feisable to test everything out there.

It's a real bitch having to write yourself up on disciplinary issues when that happens.

[Zooko]

Way to go, to everyone who contributed to the creation of the Phoronix Test Suite and to these benchmarks of using that test suite on various Ubuntu releases. Thank you!

I'm happy that future releases of Ubuntu (and probably other Linux) are going to get noticeably faster, and in large part thanks to your work. Keep it up!

[pataphysician]

An idea on where the problem possibly lays. Some posters have reported no real differences on their T60's across Ubunutu versions, and some have posted speeds that don't correlate to Phoronix's on 8.04 8.10 on T60's. But all these T60's seem to be Merom Core 2 Duo's, not Yonah Core Duo's.

I propose that the problem might lay in some Ubuntu's versions interaction with the Yonah Core Duo feature set, or possibly with chipsets, though I don't know if the motherboards are different, they very well could be the same.

I do remember some kind of problem I saw a couple years ago, with I think Mandriva, on a Core Duo, which was not related to EIST, where both cores reported as up and working, but one core ran incredibly slow, this would effect non-multi-threaded tasks more than others, if they were running only on that core. No problems in this regard when I put Debian back on it instead.

[Michael]

For those interested, the ThinkPad T60 we used in testing was a 2006 model, 2613-EJU.

Fedora benchmarks will be out tomorrow and more to come next week.

[deanjo]

One thing that occurred to me is that possibly some thermal throttling was going on which is possible if the system was being stressed for a long period of time.

[obscurite]

You missed my point.

Note that I ... didn't attempt to make any blanket claims, and I didn't start pointing any fingers around.

Michael, thanks for the response, much appreciated. However, consider that when you ask if there is something wrong in a headline, that you are making a form of blanket claim and pointing fingers. The claiming and the pointing is implicit, but it's enough to confuse a whole lot of people who take headlines at face value. This technique is used for character assassination all the time in politics and entertainment journalism to discredit someone without having to back up any claims. In fact, it's being used quite a bit in the US presidential race right now. Perhaps it's just the political climate that's making me a bit more sensitive to truth and disclosure than normal.

Anyway, thanks for a great site and for the work you're doing, and I hope I won't be complaining again any time soon.

[Beleriand]

Is the Phoronix Test wrong ?

Hello!

to show the contrary to Phoronix results I did same testing on my own computer with different Ubuntus.

Hardware:

• CPU: AMD Athlon M XP 2500+ (Barton) @ 1920 MHz (Mainboard doesn't get the name correctly)
• RAM: 2GB MDT DDR2-800
• GPU: Asus N6600GT 128MB (Nvidia 6600GT)
• MB: MSI K7N2-Delta
• HDD: Samsung SpinPoint P120 SP2514N 250 GB

Software:

• Ubuntu 6.06, Ubuntu 7.10, Ubuntu 8.04, Ubuntu 8.10
• all are standard installs and up-to-date (10-30-08), means for Intrepid it is final release
• graphic card works on all systems with the latest nvidia-glx available in official restricted repository
• extra packages are only build-essential and libgtk2.0-dev and dependencies and
• phoronix-test-suite 1.2.2 (DEB) from Phoronix and
• gtkperf 0.40 (DEB) from Sourceforge

Sure, these are only few tests. I will do some more testing when I have the time do to it.

Benchmarks:

• ffmpeg http://global.phoronix-test-suite.co...12-22916-20690
• audio-encoding http://global.phoronix-test-suite.co...62-27698-13221
• memory http://global.phoronix-test-suite.co...63-12840-24973
• gtkperf GtkComboBox http://global.phoronix-test-suite.co...153-4925-18338
• gtkperf GtkRadioButton http://global.phoronix-test-suite.co...1462-6117-5310
• gtkperf GtkDrawingArea-Circles http://global.phoronix-test-suite.co...39-22033-20656
• gtkperf GtkDrawingArea-Pixbufs http://global.phoronix-test-suite.co...6404-21546-599

Because of the gtkperf TotalTime taking too long I installed the DEB from Sourceforge gtkperf 0.40:

• gtkperf 0.40

As you can see, the trend of the Phoronix results can not be affirmed. Some of them even show Ubuntu is getting faster!

Greetz,
Beleriand