Ubuntu vs. Windows In OpenOffice.org Benchmark

ahziem writes "Ubuntu's Intrepid Ibex and Redmond's Windows XP go head-to-head in an OpenOffice.org 3.0 performance smackdown measuring vanilla OpenOffice.org, StarOffice, Go-oo, and Portable OpenOffice.org 3.0. Each platform and edition does well in different tests. Go-oo is known for its proud slogan "Better, Faster, Freer," but last time with OpenOffice.org 2.4 on Fedora, Go-oo came in fourth place out of four. Slashdot has previously reported Ubuntu beating Vista and Windows 7 in benchmarks, so either XP is faster or this benchmark carries a different weight."

Comments

Big surprise

[Bobnova]

XP faster then vista/7? I'm shocked. I've been doing some general testing between XP and ubuntu 8.10 as well as dellbuntu 8.04. Ubuntu gets 25% longer battery life on my netbook, but cannot play youtube videos (on either version) without lurching video. XP on the same netbook does youtube just fine, but has a 3 hour batter life to ubuntu's 4 hour. On an old p4 i have xp scrolls smoothly and instantly in firefox, where 8.10 has a delay before anything happens. My conclusion: On a slow system, XP is faster.

Re:Big surprise

[man_of_mr_e]

Could it be that playing Youtube videos uses 25% more cpu power? And thus, because you didn't play them on your ubuntu laptop it got longer battery life?

Re:Big surprise

[0xABADC0DA]

Some tips on netbook power. Hopefully /. will correct anything wrong here:

1) Underclocking can have huge savings... as much as the backlight being on/off. I don't mean using cpufreq to change processor frequency... the power savings apparently comes from the ram and slowing down the ridiculously bad Intel GMA945. This is generally easier on XP since the OEM will have some software to do this, and nothing pre-packaged exists in Linux afaik.

2) Use a plain background and plain graphics... no gradients or pictures. GMA can use run-length encoding to compress the display memory on a line-by-line basis, and if the line hasn't changed the display uses the compressed version.

Somebody check my numbers... assuming 666 fsb, that's 666Mhz*4 bytes per second. The display might use 1024*600*3 bytes and if it refreshes the display at 60 fps, the shared memory for the display uses:

(1024*600*3*60) / (666*Mhz*4) = 15% of fsb time

That must be wrong, because at high res it would be using all the time. But I don't know what assumption is wrong... but anyway if you can compress by say 80% by using solid colors (or vertical gradients) then you can save some power and make the system somewhat faster. This might have to be turned on with the driver, idk if linux driver can do this.

3) Some USB devices use a lot more power than you'd expect. For instance a standard USB laser mouse can use a watt from various things like having USB polling it frequently.

4) As far as I can tell from reading the web, RAM power is basically how many modules you have installed not how much memory is on them. Maybe it's based on the number of chips? Anwyay it looks like upgrading memory should increase battery life by reducing disk access. So for instance if the system has low ram, like 512mb you might see disproportionately better power on linux since it generally uses less ram, so less hd activity.

5) It's almost not worth it to put the hard disk to sleep. Modern laptop drives you might save .2-.4w over just idle, but spin up might take 5w. So telling hd to spin down every 3 min for instance might actually use more power.

Re:Big surprise

[Toonol]

Math still somewhat grates my aesthetic sensibilities in that respect. Ubuntu takes 33% longer than XP, while XP is 25% slower than Ubuntu. It just doesn't seem symmetrical... although it really is.

Re:Big surprise

[Tycho]

Facepalm! That is not much of a performance review. The reviewer gives the a score of 1 to the OS that was fastest on a given test for hardware second best OS a 2, third best a 3, fourth best a 4, and worst a five. So a test where the difference between the fastest and slowest is 5% and another where the difference is 50% both get scored the same. Add the results from the 31 tests done and a pre-release version of Windows 7 can look very nice even if the margin of final scores do not accurately represent the actual performance. Worse yet, how do we know whether of not there are still features, DRM or otherwise, that cripple performance? We don't know, and we won't know until Microsoft releases feature complete versions of the various Windows 7 version tiers.

Re:Big surprise

[jvin248]

I run Xubuntu on Pentium-3 @ 500Mhz & 256MB Ram, and a laptop with Linux Mint Fluxbox with same specs - about half of what your Netbook is probably running (most 1 to 1.6Ghz)- and can watch Youtube... Change your Ubuntu from Gnome to XFCE or Fluxbox window managers and you'll have a big speed increase. You can probably even use KDE and have success (Kubuntu). KDE runs on 700Mhz or better cpu's.

Re:Big surprise

[HiThere]

I'd heard that some DRM ran as a daemon. And consumed cycles in the background even if you weren't using anything related to it. Could be wrong, of course, but it sure sounds reasonable.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:OpenOffice benchmarks? Seriously?

[Iyonesco]

My first thoughts were also "Is speed really the issue here?" but for different reasons. I used Open Office for eight months before having to give up due to a massive number of small niggles that when combined make it very unpleasant to use. I think a lot of issues need to be addresses in Open Office before speed but sadly none of the problems ever seem to be addressed and they instead seem to focus on adding new features. In the end I had to give up and switch to Kingsoft Office 2009.

Re:Another Ubuntu-Windows Benchmark?

[Vectronic]

I've got absolutely no problem with them, provided they are reasonably accurate, and neither the summary, nor article is flamebait.

Hell, i'd probably even frequent an entire section devoted to it.

Re:Like Windows users are gonna care

[Anonymous Coward]

Who runs OO on Windows?

Any of my clients who came to me with a new computer recently.

Uninstall annoying McAfee/MS Office trial versions + Install AVG and OpenOffice = Happy client.

Re:Who cares about CPU speed if it slows your work

[DesertBlade]

I use xls on both Excel and Open Office and they are mostly compatable. If you are one of those accounting types with 100000 lines in an excel file then you you should stick with excel.

Open Office is a replacement for M$ office for 95% of the use cases. Still the proprietary formats of M$ Office made it difficult to port. Since those standards are now published I think cross program support will improve.

Re:Like Windows users are gonna care

[Anonymous Coward]

Who runs OO on anything? I use Ubuntu as my primary desktop/dev OS, but I don't use OO for anything.

It's slow, and it doesn't work for anything beyond a very trivial subset of Office functionality.

When I need Office, I go to my Windows laptop.

OO = total fail.

Re:One phrase invalidates the whole shebang...

[TheRaven64]

I came across a really fun bug in one version of the MS compiler where a colleague had tried using the SEE intrinsics to speed things up and found that it had become slower. Looking at the generated asm, it turned out that it was doing a function call for every intrinsic (while GCC just issued a single SSE instruction). I think the MS approach was to generate them as function calls and then use the inliner to turn them into single instructions, but for some reason the function inlining pass wasn't being run.

Not intending to bash the Microsoft compiler particularly here, just pointing out that the options you select for the compiler can make a huge impact on any compiler, often more than the difference between switching to a different compiler.

Re:Like Windows users are gonna care

[goltzc]

I own a small web development company. The company is basically 3 members and the occassional contractor. I run Ubuntu and the other two run vista and OSX on out development machines. We use Open Office so we can be platform agnostic.

Comment removed

[account_deleted]

Comment removed based on user account deletion