Windows vs. Linux vs Unix

Windows vs. Linux vs Unix

Post by Steve Guggenmo » Fri, 14 Mar 2003 09:51:10



We have been fortunate enough to benchmark a model's performance on
two identical servers. One with Windows NT for servers, and the
other with Linux. The two servers each have 2 processors, spreed is
1.40 GHz, RAM is 1.26 GB and also the same storage 36 GB. The Unix
machine is not identical, but similar in most specs.

The same code is submitted on both (using -nojmv), it writes thousand
of files, because I want to test file IO, with the total size of
files reaching between 20 and 30 GB.

My guess was that Linux (and Unix) would outperform (because Windows
automatically takes 1/2 of all RAM avaiable) and I thought it would
have less overhead.

The model finished in about 15 hours in the Windows environment. It
has been running twice that long on Linux, and is not even half done.
 Unix performance looks similar to Linux.

Is this surprising to anyone?
Do you think my Linux system could be improved?
If you were forced to do this much file IO,what would you do to
improve performance?
Any other insights?

Everyone's perspectives on this would be greatly appreciated.

Thanks
Steve

 
 
 

Windows vs. Linux vs Unix

Post by Joshua Baker-LePai » Sat, 15 Mar 2003 00:35:19



> We have been fortunate enough to benchmark a model's performance on
> two identical servers. One with Windows NT for servers, and the
> other with Linux. The two servers each have 2 processors, spreed is
> 1.40 GHz, RAM is 1.26 GB and also the same storage 36 GB. The Unix
> machine is not identical, but similar in most specs.
> The same code is submitted on both (using -nojmv), it writes thousand
> of files, because I want to test file IO, with the total size of
> files reaching between 20 and 30 GB.
> My guess was that Linux (and Unix) would outperform (because Windows
> automatically takes 1/2 of all RAM avaiable) and I thought it would
> have less overhead.
> The model finished in about 15 hours in the Windows environment. It
> has been running twice that long on Linux, and is not even half done.
>  Unix performance looks similar to Linux.
> Is this surprising to anyone?

Yes.  Our stuff runs faster on Linux than it does on 'doze.

Quote:> Do you think my Linux system could be improved?

More than likely.

Quote:> If you were forced to do this much file IO,what would you do to
> improve performance?

It depends.  You don't give all your stats, but make sure that your
drive controller is well supported and running optimally.  From your
disk size, it looks like SCSI, but if it's IDE, make sure DMA is
enabled.

Also, you may want to check out other filesystems.  If the files are
large, look at XFS.  If they're small, mabye ReiserFS would help.

Quote:> Any other insights?

More hardware info would be helpful, as well as Linux distro.

--
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

 
 
 

Windows vs. Linux vs Unix

Post by Anthony Gallaghe » Mon, 17 Mar 2003 14:41:43


If you do a search of the posts on the ng, other posters have
ocasionally reported similar problems. It seems that TMW Matlab Windows
version is more optimized than the Unix version for some operations.

Anthony


> We have been fortunate enough to benchmark a model's performance on
> two identical servers. One with Windows NT for servers, and the
> other with Linux. The two servers each have 2 processors, spreed is
> 1.40 GHz, RAM is 1.26 GB and also the same storage 36 GB. The Unix
> machine is not identical, but similar in most specs.

> The same code is submitted on both (using -nojmv), it writes thousand
> of files, because I want to test file IO, with the total size of
> files reaching between 20 and 30 GB.

> My guess was that Linux (and Unix) would outperform (because Windows
> automatically takes 1/2 of all RAM avaiable) and I thought it would
> have less overhead.

> The model finished in about 15 hours in the Windows environment. It
> has been running twice that long on Linux, and is not even half done.
>  Unix performance looks similar to Linux.

> Is this surprising to anyone?
> Do you think my Linux system could be improved?
> If you were forced to do this much file IO,what would you do to
> improve performance?
> Any other insights?

> Everyone's perspectives on this would be greatly appreciated.

> Thanks
> Steve

 
 
 

1. What to buy: P4s vs. PIIIs, Windows vs. Linux

OK, so here is a question that LOTS of folks are liable to
 have an opinion on:

 I'm trying to figure out what sort of a computer to buy.
 My only reason to own something other than a Mac is to
 run MATLAB, so I don't care if it does anything well
 except that, that's all it will be used for. So....if
 you had to buy something right now, would you opt for
 a P4 or still stick with a PIII? Is there any performance
 advantage of Linux over Windows 2000? (No XP for me, please!)
 If you were looking for a lot of RAM (say 1G or so), CD-RW/DVD,
 40G+ HD, and had a budget of ~$3K, what would you get?

 Thanks in advance!

 BMW

 Barry M. Wise, Ph.D.
 President
 Eigenvector Research, Inc.

2. Help with Multiprocessing

3. Unix/Linux vs. Windows portability

4. Dual-boot Dual install

5. True Unix 64 vs alpha linux and MATLAB

6. My life is drawing to a close :-(

7. C-MEX file execution speed (Unix vs. Windows)

8. Connector Cable for a NCD X-Terminal and Supra Fax Modem

9. Windows vs. Unix versions of Matlab

10. windows vs unix

11. Matlab-Windows NT vs Matlab-Unix ?

12. Matlab R13 on linux vs. windows

13. Ideal Config of a PC for Matlab - Linux Vs Windows as well as Processors?