Menu
▾
▴
colinux-users
|
From: Curtis, C. <Cra...@ec...> - 2004-09-30 16:52:30
|
I just wanted to ask about the performance of colinux. I am running win2000 on a 2.6GHz PC. I set up a linux partition with debian to be used with colinux. I also set the swap to a real swap partition. When I go through to make a project, it is MUCH slower than a 400MHz PC. I can look at my hard drive and see that the harddrive is not being touched almost at all during compilation, so this tells me I have enough memory. Is colinux just much slower than a standalone linux PC? Thanks, Craig |
|
From: Holger K. <hol...@gm...> - 2004-09-30 17:58:52
|
> I go through to make a project, it is MUCH slower than a 400MHz PC. I > can look at my hard drive and see that the harddrive is not being > touched almost at all during compilation, so this tells me I have enough > memory. Is colinux just much slower than a standalone linux PC? According to my testing, you get the full processing power of your cpu. I tested that with numerical simulation programms (fpu bound). It doesn't make a difference real linux vs colinux. But i don't use swap at all, and know nothing about the disk speed you get within colinux. |
|
From: Curtis, C. <Cra...@ec...> - 2004-09-30 18:02:03
|
Holger Krull wrote: >> I go through to make a project, it is MUCH slower than a 400MHz PC. >> I can look at my hard drive and see that the harddrive is not being >> touched almost at all during compilation, so this tells me I have >> enough memory. Is colinux just much slower than a standalone linux PC? > > > According to my testing, you get the full processing power of your > cpu. I tested that with numerical simulation programms (fpu bound). It > doesn't make a difference real linux vs colinux. > But i don't use swap at all, and know nothing about the disk speed you > get within colinux. > > > The deal is the hard drive is not really being utilized. I can look at Windows 2000's CPU utilization and during compilation, it only goes up to 1%. |
|
From: Holger K. <hol...@gm...> - 2004-10-01 05:19:03
|
> The deal is the hard drive is not really being utilized. I can look at > Windows 2000's CPU utilization and during compilation, it only goes up > to 1%. That's why i suspect disk-io problems. If the compiler is waiting for the disk it will not use cpu time. A heavily fragmented file for the simulated hard disk could be a reason. Do a bonnie test. How much RAM did you give colinux? Maybe if you have a tmpfs put all the source there and compile there. |
|
From: Curtis, C. <Cra...@ec...> - 2004-10-01 11:20:14
|
My problem had to do with network latency. I was mounting an SMB share and building off of it. When I copied it to an EXT2FS, it worked wonderfully. I am now EXTREMELY happy with colinux. This is an awesome project! Thanks, Craig |
|
From: Holger K. <hol...@gm...> - 2004-10-01 11:31:10
|
> My problem had to do with network latency. > I am now EXTREMELY happy with colinux. This is an awesome project! That's true, it is awesome, the network beeing the only roadblock. |
|
From: David B. <db...@so...> - 2004-10-01 13:18:25
|
Ahh but samba support under *nix has always sucked hasn't it ;) Colinux just inherited it ^_^;; - Dave -----Original Message----- From: col...@li... [mailto:col...@li...] On Behalf Of Holger Krull Sent: Friday, 1 October 2004 9:31 PM To: col...@li... Subject: Re: [coLinux-users] performance > My problem had to do with network latency. > I am now EXTREMELY happy with colinux. This is an awesome project! That's true, it is awesome, the network beeing the only roadblock. ------------------------------------------------------- This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more http://productguide.itmanagersjournal.com/guidepromo.tmpl _______________________________________________ coLinux-users mailing list coL...@li... https://lists.sourceforge.net/lists/listinfo/colinux-users |
|
From: peter g. <plu...@bi...> - 2004-10-01 13:22:52
|
nah colinux network performance is pretty poor in general and compiliation especilly the C way tends to hit the filesystem a lot > -----Original Message----- > From: col...@li... > [mailto:col...@li...]On Behalf Of David > Burela > Sent: 01 October 2004 14:18 > To: col...@li... > Subject: RE: [coLinux-users] performance > > > Ahh but samba support under *nix has always sucked hasn't it ;) > Colinux just inherited it ^_^;; > > - Dave > > > -----Original Message----- > From: col...@li... > [mailto:col...@li...] On Behalf Of > Holger Krull > Sent: Friday, 1 October 2004 9:31 PM > To: col...@li... > Subject: Re: [coLinux-users] performance > > > > My problem had to do with network latency. > > > I am now EXTREMELY happy with colinux. This is an awesome project! > > That's true, it is awesome, the network beeing the only roadblock. > > > ------------------------------------------------------- > This SF.net email is sponsored by: IT Product Guide on ITManagersJournal > Use IT products in your business? Tell us what you think of them. Give us > Your Opinions, Get Free ThinkGeek Gift Certificates! Click to > find out more > http://productguide.itmanagersjournal.com/guidepromo.tmpl > _______________________________________________ > coLinux-users mailing list > coL...@li... > https://lists.sourceforge.net/lists/listinfo/colinux-users > > > > ------------------------------------------------------- > This SF.net email is sponsored by: IT Product Guide on ITManagersJournal > Use IT products in your business? Tell us what you think of them. Give us > Your Opinions, Get Free ThinkGeek Gift Certificates! Click to > find out more > http://productguide.itmanagersjournal.com/guidepromo.tmpl > _______________________________________________ > coLinux-users mailing list > coL...@li... > https://lists.sourceforge.net/lists/listinfo/colinux-users |
|
From: Holger K. <hol...@gm...> - 2004-10-01 13:28:15
|
David Burela schrieb: > Ahh but samba support under *nix has always sucked hasn't it ;) No, works very well here. > Colinux just inherited it No, colinux has its specific network problems. It has serious latency problems (if you don't use bridgeing just ping host and colinux to see the difference). Every packet gets a time penalty. |
|
From: Jaroslaw F. (U. <jar...@li...> - 2004-10-01 14:04:29
|
On Fri, Oct 01, 2004 at 03:27:36PM +0200, Holger Krull wrote: > > No, colinux has its specific network problems. It has serious latency > problems (if you don't use bridgeing just ping host and colinux to see > the difference). Every packet gets a time penalty. > Hmmm, and, well, where to hunt the problem down? I'm pretty tired of connection phases to Samba on coLinux, MySQL from coLinux to host etc... BTW, I'm using bridged networking on Win2k, and coLinux 0.6.1. -- X Windows: Form follows malfunction. |
|
From: Holger K. <hol...@gm...> - 2004-10-05 16:51:23
|
Jaroslaw Fedevych (UALUG schrieb: > On Fri, Oct 01, 2004 at 03:27:36PM +0200, Holger Krull wrote: > >>No, colinux has its specific network problems. It has serious latency >>problems (if you don't use bridgeing just ping host and colinux to see >>the difference). Every packet gets a time penalty. > Hmmm, and, well, where to hunt the problem down? I'm pretty tired > of connection phases to Samba on coLinux, MySQL from coLinux to > host etc... BTW, I'm using bridged networking on Win2k, and > coLinux 0.6.1. First someone had to track down which part of the system is adding the latency to every packet. Winpcap, colinux driver or the colinux kernel itself. Can't be done without changeing the source and is therefore a little complicated. |
|
From: Chris D. <da...@ya...> - 2004-09-30 18:12:54
|
--- "Curtis, Craig" <Cra...@ec...> wrote: > memory. Is colinux just much slower than a standalone linux PC? My short answer is, no it should not be significantly slower from the perspective of the processor... In my experience. Longer answer: A few months back I did some speed comparisons between some code I wrote running under Windows, coLinux and native Linux. The program was a snippet of some AI code doing a mix of int and float math using linked lists and arrays. In any event, here is what I saw (qualatatively): - code compiled using intel's C compiler ran at pretty much the same speed under windows, coLinux and Linux. (dual boot machine) Other than the initial load of the program, there was no disk i/o. The variation in run times was only a couple percent at most. - also, code compiled using gcc ran at about the same speed under native Linux, coLinux and Cygwin. These did run significantly slower than the intel compiled binaries. This testing was under the 2.4 kernel. I have since moved to the 2.6 kernel but haven't seen significant slowdowns. As a side note, coLinux provides a nice way to run the free Linux version of Intel's C compiler under windows. The speed benefit over gcc CAN be HUGE for certian types of programs. --Chris ===== There is a very fine line between "hobby" and "mental illness." --Dave Barry Linux DVDs: http://www.LinuxDVDs.com personal pages: http://www.dahlweb.net |
|
From: Martin K. <ka...@po...> - 2004-09-30 19:25:26
|
Hi, there's just an issue, that not only CPU is heavily used when compiling. The only performance problem with CoLinux is the networking, not the throuput, but latency. Regards, Martin Chris Dahl wrote: > --- "Curtis, Craig" <Cra...@ec...> wrote: > >>memory. Is colinux just much slower than a standalone linux PC? > > > My short answer is, no it should not be significantly slower from the > perspective of the processor... In my experience. > > Longer answer: > > A few months back I did some speed comparisons between some code I wrote > running under Windows, coLinux and native Linux. The program was a snippet of > some AI code doing a mix of int and float math using linked lists and arrays. |
|
From: sophana <sop...@ya...> - 2004-10-05 16:05:27
|
Is there a plan of getting rid of that network performance problem? Maybe it is because of winpcap? I would like to use windows machines to compute for our cluster, that does lot of network activity fetching data to be processed from nfs. (later AFS I hope) network performance is definitly a big problem... Martin Kanich wrote: > Hi, > > there's just an issue, that not only CPU is heavily used when > compiling. The only performance problem with CoLinux is the > networking, not the throuput, but latency. > > Regards, > Martin > |
|
From: Nuno L. <lu...@nl...> - 2004-09-30 19:36:12
|
Curtis, Craig, dando pulos de alegria, escreveu : > I just wanted to ask about the performance of colinux. I am running > win2000 on a 2.6GHz PC. I set up a linux partition with debian to be > used with colinux. I also set the swap to a real swap partition. When > I go through to make a project, it is MUCH slower than a 400MHz PC. I > can look at my hard drive and see that the harddrive is not being > touched almost at all during compilation, so this tells me I have enough > memory. Is colinux just much slower than a standalone linux PC? What I like most with colinux is exactly it's speed. I compile the colinux kernel many times and don't notice any big difference in relation to a similar "native" linux machine I have at my side. I use Gentoo, and also don't notice any big differences, even when compiling those big packages like xorg or gcc (just updated both yesterday on both computers with no problems or very big differences). colinux is always slower than a native linux. It's just impossible to do better (we have to let windows run, don't we ;), but you shouldn't notice with normal command line usage. As a side note, I run "hdparm -tT" on the "native" system, with an old IDE 10GB disk (reiserfs) and got: * Cached reads: 360 MB/sec * Buffered reads: 18 MB/sec On a colinux "real" partition I got (with some errors ignored): * Cached reads: 340 MB/sec * Buffered reads: 11 MB/sec I don't think this are bad results, even if that partition is in a more modern IDE 7200 rpm, 40GB disk (ext2). Maybe you should run your own tests and post the results here, so we can help you more. Regards, ~Nuno Lucas |
Thanks for helping keep SourceForge clean.
X