http://www.linuxjournal.com/article/7438
So did anyone read this? Usually the comments end up being more interesting than the article. One thing that made me laugh about this one, was the mention of how difficult it is to build a "faster" system, than other distributions. Shoot, all you have to do is put -march=i686 in make.conf and the system you build is noticeably faster than 99% of the other distributions out there. I posted something to that effect.
When I tried Mandrake many many years ago, I was amazed at how much faster it was than RedHat, and it was almost identical except for being compiled for i586. I don't know why people continue to make light of the performance difference with Gentoo. Sure I think that compiling for pentium3 vs. pentium4 on a low end p4 system may not make that big a difference, but all around, is there any Gentoo user that really doesn't notice a difference from other distributions they have used, whether it's just in a few apps, or the whole system? I've even tried other distributions occasionally, since converting to Gentoo, and I'm always amazed at how slow they seem compared to Gentoo.
Changed to Is Gentoo Fast?!
Message
Changed to Is Gentoo Fast?!
Last edited by rbr28 on Sun Jan 09, 2005 5:28 am, edited 1 time in total.
Re: http://www.linuxjournal.com/article/7438
Most of my packages are compiled for i386/i486 and I don't notice much a difference from when I had Gentoo on this computer. I haven't run extensive testing, but when switching from Gentoo (i686 optimized) -> Slackware (i486 optimized) I didn't notice a speed decrease. All my programs loaded in the same time, at least as far as I could tell. I have a faster computer though, so that may be the reason why I don't see the optimization as much.rbr28 wrote:So did anyone read this? Usually the comments end up being more interesting than the article. One thing that made me laugh about this one, was the mention of how difficult it is to build a "faster" system, than other distributions. Shoot, all you have to do is put -march=i686 in make.conf and the system you build is noticeably faster than 99% of the other distributions out there. I posted something to that effect.
This may just be due to the placebo effect. (If not, it may be at least making the other distros seem so slow) Like I said above, I didn't notice a speed decrease when I was switching from Gentoo to Slackware. I've used Debian occasionally, and I didn't notice too much of a speed difference there either. Then again, I also know how to shut down processes that I don't need. It may also be that the programs on different distros were compiled with a different version of gcc. I've heard GCC 3.4.3 compiles much faster code than GCC 3.3.* I haven't tried it yet though. I think the selling point of Gentoo is more the USE flags than optimizations, because optimizations can be achieved relatively easily on other distros as well. Slackware has a Slackbuild (.build) file which is easy to modify, and integrates well into the package management system.rbr28 wrote:When I tried Mandrake many many years ago, I was amazed at how much faster it was than RedHat, and it was almost identical except for being compiled for i586. I don't know why people continue to make light of the performance difference with Gentoo. Sure I think that compiling for pentium3 vs. pentium4 on a low end p4 system may not make that big a difference, but all around, is there any Gentoo user that really doesn't notice a difference from other distributions they have used, whether it's just in a few apps, or the whole system? I've even tried other distributions occasionally, since converting to Gentoo, and I'm always amazed at how slow they seem compared to Gentoo.
- blackhorse
- Apprentice
- Posts: 225
- Joined: Sun Jul 11, 2004 3:30 pm
- Location: edge of a forest
Plus how much easier can it get for updating - ziggysquatch
- Apprentice
- Posts: 172
- Joined: Tue Nov 16, 2004 11:54 pm
- Location: /USA/Minnesota
Why is talk of the performance of our Gentoo machines a sin?
I have found a few things odd about discussions I have seen regarding the performance of Gentoo.
First, the Gentoo distribution is not itself something that speeds up or slows down a system, it merely provides the flexibility that makes it easy to build a highly optimized system.
To me, criticizing the speed of Gentoo is not criticizing the distribution at all, but making a statement about how Gentoo users build their machines, or more likely about all the other developers in the world. Why do I say that? If I build a system with gcc 3.4 and nptl, with relevant gcc optimizations, use sse and mmx optimizations, leaving out all unnecessary use flags, use a fast filesystem (reiser4 for example), use the fewest options necessary in a kernel with some nice performance patches, ...I could go on and on. All those things are easy in Gentoo, but they are not "Gentoo". If all those things are done, and the system is not any faster than another stock distro install, then what you are saying is that the work of the developers on all those projects is useless! Any idiot knows using sse and mmx optimizations won't speed up apache, but sse, mmx and 3dnow do make a significant difference in certain apps (gimp), just as so many of the other options you can build into our out of a gentoo system result in significant gains somewhere.
I also agree that for many users those differences are insignificant, or not worth the effort, but for many of us they are. People always seem to point out that Gentoo has so many other benefits, as if they have to cover up that Gentoo systems can be built highly optimized for speed. It's as if one can say that Gentoo is flexible, but if you say that flexibility allows you to build a faster, or more stable, or more secure system, then you are talking total nonsense. Very odd to me. If none of that is possible then what advantage is flexibility?
Are the critics going to say every gcc optimization, even the 3.4 compiler itself, results in no improvement? Is NPTL no improvement over linux threads? Do smaller binaries results in no gain in performance (i.e launching from a slow laptop disk for example, or using less memory)? Now I might believe some things do not result in huge performance gains, but I'm not going to believe that every developer working to write optimized code is failing. That's just nonsense. If you build a Gentoo system to take advantage of these improvements and many others, it defies logic that every one those enhancements together are going to result in no performance gain.
Sure I've seen versions of gcc that create worse code, or newer versions of apps that are worse in performance, but all this talk of how Gentoo is not faster than other distributions is as nonsensical as all the talk that it is. Anyone who really knows both Gentoo and a number of other distributions knows that it's easier to build a faster Gentoo system, than it is to build as fast a system with another distro. Sure in some apps it won't be noticeable, and you can get another distro very close, but again we are talking about the ease with which this can be done. If I took a stock Mandrake install and compiled all my apps on it, I'm sure I could get it as customized as a Gentoo install, but it sure would be a pain, and it would defeat the purpose of using Mandrake in the first place. I'm not saying performance is THE reason I use Gentoo, or even one of the top few, but the flexibility it allows me in creating an optimized system is significant.
If you think that people are not building Gentoo machines with outstanding performance, then you are really trivializing the work of gcc developers, chipmakers (intel's mmx and sse, and AMD's 3dnow), just about any application developer, kernel developlers, etc.
When I used Linux in the past, I always thought Windows was much faster. Every version of Windows seemed faster, even NT, 2000, and XP. Most of that was just desktop responsiveness type stuff, but a lot of it was the speed with which applications launched. Mandrake was the first distribution I ever used that seemed like it was catching up, and Gentoo is the first that has ever caught up in my experience.
The only thing that might ever resolve this debate is extensive benchmarking, and I doubt it's even worth the effort. Those who care will build their Gentoo systems the way they see fit, and those who don't will continue doing what they do. I thought of doing benchmarks myself, but when you get into it, it's really very complex and time consuming to do it right. It may be worth trying at work if I ever get the time. How about some of us starting a thread on running the ultimate performance comparison test?
Here's what I would start with.
1. Several identical machines (Only at work could I accomplish this first requirement)
2. A few distros in addition to Gentoo, I would take Mandrake, Fedora, Suse, and maybe 1 or two others.
3. Build each machine in a somewhat standard way, documenting how each was built. For most distros this would include whatever kernel is on the cd, building in only the applications you want to benchmark, disabling unused services, etc. For Gentoo I would want to build a few machines. Maybe 1 with pre-compiled binaries for the specific architecture, such as pentium4, 1 completely compiled from stage1 using numerous performance options as mentioned previously, and one built for i386 with almost no optimizations.
4. Document all of this extensively. For example, with Gentoo, document make.conf, build options, and special settings.
5. Compare specific scripted tasks in addition to application launch times, boot times, etc.
The hard part would be coming up with any kind of standardization in the build process. I don't think that is critical though. They idea is to compare how "typical" builds are in comparison to each other. If we needed to prove the significance of specific build options, we could simply build a gentoo machine from stage3, for i386, and then build an identical machine, with all the optimizations we wanted. That would probably be at least as worthwhile, if worth anything at all. I know none of that would ever convince anyone of anything, but I think it would be a good educational experience. Although the few performance benchmarks I see scattered around are a good start, I think they are lacking in completeness and any real attempt at a broad and fair comparison.
First, the Gentoo distribution is not itself something that speeds up or slows down a system, it merely provides the flexibility that makes it easy to build a highly optimized system.
To me, criticizing the speed of Gentoo is not criticizing the distribution at all, but making a statement about how Gentoo users build their machines, or more likely about all the other developers in the world. Why do I say that? If I build a system with gcc 3.4 and nptl, with relevant gcc optimizations, use sse and mmx optimizations, leaving out all unnecessary use flags, use a fast filesystem (reiser4 for example), use the fewest options necessary in a kernel with some nice performance patches, ...I could go on and on. All those things are easy in Gentoo, but they are not "Gentoo". If all those things are done, and the system is not any faster than another stock distro install, then what you are saying is that the work of the developers on all those projects is useless! Any idiot knows using sse and mmx optimizations won't speed up apache, but sse, mmx and 3dnow do make a significant difference in certain apps (gimp), just as so many of the other options you can build into our out of a gentoo system result in significant gains somewhere.
I also agree that for many users those differences are insignificant, or not worth the effort, but for many of us they are. People always seem to point out that Gentoo has so many other benefits, as if they have to cover up that Gentoo systems can be built highly optimized for speed. It's as if one can say that Gentoo is flexible, but if you say that flexibility allows you to build a faster, or more stable, or more secure system, then you are talking total nonsense. Very odd to me. If none of that is possible then what advantage is flexibility?
Are the critics going to say every gcc optimization, even the 3.4 compiler itself, results in no improvement? Is NPTL no improvement over linux threads? Do smaller binaries results in no gain in performance (i.e launching from a slow laptop disk for example, or using less memory)? Now I might believe some things do not result in huge performance gains, but I'm not going to believe that every developer working to write optimized code is failing. That's just nonsense. If you build a Gentoo system to take advantage of these improvements and many others, it defies logic that every one those enhancements together are going to result in no performance gain.
Sure I've seen versions of gcc that create worse code, or newer versions of apps that are worse in performance, but all this talk of how Gentoo is not faster than other distributions is as nonsensical as all the talk that it is. Anyone who really knows both Gentoo and a number of other distributions knows that it's easier to build a faster Gentoo system, than it is to build as fast a system with another distro. Sure in some apps it won't be noticeable, and you can get another distro very close, but again we are talking about the ease with which this can be done. If I took a stock Mandrake install and compiled all my apps on it, I'm sure I could get it as customized as a Gentoo install, but it sure would be a pain, and it would defeat the purpose of using Mandrake in the first place. I'm not saying performance is THE reason I use Gentoo, or even one of the top few, but the flexibility it allows me in creating an optimized system is significant.
If you think that people are not building Gentoo machines with outstanding performance, then you are really trivializing the work of gcc developers, chipmakers (intel's mmx and sse, and AMD's 3dnow), just about any application developer, kernel developlers, etc.
When I used Linux in the past, I always thought Windows was much faster. Every version of Windows seemed faster, even NT, 2000, and XP. Most of that was just desktop responsiveness type stuff, but a lot of it was the speed with which applications launched. Mandrake was the first distribution I ever used that seemed like it was catching up, and Gentoo is the first that has ever caught up in my experience.
The only thing that might ever resolve this debate is extensive benchmarking, and I doubt it's even worth the effort. Those who care will build their Gentoo systems the way they see fit, and those who don't will continue doing what they do. I thought of doing benchmarks myself, but when you get into it, it's really very complex and time consuming to do it right. It may be worth trying at work if I ever get the time. How about some of us starting a thread on running the ultimate performance comparison test?
Here's what I would start with.
1. Several identical machines (Only at work could I accomplish this first requirement)
2. A few distros in addition to Gentoo, I would take Mandrake, Fedora, Suse, and maybe 1 or two others.
3. Build each machine in a somewhat standard way, documenting how each was built. For most distros this would include whatever kernel is on the cd, building in only the applications you want to benchmark, disabling unused services, etc. For Gentoo I would want to build a few machines. Maybe 1 with pre-compiled binaries for the specific architecture, such as pentium4, 1 completely compiled from stage1 using numerous performance options as mentioned previously, and one built for i386 with almost no optimizations.
4. Document all of this extensively. For example, with Gentoo, document make.conf, build options, and special settings.
5. Compare specific scripted tasks in addition to application launch times, boot times, etc.
The hard part would be coming up with any kind of standardization in the build process. I don't think that is critical though. They idea is to compare how "typical" builds are in comparison to each other. If we needed to prove the significance of specific build options, we could simply build a gentoo machine from stage3, for i386, and then build an identical machine, with all the optimizations we wanted. That would probably be at least as worthwhile, if worth anything at all. I know none of that would ever convince anyone of anything, but I think it would be a good educational experience. Although the few performance benchmarks I see scattered around are a good start, I think they are lacking in completeness and any real attempt at a broad and fair comparison.
- /dev/random
- l33t
- Posts: 704
- Joined: Fri Nov 26, 2004 5:32 pm
- Location: Austin, Texas, USA
Gentoo feels faster than say Fedora Core or Redhat but that's because when you install Gentoo you wind up deciding what services you need to run unlike a Redhat or Fedora install where every service imaginable starts up. If I had the same services enabled on Fedora then I don't know if it would be that much slower or not. I imagine Debian, Slackware, LSF, Arch, etc. would be like this as well.
Ok, I just finished installing Gentoo Stage1 on my Mom's computer.
Its a P3 450mhz with 128mb ram and a crappy S3virge gfx card.
Not exactly a speed demon.
Before I installed gentoo, I had installed Suse9.0 for her.
Let me just say, the difference is AMAZING!
Seriously. With Suse, the hd was swapping a lot for every little app or window you openend. It was ok for her, since she only used email and liked to play solitaire. But for me, this would have made the computer unbearable to use.
Now I installed gentoo from stage 1 and also installed kde.
Amazing! I dont know if its the processor specific optimisation, the change from 2.4 to 2.6 kernel or the lack of unecessary services, or a everything in total, but DAMN! This is a definate improvement.
I am VERY impressed. I never expected something as hefty as kde to run so smooth on this hardware.
You can really tell the difference with mundane things such as moving windows, resizing, using scrollbars, switching from one program window to another, browsing your hd in konqueror,etc.
Its a P3 450mhz with 128mb ram and a crappy S3virge gfx card.
Not exactly a speed demon.
Before I installed gentoo, I had installed Suse9.0 for her.
Let me just say, the difference is AMAZING!
Seriously. With Suse, the hd was swapping a lot for every little app or window you openend. It was ok for her, since she only used email and liked to play solitaire. But for me, this would have made the computer unbearable to use.
Now I installed gentoo from stage 1 and also installed kde.
Amazing! I dont know if its the processor specific optimisation, the change from 2.4 to 2.6 kernel or the lack of unecessary services, or a everything in total, but DAMN! This is a definate improvement.
I am VERY impressed. I never expected something as hefty as kde to run so smooth on this hardware.
You can really tell the difference with mundane things such as moving windows, resizing, using scrollbars, switching from one program window to another, browsing your hd in konqueror,etc.
- /dev/random
- l33t
- Posts: 704
- Joined: Fri Nov 26, 2004 5:32 pm
- Location: Austin, Texas, USA
- zephyr1256
- Apprentice
- Posts: 170
- Joined: Mon Mar 10, 2003 10:08 pm
- Location: Kingsport, TN
I said significantly faster, but the main reason has little to do with compiler optimizations and everything to do with me having exactly what I need with each and every program I want to run. I'm comparing to my old distro Mandrake 9(this was about 2 years ago). In some cases, its a difference of Gentoo = things work and Mandrake = things didn't work(effective speed 0). In other cases, its just the advantage of customization via use flags, and the extent of that advantage varies from application to application.
The Congress shall have power...To promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries; --U.S. Constitution. Article 1, Section 8.
- Q-collective
- Advocate
- Posts: 2077
- Joined: Mon Mar 22, 2004 5:21 pm
Hey that was my last Windows OS. Took me years before I would upgrade from 2000 (Cleartype finally got me). I thought it was actually quite good in comparison to the phenominal Windows 2000 and the aweful 9x line and the even worse XP--as Windows OS' go that is...no comparison with Linux.Oktane wrote:Is it faster? My previous "distro" was Windows Server 2003
Honestly the last distro I ran before Gentoo was FC2 and it was damn fast. Suse was junk, Mandrake was actually quite OK but FC2 really impressed me. Gentoo is just so snappy and solid. All me previous binary distros felt "buggy". Things crashed to often and it just didnt feel solid, not even compared to Windows. Programs on Mandrake crashed a lot and the other binary distros werent all that hot either. I had problems with a few things and since they are all so guified, I didnt even know where to begin. Another plus on Gentoo is you just gotta hack the proper config file.
I have no documentation or tests, but I would say Gentoo is by far the most stable and solid Linux distro I have used so far.
- /dev/random
- l33t
- Posts: 704
- Joined: Fri Nov 26, 2004 5:32 pm
- Location: Austin, Texas, USA
The only thing that didnt work was one of my CD drives at the time. Other than that, everything else went great. I heard FC3 is supposed to be a great improvement on both FC and FC2. If it wasnt the for the problems with my drive, I might still be using it...hmmm, naaa, I doubt that/dev/random wrote:What the hell Fedora Core 2 cooperated for you? I'm sure I just paid the price for updating to FC2 the day it came out. It was so buggy I missed FC1 but rather than go Slackware or back to FC1 I just came on over to Gentoo.
But yeah, didnt have any problems really other than that.- /dev/random
- l33t
- Posts: 704
- Joined: Fri Nov 26, 2004 5:32 pm
- Location: Austin, Texas, USA
I had problems with random crashes (before I started installing anything exotic or tweaking). Really it felt like Windows 98 all over again. Then, once I started installing other programs via apt4rpm (once the rpm's were finally released) everything slowed to a crawl and more random crashes happened. Then I did a fresh install with the same options and settings and all was okay but there just weren't enough rpm's out and it was always a pain to hunt down the rpm's to get the development versions of programs so I could build stuff from source.
Thats what I truly love about Gentoo and Portage. The latest is always available and if it doesnt get put in Portage, you can easily make an ebuild. I found all binary distros where annoying to update programs. Most were months old and if you were lucky enough to find the latest versions, you still had to deal with the dependency hell of RPMs. Though FC2 had a pretty up to date availability compared to Mandrake and Suse./dev/random wrote:I had problems with random crashes (before I started installing anything exotic or tweaking). Really it felt like Windows 98 all over again. Then, once I started installing other programs via apt4rpm (once the rpm's were finally released) everything slowed to a crawl and more random crashes happened. Then I did a fresh install with the same options and settings and all was okay but there just weren't enough rpm's out and it was always a pain to hunt down the rpm's to get the development versions of programs so I could build stuff from source.
I was lucky enough never to have run a Windows 9x. Brought up on NT, then to 2000 (Ran that baby from Beta3), then to 2003 Server and finally to Linux. Honestly I just got bored with Windows...thats the main reason I went looking elsewhere...glad I did
What you describe FC2 to have been for you, perfectly exemplifies Suse on my box...still gives me nightmares.
I've never found Slackware to be hard to update packages. If the latest version of software isn't in the package browser, you can either make a Slackbuild file, or compile it yourself and use checkinstall to make a package, (which is actually easier than making an ebuild file) I've never experienced "dependency hell" on Slackware... On RPM distros, when using unofficial packages, yes, but not otherwise.Lokheed wrote:Thats what I truly love about Gentoo and Portage. The latest is always available and if it doesnt get put in Portage, you can easily make an ebuild. I found all binary distros where annoying to update programs. Most were months old and if you were lucky enough to find the latest versions, you still had to deal with the dependency hell of RPMs. Though FC2 had a pretty up to date availability compared to Mandrake and Suse.
- PapaSMURFFS
- n00b
- Posts: 53
- Joined: Thu Sep 23, 2004 4:05 pm
- Location: Toronto, Canada
- ChrisWhite
- Retired Dev
- Posts: 399
- Joined: Thu Jul 08, 2004 7:19 pm
- Location: Stockton, CA
- Contact:
I really think that some people take the perspective of Gentoo from a somewhat tunneled vision. Gentoo really isn't JUST about speed, it's about a few more things:
<pr mode
>
1) First off, it's really about choice. While prepackaged binaries choose configuration flags for you, you can choose what gets included and what doesn't, expecially in server situations where you may not need certain functionality. Big example I can think of is being able to build a headless server and getting all your packages without X support. Think of how much blot THAT saves! Also you can save a bit more by taking out ipv6 support (I've done that myself).
2) It's about extendability. The ebuild systems is a nice and easy way to ensure the somewhat tedious proccess of compiling is somewhat simpilified. With a basic script, one can easily keep consistent installation for quite a large number of packages with basic configuration.
3) It's about knowing where your files are! With Gentoo keeping track of files, it makes uninstalling a LOT easier. That nice little revdep_rebuild script also helps to rebuild stuff that was using that support so stuff doesn't get broken. I myself like to go source-y sometimes, and that whole "make uninstall thing"? Yah, I won't touch that with a ten foot pole now.
4) It's about centralized packaging. Packages are kept on a centralized server and then propagated to our other mirrors. By doing this, you don't have to hunt around that hard to find package x, whereas back in my redhat days, I remember having to abuse rpmfind and even then getting mixed results...
That's my 2 cents. Gentoo is more than just speed. Hopefully people will better realize that.
</pr mode>
<pr mode
>1) First off, it's really about choice. While prepackaged binaries choose configuration flags for you, you can choose what gets included and what doesn't, expecially in server situations where you may not need certain functionality. Big example I can think of is being able to build a headless server and getting all your packages without X support. Think of how much blot THAT saves! Also you can save a bit more by taking out ipv6 support (I've done that myself).
2) It's about extendability. The ebuild systems is a nice and easy way to ensure the somewhat tedious proccess of compiling is somewhat simpilified. With a basic script, one can easily keep consistent installation for quite a large number of packages with basic configuration.
3) It's about knowing where your files are! With Gentoo keeping track of files, it makes uninstalling a LOT easier. That nice little revdep_rebuild script also helps to rebuild stuff that was using that support so stuff doesn't get broken. I myself like to go source-y sometimes, and that whole "make uninstall thing"? Yah, I won't touch that with a ten foot pole now
.4) It's about centralized packaging. Packages are kept on a centralized server and then propagated to our other mirrors. By doing this, you don't have to hunt around that hard to find package x, whereas back in my redhat days, I remember having to abuse rpmfind and even then getting mixed results...
That's my 2 cents. Gentoo is more than just speed. Hopefully people will better realize that.
</pr mode
>Thats great. I never used Slack before and dont know where this big defense in the name of S came from...actually I havent even seen Slack being mentionned in this thread...Nightgrave wrote:I've never found Slackware to be hard to update packages. If the latest version of software isn't in the package browser, you can either make a Slackbuild file, or compile it yourself and use checkinstall to make a package, (which is actually easier than making an ebuild file) I've never experienced "dependency hell" on Slackware... On RPM distros, when using unofficial packages, yes, but not otherwise.Lokheed wrote:Thats what I truly love about Gentoo and Portage. The latest is always available and if it doesnt get put in Portage, you can easily make an ebuild. I found all binary distros where annoying to update programs. Most were months old and if you were lucky enough to find the latest versions, you still had to deal with the dependency hell of RPMs. Though FC2 had a pretty up to date availability compared to Mandrake and Suse.
I haven't bothered to measure anything, but since the previous distro I used was SuSE I've experienced performance boosts merely from not having a system full of useless bloat I never use. One particular area of improvement was boot time; the Gentoo system I have now is faster booting by far than any other distro I've used.
Last edited by Imek on Mon Jan 10, 2005 5:37 pm, edited 1 time in total.
I don't know if I notice much of a speed difference between Gentoo and my previous distro, Mandrake.
But my main reason for switching was bloat and dependency problems.
With Gentoo I can pick the options and customize all I want without fear of many problems. I can try newer packages (aka ~x86) easily and don't have the massive problems I ended up with trying bleeding edge stuff with Mandrake. Always ended up with leftover files that caused problems when downgrading back to original system when problem occurred. With Gentoo, just unmerge ~x86 and emerge x86 and back to normal. I have never had a problem restoring the system.
Gentoo for me has been rock stable too. More stable than any OS I have used and I push it with virtual machines, multiple compiling, multimedia apps, tv, graphics, all at the same time.
But my main reason for switching was bloat and dependency problems.
With Gentoo I can pick the options and customize all I want without fear of many problems. I can try newer packages (aka ~x86) easily and don't have the massive problems I ended up with trying bleeding edge stuff with Mandrake. Always ended up with leftover files that caused problems when downgrading back to original system when problem occurred. With Gentoo, just unmerge ~x86 and emerge x86 and back to normal. I have never had a problem restoring the system.
Gentoo for me has been rock stable too. More stable than any OS I have used and I push it with virtual machines, multiple compiling, multimedia apps, tv, graphics, all at the same time.
- HydroSan
- l33t
- Posts: 764
- Joined: Thu Mar 04, 2004 5:43 pm
- Location: The Kremlin (aka Canada)
- Contact:
The only way Gentoo feels 'faster' is because I can specify which services to run when, without having to hack at configurations just to get things working. I can have the most stripped-down or loaded-up installation I want, just by going 'emerge package1 package2 etc'.
It's also flexible. All those security options such as GRSec, PaX, SELinux and PIC/PIE give me a sort of peace of mind.
I also work with Debian. Debian is okay. But it DOES feel old, even with Sarge. Some sections don't feel as clean or intuitive, but it gets the job done nicely.
SuSE used to be my favourite, back in the 9.0 Pro days. Now SuSE seems to have fallen out of grace.
It's also flexible. All those security options such as GRSec, PaX, SELinux and PIC/PIE give me a sort of peace of mind.
I also work with Debian. Debian is okay. But it DOES feel old, even with Sarge. Some sections don't feel as clean or intuitive, but it gets the job done nicely.
SuSE used to be my favourite, back in the 9.0 Pro days. Now SuSE seems to have fallen out of grace.
I was a Gangster for Capitalism, by Major General Smedley Butler.
Server status: Currently down, being replaced with fresh install - 20% completed.
Server status: Currently down, being replaced with fresh install - 20% completed.
I had a very similar experience in August 2003, when I first started to use 2.6 instead of 2.4. The swappiness problem almost completely went away thanks to the new I/O layout.ChojinDSL wrote:Ok, I just finished installing Gentoo Stage1 on my Mom's computer.
Its a P3 450mhz with 128mb ram and a crappy S3virge gfx card.
Not exactly a speed demon.
Before I installed gentoo, I had installed Suse9.0 for her.
Let me just say, the difference is AMAZING!
Seriously. With Suse, the hd was swapping a lot for every little app or window you openend. It was ok for her, since she only used email and liked to play solitaire. But for me, this would have made the computer unbearable to use.
Now I installed gentoo from stage 1 and also installed kde.
Amazing! I dont know if its the processor specific optimisation, the change from 2.4 to 2.6 kernel or the lack of unecessary services, or a everything in total, but DAMN! This is a definate improvement.
I am VERY impressed. I never expected something as hefty as kde to run so smooth on this hardware.
You can really tell the difference with mundane things such as moving windows, resizing, using scrollbars, switching from one program window to another, browsing your hd in konqueror,etc.
Last edited by etnoy on Mon Jan 10, 2005 8:36 pm, edited 1 time in total.
The md5sum of the above post is 06280ccd85ef9deb49c336e7945f4b5c
God is dead! - Nietzsche
Nietzsche is dead! -God
God is dead! - Nietzsche
Nietzsche is dead! -God