2.4 vs 2.6 Linux Kernel Shootout

FyRE666 writes "Infoworld are currently running an interesting comparison of the 2.4 series kernel against the new 2.6 release on Xeon, Opteron and Itanium CPUs with some surprising benchmark results for common server-related tasks. Basically the new scheduler helps the 2.6 kernel to cream the old 2.4: Samba tests showing up to 73% speed increases, MySQL showing up to 29% and Apache serving dynamic content up to 47% faster!"

Comments

Error

[popa]

tried to get this in before you posted it... but dynamic only went up 22% for apache.... static went up 47%

Re:Error

[t0ny]

or maybe you are more like the gay guy who everybody beat up in high school, and posting to things that t0ny writes is the only thing a passive-agressive loser like yourself is capable of.

Re:Error

[t0ny]

I could care less what online comic artists make. It obviously must make some kind of money, because they are still doing it after what, four or five years?

However, I just found the comic itself both funny and relevant. Because we both know the only things which really have any value are ideas, right? Im going to buy a pizza right now with some ideas.

And SCO shows...

[Black Parrot]

...a stunning 129% increase on SPEClawsuit!

Re:And SCO shows...

[Dreadlord]

how generous the cool guys at SCO are, we can get all those goodies by upgrading to 2.6, without any additional fees.

Okay, so that's with the latest and greatest..

[spitefulcrow]

But how much of an improvement does it get on older hardware and/or software packages?

Re:Okay, so that's with the latest and greatest..

[iabervon]

On older hardware, it isn't noticably better. But that's just because it was already fine on such hardware. For such hardware, the main interesting thing is that the CPU scheduler does a better job of recognizing tasks which need a bit of processor time at precise intervals, so that you can play audio and video without skips while under more total load.

It should also help with the ~5ms delay when you interact with the system, so that it "feels better", regardless of the actual system speed.

But these aren't things that are easy to benchmark, anyway. (At least, unless you're writing code to poll a sensor at precise intervals)

Re:Okay, so that's with the latest and greatest..

[mcbridematt]

I tried 2.6.0-test8 when I had my Dual PII box a few months ago. Performance difference for KDE between it and my 800@920MHz spitfire Duron was pretty much unnoticeable. Recommended for older boxes.

Also, for a more current comparison:

distributed.net rc5-72 Duron 1.4@ 1.9 (2400+) w/256kb l2 cache mod:

Windows: 5,500-6,000/sec
Linux 2.4.20-debian: 3,400-3,700/sec
Linux-2.6.1-rc1: 5,500-6,000/sec.

The upgrade to Linux-2.6 is worth it.

Re:Okay, so that's with the latest and greatest..

[JDWTopGuy]

You should be getting a few million keys/sec, not 5,500 keys/sec. Perhaps you made a mistake? I get 4.5 million keys/sec on my XP 1600+ running 2.4.23.

Maybe I'll give 2.6 a try, it's worth it for faster RC5... :D

Re:Okay, so that's with the latest and greatest..

[Anonymous Coward]

The 2.6 kernel has made one of my desktop machines a pain to use. Music skips, windows don't refresh fast, etc. This is on a memory starved system, so the VM is getting heavily utilized. 2.4 provided a much more responsive desktop on this machine, an AMD1600 with 256M ram. I am not sure why this occurred, and haven't had time to investigate it...

Re:Okay, so that's with the latest and greatest..

[Anonymous Coward]

Following up to my own post. I just did some quick checking and this is a known issue. Apparently the 2.6 VM sucks when it's doing heavy swapping with multiple processes competing for CPU time. There are a number of posts on LKML about it; see the thread "poor swap performance on low end machines". The quick summary is that 2.4 gives much better performance on machines that swap a lot.

Re:Okay, so that's with the latest and greatest..

[SiliconJesus101]

Hmm, here's my cute lil' Powermac G4 1.2Ghz. This puppy hauls ass in math. Beats x86 CPU's that are almost twice the clock speed. I wonder how those fancy new dual G5 machines do.

dnetc v2.9007-486-CTR-03110414 for Mac OS X (Darwin 7.2.0). Please provide the *entire* version descriptor when submitting bug reports.
The distributed.net bug report pages are at http://www.distributed.net/bugs/
[Feb 01 08:35:57 UTC] Automatic processor type detection found a PowerPC 7455 (G4) processor.
[Feb 01 08:35:57 UTC] OGR: using core #0 (GARSP 5.13 Scalar).
[Feb 01 08:36:16 UTC] OGR: Benchmark for core #0 (GARSP 5.13 Scalar) 0.00:00:16.62 [14,449,071 nodes/sec]
[Feb 01 08:36:16 UTC] RC5-72: using core #4 (KKS 7450).
[Feb 01 08:36:35 UTC] RC5-72: Benchmark for core #4 (KKS 7450) 0.00:00:16.92 [12,570,753 keys/sec]

drugs are bad

[POds]

Okay, who's been feeding 2.6 speed?

Re:drugs are bad

[Rosco P. Coltrane]

Okay, who's been feeding 2.6 speed?

IBM, according to SCO.

Re:drugs are bad

[requim]

mod parent up funny..

Re:drugs are bad

[ultrabot]

Okay, who's been feeding 2.6 speed?

It was widely considered a smart move - feeding it crack would have further aggravated SCO, on the basis of Linux using their methodologies and ideas to bring enterprise functionality to the kernel.

May try it then

[Anonymous Coward]

I was wondering about upgrading to 2.6 from 2.4 with XFS on my box, with the improvements to SCSI support and the CPU speed ups it sounds promising :D

Then again BSD is very nice on the same hardware. Wonder how 2.6 linux & (free)BSD compare for those tasks.

Re:May try it then

[Doomdark]

Of course, the 5.2 schedULEr was well thought out and designed before being coded and will probably last quite a while. The linux 2.6 one will probably need major patches for 2.8 and will need to be rewritten for v3.

Care to elaborate what is wrong with linux 2.6 scheduler design (or implementation)? And in comparison what is elegant about FreeBSD's?

Re:May try it then

[MikeCapone]

One thing you can do to speed up your system is to dump XFS.

Man that thing chews CPU like nothing else (for a slight gain in file I/O, yes).

I've been interested in XFS for a while but haven't taken the plunge yet; care to elaborate a bit on what you say?


On what kind of CPU does the difference shows? Recent stuff or only on old hardare?

Thanks,

Re:May try it then

[Nothinman]

The 2.6 Linux process scheduler is O(1) meaning it'll scale to as many CPUs as you want and it's NUMA aware already.

I wouldn't be surprised if the FreeBSD scheduler was O(1), hell even NT's scheduler is O(1), but is it ready for NUMA or HT CPUs yet?

Re:May try it then

[anthonyrcalgary]

Try them both to see which one is faster/you like more. And don't listen to anyone who tells you not to try the alternatives, they're probably zealots.

Don't let my sig fool you, I like Gentoo better on the desktop. It's just that little bit more convenient to get the latest everything.

Some stuff I can tell you:

• With ULE enabled on FreeBSD 5.2, I can't tell the difference. No formal benchmarks, but I did everything I could to expose stuff that might annoy me later on. Dragging windows around really fast with "display content in moving windows" on and music playing. The sort of thing that would bring Linux 2.4 to its knees easily
• FreeBSD 5.2 has some annoying things. Like you've got to manually enable the drivers for sound devices and USB mice. And you've got to recompile the kernel to get ext2 support. It's pretty easy and well documented, but you can't just install and have a working system with no additional work.
• Gentoo didn't (for me) have any major issues. I mean, compile times can be a bitch, but you'll be doing that on FreeBSD eventually too.
• FreeBSD can do some really nifty stuff that Linux can't. Like jails.

My advice is to wait until FreeBSD 5 hits STABLE and Linux 2.6 matures a little bit, install both, and see what's more fun.

Re:May try it then

[anthonyrcalgary]

ULE has optimizations for HT/SMT and NUMA.

http://www.chesapeake.net/~jroberson/ULE.pdf

And FreeBSD also has some stuff that implicitly takes advantage of multiple CPUs, like a multithreaded network stack.

Re:May try it then

[drinkypoo]

You don't need to suffer the compile time shuffle with Gentoo if you download/otherwise acquire both CDs of Gentoo 1.4, and install the GRP, aka Gentoo Reference Platform. GRP users also get a bunch of tools to make things like building a kernel easier - it produces a kernel with the same sort of functionality as the one on the install CD, but that one is based on xfs-sources, and you can use whichever source tree you prefer.

As for jails, well, user mode linux, initial ramdisk fs, loopback filesystem. You get a whole separate operating system. But, jails are cool.

Re:May try it then

[Mr Z]

Keep in mind that ULE is derived from Linux 2.6's scheduler. Take a look at the last sentence of Section 2 of the BSD white paper on UL.

--Joe

Re:May try it then

[Mr Z]

ULE is becoming the default scheduler in -current.

Of course, things may have changed since then. I don't follow BSD all that closely, I just hang out on KernelTrap.

--Joe

Re:May try it then

[Mr Z]

It's definitely a bigger hammer. I wouldn't call it more elegant, though.

Re:May try it then

[anthonyrcalgary]

Yes, I know. Linux's scheduler is very good, and FreeBSD builds on it. I'm sure they'll exchange ideas more than once before it settles down. The issue I'm talking about is the locking, which is more finely grained in FreeBSD 5. It's possible to have more system calls in flight at once, especially with things that are critical to performance, like the network stack.

Re:May try it then

[Mr Z]

Hold on... Last I knew (which was awhile ago), FreeBSD (even in 5.x) still has more instances of the Big Kernel Lock/Giant Lock than Linux does. As I recall, Linux fully threaded their networking stack back in the 2.4 days.

I know earlier on in the 5.x series, Matt Dillon and others acknowledged that FreeBSD was at least a year behind Linux in their SMP support. It's been a couple years since then, but Linux hasn't sat still. It appears BSD hasn't either. Looking at FreeBSD's SMP status page, it appears that they've made a lot of progress. Their networking stack (at least, that portion of which that is called out as line items in a table) is nearly fully threaded. Coolness.

--Joe

Re:May try it then

[bill_mcgonigle]

FreeBSD can do some really nifty stuff that Linux can't. Like jails.

I'm curious, how are FreeBSD jails different than a chrooted process or a jailshell under Linux?

Re:May try it then

[anthonyrcalgary]

Jails basically create a complete virtual computer with less overhead than a user space linux kernel. Also, a jail keeps processes from using any network address except for a specified address, and a private loopback address. The jailed process can't see processes outside, and couldn't send signals to them if it could. Basically, a jail can be treated as a completely seperate computer that shares some hardware. It'll have (if configured correctly) its own sshd, accounts, and so on. You can even install the gnu tools and use the linux compatability features to have an almost-linux box inside your FreeBSD box.

chroot can be excaped by a user with root privileges, or by an executable with root privileges or that is suid root. chroot does not block system calls, but jails do.

http://www.freebsd.org/doc/en_US.ISO8859-1/books /a rch-handbook/jail.html

Impressive

[mgv]

These are impressive improvements.

Its actuallly hard to believe that there is that much more improvement to be gained - it will leave the microsoft servers even further behind as I don't think that they are improving their kernel that fast.

One question:

Does this mean that we can see improvements in low end systems for desktop use, or is the benefit only for servers. Because if this helps low end machines, it extends further the number of machines that can move from (say) win 98 to a real OS, whose hardware has long been abandoned by microsoft.

Michael

Re:Impressive

[JebusIsLord]

well i know the preemtive and low-latency changes (while available as patches to 2.4) are included in 2.6, and make a significant difference to GUI responsiveness.

Re:Impressive

[cide1]

Alright, let me be the first to say: WTF? Sorry big guy, I dont know if your trolling, dont speak english, or just have no idea what your asking, and cant be bothered to fix your broken keyboard.

With that said, a kernel that provides more performance will allow a slow machine to do more work in a said time period. It should be noted than none of these tests said anything about XFree86 performance, and that is the number one thing that will make a slow machine "feel" faster.

Re:Impressive

[Lussarn]

I'm just done with the transition of my home box. Took a few hours to get everything up, needed patches for lirc support. But it seems very well. UT2003 runs like a god with zero hickups. The thing I like the most is the new build system in 2.6, much better checks of what need to be compiled. To insert a new thing in the kernel takes about 20sek compile compared to probaly 5-10 minutes on 2.4 on my XP1800+.

Re:Impressive

[zmooc]

I can now watch divx on my pII-233 with it. Xmms never skips and X remains useable under high loads. It's an incredibly huge difference to 2.4. But I only noticed when I had to go back to 2.4.22 to try out openmosix. It was absolutely annoyingly slow compared to 2.6.

Re:Impressive

[the arbiter]

Nice try, thanks for playing. It's a troll. It hits every article on /.

Re:Impressive

[Serveert]

The pre-emptible kernel(you must enable this, it's off by default) will blow your mind when using a GUI. It is beyond impressive, it's like you're neo and you can see the mouse pointer before it gets to where you move your mouse.

You can grep your hard drive many times yet still browse the web with no slow down.

Re:Impressive

[GreatDrok]

The main difference I am seeing between running 2.4 and 2.6 on my system (Fedora Core 1 Athlon 2200+ 512MB) is that with 2.6 the system never seems to get bogged down.

I am able to continue using the system normally even when disc intensive tasks are running, something I could never do with 2.4. For example, try running updatedb as root and see what that does to your GUI responsiveness. Do the same with 2.6 and you will be hard pressed to even notice.

This is an enormous improvement and will likely make all the difference in the world to the use of Linux on the desktop. This machine now feels more responsive running Linux/X11 than it does with WindowsXP, a major milestone. There were always some things that Linux was fast at but a responsive GUI wasn't one of them. I would be interested to see how this works on older hardware, boxes that had only been able to run Windows98 for example should now be able to run a brand new OS. Cool!

Re:Impressive

[xcomm]

>>>Does this mean that we can see improvements in low end systems for desktop use, or is the benefit only for servers. Because if this helps low end machines, it extends further the number of machines that can move from (say) win 98 to a real OS, whose hardware has long been abandoned by microsoft.

Unfortunataly not!

I personally have not made the same exiting experience about the new 2.6.x kernels spead on my old PII 400 and 320MB (normally relatively fast due SCSI and 3Com TCP support). Especially if you have open some windows it reacts really harder than under 2.4. Different is sometimes only, that it seems to suggest you a better reaction, which is like the new scheduler is itnended to work.

So I'm running 2.6.x at home, but it's not looking to be the fastest kernel on small consumer gear. It looks designed for the big iron in the enterprise or at home. All other suggestions looking to me like propaganda.

Greetings from a GNU/Linux geek!

Re:Impressive

[fucksl4shd0t]

"I like my women like I like my coffee...ground up and stuffed into a plastic bag, which I keep in the freezer"

I like my women like I like my tea. Boiled and with sweetness added in measured amounts.

Re:Impressive

[Mr Z]

Apparently, the VM tuning in Linux 2.6 is geared towards machines with 128MB of RAM or more. (Sorry I don't have a link handy.) So, I'd say there's a definite cut line of old vs. new for 2.6.

--Joe

Linux in cache?

[SHEENmaster]

What I'd like to see is Linux that could run entirely within cache on the higher end chips. Even dated UltraSparcII chips can have up to 8M/cache. That's 64M in an 8-way box, allowing for some truly awe-inspiring performance on mathematical problems if RAM is ignored.

I haven't looked into sparc assembly enough to know if this is possible.

Re:Linux in cache?

[Aliencow]

Maybe you should try out FreeDows ;-)

Re:Linux in cache?

[ParisTG]

Caching is controlled completely by the CPU, transparent of the programmer.

Assuming that the kernel is the only code running, and it is small enough to fit into cache, then it will get there eventually.

However, it would make no sense to keep the entire kernel in cache, since most of that code isn't used most of the time. Also, application software is running at the same time, which needs to be cached as well.

In other words, just trust the CPU. It knows what it's doing :).

Re:Linux in cache?

[be-fan]

Why would I want my AmigaFS modules in cache? Or anything that touches something slow like a hard-drive? The whole point of cache is so the code that gets used the most stays in cache. If certain kernel code gets used often enough, it'd stay in the cache.

Re:Linux in cache?

[SHEENmaster]

This would not be for general use, it would be for running one specific, time-critical, application. If I'm going to sacrifice my 4G of ram to run everything in cache, it would not be to keep AmigaFS in my kernel.

Re:Linux in cache?

[runderwo]

Caching is controlled completely by the CPU, transparent of the programmer.

Only in some designs. Architectures like MIPS allow for both a cache-coherent or non-cache-coherent design. In a non-cache-coherent design, the cache is not transparent, and the kernel programmer is responsible for cache management; marking pages as dirty, flushing cache, etc. These designs are significantly more difficult to program and are present on some SGI machines, making porting to those machines a significant task.

Theoretically, higher performance can be achieved in a non-cache-coherent design, since the programmer would ostensibly know more about which data is most frequently used on his system and be able to customize his kernel for that. Also, it requires less glue logic on the board. However, the intent may be thwarted if the programmer doesn't have all the documentation (or skills) necessary to make efficient use of a software controlled cache.

Re:Linux in cache?

[calidoscope]

There was some guy posting to comp.arch by the name of Linus Torvalds saying that 10 MB was more like it for cache size to do what you're proposing. He was making noises about taking a break after 2.6 was released, starting work on 2.7 and needing an avenue to take out his aggressions.

The CDC-6600 had the equivalent of 1 MB of main core with maybe 10X that amount in extended core storage. Other examples of big iron from that era had similar sizes of main core and extended core storage.

Re:Linux in cache?

[dnoyeb]

Cache is not just faster memory. Just as there are things that might as well be on the hard drive and not in memory, there are things that do not need to be cached.

cache captures instructions, and sometimes data too, not programs. Its not meant for large routines.

Re:Linux in cache?

[SuperFrink]

If you had a larger cache than program code and memory requirements (and I guess I/O buffers) then you _could_ get everthing into a CPU level cache. That said caches are typically much smaller than the actuall RAM size so to fit the memory contents into cache a hashing algorithm is used. (eg use some of the bits from the memory address to index into the cache.)

Because of the hashing you might still get collisions, it depends, but when you get a collision then the "old" contents already in the cache have to be written back to memory or disk (if they are different then the original copy aka "dirty") or they could just be written over. This is why using a cache can possibly give you really bad performance if things in your code happen to occupy the same cache line in say an often repeated loop. (They keep bumping each other out of cache.)

Your idea leads to another interesting idea; what if the OS got it's own chunk of fast memory that the CPU could access quickly, possibly on a second bus. This relates to the LinuxBIOS project and something called a Harvard architecture (first link I found).

Re:Linux in cache?

[lars_stefan_axelsson]

Caching is controlled completely by the CPU, transparent of the programmer.

Not really true these days. Some architectures do have an explicit 'cache preload' instruction, such as the SPARC V9 and the ARM9E to mention two. These allow the programmer to preload a D-cache line before it is needed.

As the speed of the CPU has increased much faster than the speed of main memory (and hence increasing the relative cost of a cache mis), compiler based techniques to emit cache preload instructions in advance, before the data is actually needed, has been the subject of some research in the past 7-8 years. The main reason to do it is software, instead of hardware, is that the compiler have a greater knowledge of the layout of the entire task, as it can 'look ahead' in the source code. The main disadvantage is that any static analysis, of course won't have access to dynamic (run time) data about the program as it is running.

If you wish to go further, you could do worse than to start with my former colleague Magnus Karlsson's PhD thesis on the subject:

Magnus Karlsson: "Data Prefetching Techniques Targeting Single and a Network of Processing Nodes". Ph. D. thesis, Department of Computer Engineering, Chalmers University of Technology, December 1999.

And of course as always, Google and citeseer is your friend.

In other words, just trust the CPU. It knows what it's doing :).

Well, actually, it doesn't these days... :-) Trust the compiler instead :-) (Yeah, right).

Re:Linux in cache?

[SuperFrink]

using a cache can possibly give you really bad performance if things in your code happen to occupy the same cache line in say an often repeated loop.

Not to say you'll get worse performace by using a cache but rather to say it won't be as amazingly fast as you might guess. Though I suppose if your memory got one address and sent back several words they it's _possible_ because you'd have to send a couple requests.

I'm getting too theoretical here. Cache predictions can get really complex and hard. It depends on your application and the code and the compiler and the hardware and maybe phase of the moon too.

Re:Linux in cache?

[runderwo]

Try implementing a MP system with hardware snooping. Then try it again with a software-controlled cache. See which one scales better, and get back to me on that.

Re:Linux in cache?

[runderwo]

BTW, s/mark pages dirty/invalidate cache lines/ in case you couldn't figure that out.

Re:Linux in cache?

[statusbar]

In addition, the G4 and I believe the G5 processors have cache control instructions as well - So you can start 4 fragmented streams of data going to and from the cache before you need the data. Sometimes, with proper tuning you can achieve 50% faster results.

--jeff++

Re:Linux in cache?

[Nexx]

cache captures instructions, and sometimes data too, not programs.

What're programs if they're not sets of instructions?

Re:Linux in cache?

[mrm677]

Only in some designs. Architectures like MIPS allow for both a cache-coherent or non-cache-coherent design. In a non-cache-coherent design, the cache is not transparent, and the kernel programmer is responsible for cache management; marking pages as dirty, flushing cache, etc.

There is a big, big difference between caching and keeping caches coherence in a multiprocessor system. You are correct that higher performance can be obtained by doing away with cache coherence. The Cray T3D and T3E machines have no cache coherence and the latter scales to 4096 CPUs I believe.

However not considering MP software with sharing between processors, there are very few uniprocessors that allow software to control caches. The ones that do exist, for example some recent ARM cores, do allow software to pin certain blocks/pages/whatever in cache. This is certainly impractical for user applications and probably too impractical for a kernel as well. It creates a tight coupling between hardware and software.

My thoughts...

[big_groo]

I haven't noticed *any* differences on the desktop. ALSA is nice, kernel config is easier, but other than that...nothing noteworthy over 2.4.19 (my last kernel). Am I missing something? Oh, and I *still* can't get my USB mic to work (help!).

Slackware with Dropline, btw. I do notice that Java tends to take up 250MB of RAM every once in a while while running Firebird. I didn't have that problem with 2.x.x.

Re:My thoughts...

[pacman on prozac]

My thoughts were, wow, much faster. I'm now running 2.6 on all my desktop machines and it flies. They "seem" much more responsive with 2.6 than 2.4, especially under load.

The initial boot time to load the kernel seems to have massively dropped although I could be imagining that.

The new build system in 2.6 definately rocks, forgot to compile something important in? No need to wait for * to recompile anymore, just the vital parts are re-done.

Re:My thoughts...

[big_groo]

I *thought* boot time was a bit faster, but I for some reason I thought it was just me. I guess it is a *bit* faster on boot. I honestly didn't see an appreciable increase in speed though. It just could be GNome :) (SMACK -1 troll, I know, but I don't use KDE)

P4 2.2, 768MB RAM. NO swapping :).

Re:My thoughts...

[echeslack]

I don't notice any real speedups, but I do not really notice updatedb running anymore. Actually that's not entirely true, I just notice it a lot less. Many times it will run and I won't notice it. The mouse used to start getting all sticky when I had too much disk activity. Now it *mostly* runs smoothly.

Re:My thoughts...

[Coryoth]

I haven't noticed *any* differences on the desktop. ALSA is nice, kernel config is easier, but other than that...nothing noteworthy over 2.4.19 (my last kernel). Am I missing something?

Well, it's not exactly something you notice on the desktop, but having LSM built into the kernel now is definitely a good thing. This is basically SELinux folded into the mainstream kernel - and the improved security available from that is impressive. It is the sort of thing you really OUGHT to be using, even on a desktop machine. Think of it as kind of a like firewall that screens kernel operations instead of packets: you can set up fine grained policies that prevent processes doing anything they shouldn't. That means that, for instance, with a well configured system, a buffer overflow DOES NOT provide root on the box - it just crashes the process that was overflowed.

So, everyone out there running 2.6 kernels, go and get some configuration tools and set up your LSM security. The sooner everyone is running LSM security, the sooner we can say that Linux really does have provable significant security benefits over Windows.

Jedidiah.

Re:My thoughts...

[mgv]

I *thought* boot time was a bit faster, but I for some reason I thought it was just me. I guess it is a *bit* faster on boot. I honestly didn't see an appreciable increase in speed though. It just could be GNome :) (SMACK -1 troll, I know, but I don't use KDE)

P4 2.2, 768MB RAM. NO swapping :).


I think that your absense of a swap file may explain this. 2.6 works alot faster on disk I/O.

This suggests to me that older low memory machines that use alot of swap file access may be much faster. Which would be incredibly useful to legacy machines.

Anyone else using an older machine with smaller amounts of ram notice an improvement in 2.6?

Michael

Re:My thoughts...

[owlstead]

No, you aren't missing something. Applications will still run as they did, except when they are heavily multithreaded. Starting up something CPU-cycle expensive and typing in open office might give you an idea.

But all in all, it's the kernel. End users should be nicely unaware of it. Don't expect any fireworks to go off, most of the time you notice a kernel you will have hoped you didn't :).

Re:My thoughts...

[big_groo]

I have a swap file. It's almost 1GB. The OS just doesn't use it.

Re:My thoughts...

[Anonymous Coward]

You sure, must OSes will make use of a swap file to swap out anything that hasn't been accessed for a while and basically doesn't appear to be needed for the moment.

Take a look at /proc/meminfo

Here's what mine says:

cat /proc/meminfo
total: used: free: shared: buffers: cached:
Mem: 528654336 507162624 21491712 0 25026560 293031936
Swap: 671399936 127127552 544272384

So i'm using virtually all of my 512meg of RAM and almost 128 meg of swap. Most of the RAM allocation will be IO caches and the swap will be things i'm not making use of.

btw i'm using 2.4.

Re:My thoughts...

[Robo210]

I use 2.6 on a Pentium 133 with 48mb of RAM, and I have noticed *huge* differences in the boot time. KDE seems to start up slower, but run a bit faster... well, at least the mouse seems to lag a bit less... its hard to tell when everything runs so slow to begin with. Running either 2.4 or 2.6 still eats my swap file, but I guess thats what I get for using such old hardware. Oh, and Gnome acts the same.

Re:My thoughts...

[shadowbearer]

Ditto here. I've been running the -test kernels on the faster machine since summer, so I can't remember just how much difference there was. I do know it was noticeable, tho. In particular setiathome used to noticeably slow the machine; now I don't even notice whether it's running or not. *grin* 2.6 definitely WU'ed me there *grin*

On the laptop I just compiled 2.6.1 for, however, (a 200mhz DEC HiNote) the speed increases are huge. You're not imagining the boot time drop - it's easily twice as fast on the laptop as 2.4.20 was. The GUI is also noticeably more responsive.

The new build system is great, especially on a slow machine :) Kudos to the kernel people, and thanks!

SB

Re:My thoughts...

[MikeCapone]

Oh, and Gnome acts the same.

On such old hardare, may I suggest you try XFce 4?

I tried it at first because I, too, have old hardware, but endend up liking it more than both KDE and GNOME (independantly of speed); once it's set up properly and you learn the neat things about it (like switching between virtual desktops with the mouse wheel), it's hard to go back.

Re:My thoughts...

[Zebbers]

are you kidding? With the new scheduler and premption how can you not notice 2.6? Get two kernel images, one 2.4 and one 2.6 and have them up in Lilo ready for loading. There is a difference. It may be hard to measure but its easy to *feel*. Kind of like KDE3.2rc1, it feels much better than 3.1.5 did....

Alsa and everything is just part of it too but that stuff is great...it seems like it took less work to get a lot of functions up and going running 2.6.1 than it would have under 2.4.....

Re:My thoughts...

[Pharmboy]

He probably is rebooting more frequently than he needs to. I run a gig and half of ram on my dual p3/1000 and it uses swap after about a week. Then it will flush out, use swap, etc. I never actually use half the ram, its just buffer/cache, but it does use the swap. Basically, Im guessing his uptime isn't long enough to cache that much data. (mine is about 100 days right now, a little under average.)

Re:My thoughts...

[mellon101]

CD and DVD burning used to absolutely kill my system with 2.4 kernels. High cpu usage and everything was jumpy and slow. With 2.6 I just cant seem to bog it down ....no matter what I do. I can be compiling GNOME, burning a DVD, and doing a number of other things ....and it all is smooth as silk. I love this kernel!

Re:My thoughts...

[be-fan]

Applications will generally run at the same speed, but lots of things should be visibly faster. I've noticed the following things on my machine (2.0GHz P4 640MB of RAM, KDE 3.2):

- Better interactivity and elimination of the "nice -10" hack for XFree.
- Much nicer behavior while running compiles or apt-get upgrade. I can run a compile in the background and not even notice it in the foreground. Its nearly impossible to get audio to skip.
- Better I/O performance. Doing a large disk write doesn't make things stutter. With the CFQ scheduler, heavy disk activity doesn't adversely impact startup time of newly launched apps.

Re:My thoughts...

[dnoyeb]

I have a swap file. It's almost 1GB. The OS just doesn't use it.
Which was exactly his point. If it were you'd see an improvement.

Re:My thoughts...

[rgmoore]

That means that, for instance, with a well configured system, a buffer overflow DOES NOT provide root on the box - it just crashes the process that was overflowed.

That's not quite right. What it really means is that it's now possible to set up a system where being root doesn't give you complete control over the system. IOW a cracker can open up a root shell on your box and still not be able to do much damage.

Re:My thoughts...

[Coryoth]

Yes, quite correct, thanks.

I guess I meant that having the root account doesn't amount to having what one traditionally consider "root" on the box.

Re:My thoughts...

[bonch]

So far, all I've ever seen regarding desktop improvements with 2.6 are anecdotal "it seems faster" stories that are probably based on the fact you're expecting it to be faster and so convince yourself that you notice such.

I see no reason to upgrade for me.

Re:My thoughts...

[SuperFrink]

The initial boot time to load the kernel seems to have massively dropped although I could be imagining that.

Funny you mention that. I've been thinking of using a watch to see if was imagining things. It sometimes takes me longer to get an IP via DHCP then it does to boot up. :^) I don't start much more than ssh at boot though. (Linux 2.6, Slackware 9.1, runlevel 3)

Re:My thoughts...

[stor]

I find the desktop generally a bit more responsive. The mouse feels a lot smoother under 2.6 for one thing.

I _really_ notice the responsiveness, though when I start up a I/O-hungry job on one workspace, switch to another workspace and continue working.

With 2.4, the background job would tend to starve the system of resources, making the mouse jerky and windows unresponsive. With 2.6, often I forget I'm running the background job.

This is FC-1, 2.6.1 vanilla with modules, preemption ON. Measuring device (my perception) may not be accurate, but I'm not going back.

Cheers
Stor

Re:My thoughts...

[Serveert]

Enable the preemptible kernel option before compiling, you will see GUI improvements.

Re:My thoughts...

[Serveert]

Are you a troll? If not, learn about what "preemptible kernel" means. I'll give you a hint: it means the GUI will be more responsive.

Re:My thoughts...

[vigilology]

The new build system in 2.6 definately rocks, forgot to compile something important in? No need to wait for * to recompile anymore, just the vital parts are re-done.

Hmm, make has always been like that for me, no matter what I compile. It only recompiles the bits needed for the change, never the whole lot...

Re:My thoughts...

[Creedo]

For gui responsiveness, yes.

Re:My thoughts...

[Wolfrider]

--Trim it down to ~200MB then, and give yourself enough space to store another Linux live-cd ISO image. ;-)

Re:My thoughts...

[aulendil]

it means the GUI will be more responsive.

Why?

Re:My thoughts...

[Serveert]

The kernel can be pre-empted by syscalls. Before if you call, say, write() to write to a file, the kernel must write to the file before anything else can happen. Interrupts have always pre-empted syscalls, but now we can pre-empt write() to execute a write() or read() for another process, or we can pre-empt it to move the mouse. Hence one reason why GUI should be faster.

2.6 on server?

[black666]

Those benchmarks are nice, but who runs kernel 2.6 on production servers that need every speed they can get? It will be a few more 2.6.x releases until I consider running one of my servers with a 2.6 kernel.

Re:2.6 on server?

[Maserati]

Start planning, it's a safe bet they'll get there. Put a game server on 2.6 and see how it plays, or do something with a 2.6 box, you *will* be upgrading eventually.

Re:2.6 on server?

[popa]

i do

Re:2.6 on server?

[1000StonedMonkeys]

Well, yes, but it's a safe bet that the speed improvements in 2.6.0 won't disappear in later versions of the kernel.

Re:Where's the distros

[deadsaijinx*]

2.6 hasn't had the long real-life test use that the 2.4 family has. However, that isn't to say that 2.6 has been ignored by the distro kings, just not their staple offering, which would need to be absolutely assured of stability and compatibility.
Example, Suse 9 came with a copy of 2.4.21-144 which it installed and then gave you the sources and information needed if you wanted to update to 2.6. So, yes, it's out there, and it's in major distros, but not having the absolute assuredness of the 2.4 line, 2.6 is left as a secondary option.

BSD Vs Linux Vs Windows 2k3

[Anonymous Coward]

How does 2.6 compare to Free or Open BSD & how do they compare to windows 2003 server doing the same job?

NTFS Read/write support?

[lawpoop]

The chart on the first page says that 2.6 supports read and write for NTFS. Is this really the case? Does anyone trust NTFS writing if it's in the kernel?

Re:NTFS Read/write support?

[jensend]

The native ntfs driver supports writing only if the write operation wouldn't disturb any of the metadata- that is, you can't create or delete any files but you can safely overwrite a file if you don't change its size. This is more useful than it sounds at first- you can use a loopback file on the ntfs partition for anything you want.

Re:NTFS Read/write support?

[ensignyu]

Write support is (supposedly) safe, because it will *only* let you do things that are safe. Pretty much limited to writing to an existing file without changing the size.

So it's not really full NTFS write support yet.

Re:NTFS Read/write support?

[Cocodude]

Personally, I think this should be mentioned explicitly in the kernel configuration. When I enabled write support, I thought it didn't enable anything above read support, but thank you for your comments. It was one reason I upgraded to 2.6, but I am still glad I did so, it feels more responsive in general.

Why don't we have NTFS support yet though? I know NTFS may be far more complicated than FAT, but we do have read/write access to this (at least for floppy discs!).

Re:NTFS Read/write support?

[r00zky]

Why don't we have NTFS support yet though?

Ask Microsoft.

Re:NTFS Read/write support?

[damiam]

The Captive project uses Wine technology to natively load the Windows NTFS driver. It's not quite as good as native, but it might fit your needs.

Re:NTFS Read/write support?

[r00zky]

Since when is it Microsoft's job to provide drivers for a competing operating system?
Not the drivers, the specification...
oh wait, that would go against their monopolistic principles, nevermind.

Re:NTFS Read/write support?

[aiyo]

A good company would provide both specs and cross platform drivers especially for something that is of critical importance such as the file system. Really, the only people they hurt is their own users since *nix people can do without ntfs support. We're so lucky to have better options.

Re:NTFS Read/write support?

[RedWizzard]

And why should they be required to give you specs for their file system?

No one said they should be required to give out the specs for their file system. The question was "why don't we have write support?". The answer is "because Microsoft haven't released the specs".

Re:NTFS Read/write support?

[Serveert]

If you do a make menuconfig and click on help for ntfs write support it will spell this out for you.

Re:NTFS Read/write support?

[stock]

furthermore, why would a tool like captive-ntfs need things like gnome-vfs-2.0 ? Why does something that should work flawlessly from the commandline need a full GNOME layer?? :

[jackson:stock]:(/usr/local/lib)$ ldd libcaptive-1.1.5.so
libgnomevfs-2.so.0 => /usr/lib/libgnomevfs-2.so.0 (0x400bd000)
libgconf-2.so.4 => /usr/lib/libgconf-2.so.4 (0x400fa000)
libpthread.so.0 => /lib/i686/libpthread.so.0 (0x40130000)
librt.so.1 => /lib/i686/librt.so.1 (0x40180000)
libbonobo-activation.so.4 => /usr/lib/libbonobo-activation.so.4 (0x40193000)
libORBitCosNaming-2.so.0 => /usr/lib/libORBitCosNaming-2.so.0 (0x401a7000)
libORBit-2.so.0 => /usr/lib/libORBit-2.so.0 (0x401b5000)
libpopt.so.0 => /lib/libpopt.so.0 (0x40205000)
libssl.so.0.9.7 => /usr/lib/libssl.so.0.9.7 (0x4020d000)
libcrypto.so.0.9.7 => /usr/lib/libcrypto.so.0.9.7 (0x4023f000)
libxml2.so.2 => /usr/lib/libxml2.so.2 (0x40342000)
libz.so.1 => /lib/libz.so.1 (0x40443000)
libm.so.6 => /lib/i686/libm.so.6 (0x40451000)
libgmodule-2.0.so.0 => /usr/lib/libgmodule-2.0.so.0 (0x40474000)
libdl.so.2 => /lib/libdl.so.2 (0x40478000)
libgobject-2.0.so.0 => /usr/lib/libgobject-2.0.so.0 (0x4047b000)
libgthread-2.0.so.0 => /usr/lib/libgthread-2.0.so.0 (0x404b1000)
libglib-2.0.so.0 => /usr/lib/libglib-2.0.so.0 (0x404b6000)
libc.so.6 => /lib/i686/libc.so.6 (0x40522000) /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x80000000)
[jackson:stock]:(/usr/local/lib)$

A historic design failure... sad but true

Robert

Re:NTFS Read/write support?

[bmzf]

I found the Captive project quite impressive. It's a pretty neat idea, I'll agree to that.
However, if you have anything on your NTFS partition that you actually like/want, don't use Captive. I tested it out and here's what I can conclude:

Reading NTFS: works fine.
Writing NTFS: looks like it works fine on the linux side, but as soon as you reboot into your windows system, expect the file to be missing/corrupted, or for windows to refuse to boot. You'll need the install CD to boot into repair mode and do a check disk. This is especially likely to happen if you try to copy some files to the NTFS partition, or create new ones.

After frantically getting my windows system to boot (I need the Visual Studio .NET on there), I went back and pulled Captive and reinstalled my Linux-NTFS rpm from linux-ntfs.sf.net

It obviously loads much much faster, and I don't have to worry about whether I can boot into my other OS again.


Hope this helps. Don't say no one warned you ;)

It's shame, really.

[Anonymous Coward]

It's not Linux 2.6.x, it's SCO/Linux 2.6.1.

Re:It's shame, really.

[shfted!]

Tell that to GNU/RMS!

Also notable....

[haggar]

..is the parformance of the Opteron. Looks like Linux 2.6.x and Opteron are a great combo. Okay, I admit, I was a bit skeptical regarding Linux 2.6, but it seems it might actually deliver.

I'm looking forward for Solaris + Opteron servers. Should be another interesting combo, performance wise. For one, Solaris 9 has some fantastic scheduling for multiprocessor machines. Additionally, it has been implemented in 64 bit for many years.

Re:Also notable....

[Doomdark]

I'm looking forward for Solaris + Opteron servers. Should be another interesting combo, performance wise. For one, Solaris 9 has some fantastic scheduling for multiprocessor machines. Additionally, it has been implemented in 64 bit for many years.

One potential concern, though, is that while there has been Solaris for x86 for a while, it's not really its main platform... so it's bit like waiting for 2.4.6 kernel on, say, Sparc systems. So I'm wondering if Solaris design of scheduling (and other kernel parts) takes some advantage of Sparc processor's design, that wouldn't map nicely into Opteron? In future this should get better (since Sun is now allied with AMD), but Solaris 9 was written probably well before x86 was seen as strategic platform for Sun.

Re:Also notable....

[paitre]

Sun is doing an Opteron-optimized build of Solaris, but don't expect it in the short term.
That's about all I can say ATM, but it -is- on the way...

Re:Also notable....

[Anonymous Coward]

Sun does sell some machines that run Solaris and have x86 processors in them. They sell one Athlon (Mobile Athlon, actually -- it's in a blade server) and several dual Xeon systems. (See here.) So it's not as if the x86 version of Solaris is totally the ugly stepchild of the Sun family.

Plus, yeah, it's helpful to get into assembly to write good context switch code and MMU stuff and whatnot, but how much of the scheduling and scalability is really based on platform-specific optimizations? I think most of it is due to choosing a good design and good algorithms. Especially these days, when raw CPU is no longer the problem because CPUs are so fast. My point is that I think lots of the value in the Solaris kernel translates to any architecture -- like a fully pre-emptable kernel, and design decisions that minimize contention for locks and things like that.

Re:Also notable....

[FuzzyBad-Mofo]

Indeed, the Opteron was kicking major ass. w00t!

Re:Also notable....

[Serveert]

ouch.

Re:Also notable....

[Haeleth]

Additionally, [Solaris] has been implemented in 64 bit for many years.

So has Linux; it's been running on the Alpha architecture since the mid 1990s.

Re:Where's the distros

[MarcQuadra]

Wow, you need to quit bitching. 2.6 will be in the distros 'when they're ready', do you remember all the really broken 2.4.x stuff? It was REALLY bad press for folks who COULDN'T UNMOUNT DRIVES safely.

Let the ubernerds self-build 2.6 systems for a while and work out more bugs. If you want it you can have it, but mass-distribution before we even hit 2.6.2 might be a BIG mistake.

I can't believe these results

[corebreech]

From the article:

For instance, a simple read of a 500MB file during a streaming write with a 1MB block size on my Xeon-based test system took 37 seconds with v2.4.23, and 3.9 seconds with v2.6.

Huh? That just can't be right, can it?

Also, a lot of us have been running with NPTL for some time now before shitcanning our Red Hat installs... I would've like to seen a comparison between 2.6 and a NPTL 2.4.

Re:I can't believe these results

[JumboMessiah]

Actually, it wouldn't suprise me if this is correct. If you notice, he was reading the 500MB file while a continuous streaming write was going on in the background. On 2.4.x, a write streamout will kill read performance drastically. Mostly due to the way the I/O scheduler schedules the read. Which, most of the time, is to stash it at the end of the writes.

The two new I/O schedulers in 2.6.x help to resolve this. For more info, check here.

Re:I can't believe these results

[ron_ivi]

Ah, then it'd be interesting to know how much the write was delayed. If the write finished 40 seconds later then it's easy to understand what happened.

Re:I can't believe these results

[autocracy]

I think it's more likely that the reads and writes were done in larger blocks. Nothing kills disk speed like unnecesary seeking across the platter. Of course, it should be noted that this is informed speculation, and not a certainty.

Re:I can't believe these results

[segfaultcoredump]

In tests that i've done on solaris boxes (a very small 2 cpu UltraSPARC III server, not the big iron) I often see 'strange' results like this from benchmarking utilities like iozone. (as in 600+MB/s read and write speeds)

In general, if the file fits into memory, read and write speeds can often exceed the theoretical limits of the underlying array. This is because the OS buffers the hell out of the I/O (unless you use something like UFS's 'forcedirectio' mount flag to avoid the OS buffers). This is good and bad. Good for throughput but bad if the buffer is allowed to grow unchecked and forces applications into swap as one tries to write a huge file that is larger than the RAM on the system. (think DVD iso's on a box with only 4G of ram).

In the case of the 2.4 vs 2.6 shootout, I'd be much more interested in seeing a full iozone graph showing block size vs file size vs performance. ( If you do this, make sure that you test for file sizes at least twice as large as the RAM on the box to see what happens when the system runs out of space for a buffer. )

All of that said, the best I've seen so far (same 2 cpu sparc box) in actual disk performance is 180MB/s (and that was with a NexSAN ATABeast that had 42 300G IDE drives). The load was spread over two 1Gb fibre channel links, so i was starting to approach the max throughput of the HBA's at that point :-)

Good to know that.

[nefele]

So if I use the new and improved herbal 2.6 kernel my processing power will be UP TO 150% BIGGER and my UPTIME will be 200% LONGER!!
And it's only $699 a box! ;-)

Time for some more FAIR benchmarks

[haruchai]

I'd like to see how Linux 2.6 stacks up against Windows Server 2003 now. This time, let's have Microsoft and Redhat or some other Linux gurus go head to head.
One of the good things of benchmarking at an early stage is that it may expose some hard to find weaknesses, much like the first Mindcraft tests exposed a kernel limitation which hampered Apache's performance.

Re:Time for some more FAIR benchmarks

[Richard_at_work]

There will really be no fair benchmarks until a farmer in the back of beyond who doesnt normally touch computers wakes up one day and thinks

"Well, shucks, today im going to buy a new computer, Redhat9, Win2K3 and spend some time installing them, running the same sort of tasks on each, and jotting down the results. I have no idea why, but lets do it anyway."

Until someone who has no affiliation with either runs the benchmarks, you jsut cant rely on the results. Im not saying they would be messed with, but when you like one OS more than the other, its easy to introduce discrepancies.

Re:Time for some more FAIR benchmarks

[haruchai]

That's why I proposed this in a thread on osnews.com

Read the comment by my ( as Birdie.P) here:

http://www.osnews.com/comment.php?news_id=5868&o ff set=15&rows=30

Mystical Mozilla Speed Increases...

[GameGod0]

Linus said in an interview somewhere that he was personally working on stuff in the 2.6 kernel to make Mozilla load faster.

I can tell you, boy, was I disappointed. When I tried 2.6.0, I couldn't tell an ounce of speed difference from 2.4.x...

Has anyone seen this mystical speed increase?

Re:Mystical Mozilla Speed Increases...

[Garg]

Sure, it's right there in the kernel:

if (strcmp(username,"ltorvalds")==0) {
gooseMozilla();
}

Garg

Re:Mystical Mozilla Speed Increases...

[GameGod0]

Maybe it just doesn't work with Mozilla Firebird... (I thought I tested vanilla Mozilla as well though...) Ok, thanks anyways.

Re:Mystical Mozilla Speed Increases...

[christopherfinke]

...stuff in the 2.6 kernel to make Mozilla load faster.

Hmmm... building the Web browser into the OS. That couldn't cause any problems, right? Right?

Re:Mystical Mozilla Speed Increases...

[wdnspoon]

For most people out there, running Athlon's with 2Ghz+ clocks and 1Gb DDRAM you're not going to see such a big change that you can really 'feel' the difference right away. Then again, on most modern systems starting Mozilla is not such a huge deal (mostly a function of how fast the heads on your drive move.) Running a Celeron 600 w/ 256mb SDRAM I can really tell that things have improved. Mozilla 1.6 is loading as fast under 2.6.2-rc3 as Firebird 0.7 did on 2.4.22. The performance of everything GUI is also much, much, much smoother under heavy loads. I take advantage of this new kernel power by being able to recompile my kernel while running OpenOffice.org and listening to mp3's on XMMS. Pretty impressive for a piece of hardware that could barely run a Flash movie and a Java game at the same time under Windows 2000.

Re:Where's the distros

[Thomas M Hughes]

Debian/Unstable can install 2.6.0 a regular apt-get install at this very moment. Its just something like:

apt-get install kernel-image-2.6.0-1-processor type

If you're running stable, you shouldn't be running a 2.6 kernel anyway.

Tests

[Sir Pallas]

I guess it's important to ask the following question: was 2.4 ever designed to run on those kinds of processors? I mean, the O(1) scheduler is a pretty cool, processor independant change; but was 2.6 designed with specific optimizations for newer processors (and newer instructions) in mind? I'd be interested to see benchmarks from old hardware -- i.e., stuff like I've got sitting around. (If only I had a bit more time. Maybe I can borrow some cycles from 2.6 Linux boxen.)

Re:Tests

[chunkwhite86]

I guess it's important to ask the following question: was 2.4 ever designed to run on those kinds of processors? I mean, the O(1) scheduler is a pretty cool, processor independant change; but was 2.6 designed with specific optimizations for newer processors (and newer instructions) in mind? I'd be interested to see benchmarks from old hardware -- i.e., stuff like I've got sitting around. (If only I had a bit more time. Maybe I can borrow some cycles from 2.6 Linux boxen.)

I'm guessing that the answer is yes and no.

When you roll your own kernel, you can select which CPU you are building for. This enables optimizations for that specific CPU (but usually breaks compatibility with older CPU's). So no, I don't think it was build with these new CPU's in mind, but it does have the optional optimizations in there to take advantage of them - just like it does for 486 and pentium and Athlon.

Keep in mind that your compiler must also have approperiate optimizations for your CPU to see the best performance. I'm not sure how far gcc has come on PPC970, Opteron, and Itanium these days.

early 2.6 much better than early 2.4

[Goonie]

I've personally had no trouble with 2.6 since 2.6-pre. While a sample size of 1 isn't much use, the lack of hasty releases of new 2.6 versions, and the lack of bitching about showstopper bugs in 2.6, suggests it's not in too bad shape, IMO.

Re:early 2.6 much better than early 2.4

[MarcQuadra]

I think that has a lot to do with chicken-and-egg though, there aren't many people having trouble because very few people are actually running it, because the distros are holding off.

When 2.4 came out it had stuff people needed right away on their desktops, USB, many more drivers, DRI for XFree86, and a whole slew of other things that we were playing catch-up with. 2.6 performs better, but it offers very little in terms of desktop-relevant improvements. The move to 2.6 will naturally be much slower than the move to 2.4, because there's less reason to go there right away.

I do think 2.6 will be less painful than 2.4 was, because of the big vendor support from IBM, and Redhat's closer involvement, amongst others. But just because it will be less trouble doesn't mean it won't be any. I'll personally be waiting for several releases before I take the plunge on any production boxes.

Re:Who cares? OS X kicks both.

[mgv]

OS X is far better than either "kernel". With OS X you get a better GUI, more apps, and an OS that is backed up by a fantastic corporation, rather than a bunch of hackers who have dubious backgrounds and sketchy credentials.

You aren't comparing like with like. I have windows, linux and OSX running on different laptops at home (ok, I have a problem with needing toys, but at least I have insight).

OSX is very nice - but you don't buy it for speed. In fact, the sort of people who buy it often gloat at how you don't need to worry about that sort of thing with a Mac. I don't doubt that for many windows users, they would be much better off with a mac, as they are pretty clueless, and Mac's are a very nice implementation of the BSD core.

However, this isn't what this post is about. It could be argued that any server that runs a GUI is wasting resources. It depends on what you are asking your server to do.

There are clear improvements in the 2.6 kernel, especially with regard to Disk access from what I can see in the article.

This is totally different to which OS provides the better GUI. In any case, OSX doesn't run on a vast amount of hardware out there, and your attitude is that all that intel stuff is only good for landfill. If you accept that there is alot of hardware out there that will NEVER run OSX, then you should also accept that for those people, its very useful to know if 2.6 is a better performer.

Michael

Re:Where's the distros

[mindstrm]

Tried it a week ago, and I did have to separtely download and build the module utilities.... the versions in unstable were not current enough at the time.

Perhaps that changed in the last week.

Either way it's not a big deal... people should give 2.6 a try, it's not that scary.

2.4 versus 2.6: file system performance?

[Debian Troll's Best]

These are some pretty encouraging results. The hard work put in by all the kernel developers has obviously paid off in a big way. However, after reading the article I still have a few questions about kernel 2.6 performance, namely filesystem performance. Rapid random read/write access is obviously highly critical for enterprise type applications, such as apt-get package management and package database updates. Basically with the 2.4.x series of kernels, filesystem performance using either the ext2 or ext3 schemes could drop to below 5 apt-get package installs per second, even on large SMP/RAID systems. I have been investigating the use of raw disk I/O (similar to that used for high performance table spaces in products like Oracle and DB2) to reach my target of 100 apt-get package installations per second on commodity level hardware, via custom kernel level ATA and SCSI chipset drivers. But I'd love to hear that FS speeds have been improved in the 2.6 kernel. Has anybody benchmarked this aspect of the new kernel? And if so, when could we expect to see Debian start shipping with the 2.6 kernel? I look forward to reading the community's response.

Re:2.4 versus 2.6: file system performance?

[hdparm]

I was wondering about those same issues but my 8 CPU boxes are strained even more - due to a tail combining bug in a big_mem kernel, I had to load ReiserFS driver as a C# compiled module to try to up the apt-get installs to 130/sec, which is of essence for distributed transactions our MySQL driven portal is running. Below, you can see just a beginning of REAL_TIME dumps, unfortunatelly Synaptic crashed under heavy load and I couldn't post graphs :o(

REAL_TIME CPU_TIME DFA BFA

CREATE 29.16 1.220 1.422 3.779 1.491 1.645 13.52 1.182 2.013 2.087 1.997 0.657 789364 1.208 1.180 1.180 1.180 1.181
COPY 119.64 1.211 1.191 1.473 1.230 7.288 21.98 1.152 1.515 1.746 1.520 0.695 1578116 1.208 1.180 1.180 1.180 1.182
READ 116.55 1.213 1.177 1.025 1.134 6.850 18.35

Anyway, hope this helps in establishing what patches you need to apply. Let us know about your progress and please ask if you need any additional help.

Re:2.4 versus 2.6: file system performance?

[Debian Troll's Best]

Thanks for the suggestions hdparm, but since I posted my original comment my colleagues and I have decided on taking a hardware-oriented approach to solve our problem. Specifically, this will involve careful removal of the outer metal shell of each of the Seagate Cheetah SCSI hard drives from our RAID1+0 setup and making a few changes to the internal structure of the drive to effectively 'overclock' the hard drives. Let me explain

Stock Seagate Cheetahs use a fairly standard aluminium drive shaft, much like the one in a consumer grade piece of rubbish. We are replacing each of these with a carbon propeller shaft and light-weight fly wheel, which will increase initial acceleration of the drive platters, and will allow them to spin at a maximum speed of 17,500rpm versus the standard 10,000rpm. This should see our rate of apt-get transactions improve dramatically. But that's not all. As any good CPU overclocker knows, 'lapping' the contact surface of their heatsink will remove microscopic imperfections and result in a closer contact between heatsink and CPU. We too will be 'lapping' each hard drive platter. Of course this is dangerous to the platters, so we are always sure to use a fresh Kleenex each time. Once the platters are lapped, we can alter the suspension and damping characteristics of the read/write heads, making them float even closer to the platter and resulting in sportier turn-in, less body roll and more predictable handling even when dealing with 'rough' packages such as Troll Tech's Qt libraries which still have an aura of 'non-free' about them.

Finally we short-circuit resistor A24-J, which amazingly unlocks a special 'developers' mode of the hard drive, and firmware commands may be directly inputted using a text editor. We have developed a set of SCSI firmware routines which recognise the apt-get and .deb file formats even at the lowest level of hardware, offering stellar apt-get-goodness. Using a customised version of apt-get implemented in a mix of x86 assembler and Python (for the performance critical parts), apt-get is now able to bypass the Linux kernel, PC BIOS and the SCSI controller card, and communicate with the hard drive mechanics directly. This adds approximately an extra 60% to overall performance, to say nothing of the improvement in overall reliability and robustness.

We feel that these modifications will result in a drive array that will provide a superior platform for high-throughput enterprise level apt-get package management, regardless of filesystem. In fact we have very little choice about filesystem, since the lapping procedure with the Kleenex irreversibly etches tracks and sectors onto the drive surface. No need to worry about 5% of the drive being wasted on superuser-only space after a reformat! Now, I realise that these types of hardware mods may not be in the reach of all Debian users out there. I'm happy to discuss this further with the community if necessary. I am also creating a HOW-TO, which will be distributed via apt-get mirrors in the form of an 'info' document (man pages are filled with inaccuracies due to the inherently lossy compression techniques used in their production. RMS was really onto something with info!!!).

I look forward to the GNU/community's feedback.

Re:2.4 versus 2.6: file system performance?

[hdparm]

Let me tell you - AMAZED AND BOUNDLESSLY EAGER are less than insignificant words to describe the feeling and energy that simply exploded throughout our 'Zone 3' (slightly enhanced security level and sterile environment - you guessed it, we use Kleenex as well) lab when we saw your post. To think that someone, somewhere draws an inspiration from similar sources and thinks on the same level as we do, sounded like utopia. Until now!

Your post prompted an immediate decission by our executives that we lift all wraps of the project and nullify all applicable disclosure agreements.

I see that you, guys, have pretty good idea as to where large performance gains can be achieved - most people do not understand the importance of digging deep into the machine's guts. My heart kept tearing appart for all the promissing young men and women who are now ultimatelly lost for science. Lost in the underground world of Perl and Larry Wall's cryptic labyrints. Too many times the only answers to their cries for knowledge in various forums were "Why don't you just write a Perl script?" and "RTFM!". They just HAD to learn it, while completely losing focus on more important issues. But... let's go back to what matters - direct access apt-get.

Since you've already covered pretty much every essential aspect, I would just like to offer few tips - tweaks that we have come up with. Tiny ones, true but unbelievable performance enhancers:

a] Carbon shafts worked great in most cases, however - Cheetah's spin rate drops to below 14200 rpms on NTFS writes (be careful, though, still experimental!). We've found a workaround - injecting 3.75% solution of Good Year re-threads into carbon shafts seems to reverse gyro-effect created by propellers, thus reducing centrifugal force to tolerant levels. There are still some rough edges on FAT32 installs but we're exprimenting with different tyre concentrations. Please, don't get me started with why on Earth are we doing this with proprietary file systems. I am still furious with them filing for an XML-->Word document patent and now we have to make sure that apt-get install VB .Net will work flawlessly. Why now when they've finally made such a great progress with Trustworthy Computing, bringig back confidence in their integrity and innovative spirit?

b] Platter lapping was obvious solution from the beginning - what real modder would miss that? But did you know that lapping them in reverse order makes working with .rpms much better, while not affecting .debs at all? Thanks for the dampness alteration tip - we still haven't managed to apt-get update KDE_3.2-nightly.build. I'm sure it'll work now.

And finally:

c] Shorting that resistor never came to mind as a solution. Therefore, we were not able to take advantage of speedy Python execution. Text mode access will immensly improve our turnaround times - serving assembly generated raw XHTML of the Adaptec chip was a real bitch!

It's such a pleasure exchanging cutting-edge knowledge with like-minded experts and colleagues. Keep up the good work and well done!

backports

[POds]

for those who dont know, you've been able to get a back port of 2.6 on woody for the last month (almost).

so go get it, and tell me how it will effect my surfing, emailing, mp3ing and general userish behavour on my P2 400 128RAM...

Go on, get to it! :)

Re:backports

[shadowbearer]

I just compiled 2.6.1 for my 200mhz laptop (Debian unstable) and the speed increase - especially at boot and for Fluxbox - was very, very noticeable, particularly for cpu intensive apps.

I haven't noticed any breakage - not yet - the machine has only been up for 4 days running 2.6.1. But so far it's great :)

BTW I used the kernel source from debian, not the backport.

A question for anyone out there with a Digital HiNote 7xx series laptop; any idea which sound chip it uses, and how to set up sound? Google hasn't been very informative. (Not a 2.6 problem, I can't figure out which driver to use; most people seem to be using old SB compatibility, but I can't make it work :( TIA )

SB

Re:backports

[Cire668]

A simple way to identify the sound chip on any computer would be to use lspci; you can either install lspci or do "cat /proc/pci". This will list all PCI devices; you should be able to locate the sound card among these.

Re:backports

[POds]

i just installed (bit the bullet) the 2.6 kernel via backports and well.. the system loaded but i've gota compile my own to get it the way i like it and fully functional...

also, whats interesting, i can use 2.4 along side it! So thats fantastic. And the only manual thing i had to do, besides apt-get is put an initrd=/initrd.img (which is a symlink i had to make to /boot/initrd-2.6.0-bla) in my lilo.conf.

Now to compile and see how well it really works

Re:backports

[Platinum Dragon]

Toss me in the "me t001!!1!!" category, with one caveat.

PII 266, Red Hat 9, vanilla 2.6.1 kernel source, been running the 2.6 branch since the .0-testx released, and for the most part I've been pretty happy with performance, especially considering how I've watched faster machines chug along running Windows and the bloatware it seems to attract. However, I've noticed that gnome-terminal's output is noticeably slower than in 2.4. Does this have to do with the new scheduler? Output seems to come in chunks, where before the lines scrolled smoothly. I also encounter skipping in mp3 playback when I access certain webpages in Galeon, and again, I suspect it's something funny with the new scheduler. However, I'm not going back to 2.4.x any time soon, as I'm happy with this version, and building a kernel seems to have become much, much easier for newbies and grizzled kernel hackers alike.

Re:backports

[shadowbearer]

I haven't tried lspci yet; I'll try later.

There's no info on the sound card in /proc/pci under Debian. That is what has thrown me so far.

SB

Re:backports

[shadowbearer]

I don't run Gnome, so I can't comment on experiences under it. But wrt to music/video playing, I've actually noticed much better performance with 2.6.

I wonder if it might make a difference if you recompiled the apps in question? It sounds like a userspace problem, assuming you haven't seen any other oddities. Do you experience the same thing in other terms? That might at least localize it - I suspect it's probably userspace code, but IANAKernelDev :)

Just a thought.

Cheers
SB

Re:backports

[Spacejock]

I have a Digital Hinote vp717, just upgraded the ram from 32mb to 140mb yesterday. (Want a 16mb stick, got one spare now ;-)

Anyway, the soundcard is a 'Crystal PNP audio system' (how useful is that?) The Windows driver says CS4236/37/38

Fitting a 30gb drive to it this week, then I'll be able to install an icewm setup and I'll mess with the linux sound drivers.

Cheers
Simon
Hal Spacejock Lives!

Re:backports

[shadowbearer]

Thanks, I'll give the Crystal drivers a try. Not very useful, but hey, it's something to try.

Everything I've seen indicated it's some kind of SB16 compat chipset, but nothing I've done works. It's damned hard to find solid info on this laptop....sigh. That's too bad, it's a nice laptop. I may just try to install Mandrake on it, after imaging the Debian install; I'm curious as to whether or not Mandrake's autodetect will recogonize the chip. Alas, I don't have a cdrom, so I can't boot a knoppix cd, which might have helped.

Thanks for the info on the large drive; I was wondering about that.

If you have any luck with the sound drivers, email me

andyg at SCOSUCKS spe midco net

SB

Re:backports

[Spacejock]

Agreed about the quality of the laptop - I've written two novels on it, finishing the third now.

I'm hoping the 30 will install ok. I have a USB external 40gb partitioned 16/16/7 and it can't see the last partition, so I assume 32gb is the limit. I have a 2gb drive in there now, I'm also hoping the 30gb will be a bit quicker, quieter and economical on power. Don't want much, eh?

I have a CD rom in the machine, booted knoppix but it's very slow (didn't have the patience to try audio) I tried 'damn small linux', but it won't see the usb side of things, let alone the audio.

Just tried knoppix again, don't know if cs4236 is a kernerl driver but it's not included with the distro. (Checking the kernel options now)

Cheers Simon

Re:backports

[stor]

Seems to be a known problem.

This seems to be the best explanation of this oddity.

I quote Tim Connors:

AFAIK the definition of jump scrolling is that if xterm is falling
behind, it jumps. Jump scrolling is enabled by default.

What this slowness means is that xterm is getting CPU at just the
right moments that it isn't falling behind, so it doesn't jump - which
means X gets all the CPU to redraw, which means your ls/dmesg anything
else that reads from disk[1] doesn't get any CPU.

Xterm is already functioning as designed - you can't force jump
scrolling to jump more - it is at the mercy of how it gets
scheduled. If there is nothing more in the pipe to draw, it has to
draw.

These bloody interactive changes to make X more responsive are at the
expense of anything that does *real* work.

[1] Others say that it only affects them first time through - after
something is cached, it goes back to normal speed. For me - it is slow
*all* the time. If I pipe it to cat or tail or something, it is a
*lot* quicker.

There's a few flames in that thread ;)

Cheers
Stor

Re:backports

[shadowbearer]

32Gig limit would make sense.

Odd that DSL wouldn't work with USB for you - it did here (well, a USB mouse, anyway - which is nice, as the touchpad on mine sucks, it works, but it's, um, touchy :)

I have the day off, gonna take another shot at the sound drivers. I might even try to install Mandrake...

Cheers
SB

Re:backports

[Malc]

The documentation (HOWTO & FAQ) links seem to be broken. What exactly would I add to my apt sources file to install the kernel. What about bind or Samba? Would I need to uninstall the existing packages first?

Re:backports

[Spacejock]

The hard drive went in fine, although bios says it's 8gb. I partitioned it using a 2.5" external USB drive case, then dropped it into the laptop. I used 16gb fat32, then an extended parition with 384mb swap, 5gb linux and another ~7gb fat32. It's much quieter than the original 2gb, and about half the weight, too. (And 2/3 the height) Less luck installing linux. Tried Mandrake, Knoppix, Gentoo and they all come up with CDRom/Loopback errors. One problem is that they keep choosing 1024x768, which shows up as an oversized screen (I think the panel can only do 800x600, looks like a bug) Will keep messing around with it. Cheers Simon

Re:backports

[shadowbearer]

Check out the Knoppix Cheatcodes.

Yes, the panel can only do 800x600; the vidchip can do 1024x768 on an external monitor. Not a bug, all those cds boot up that way, they see the vidchip capabilities and use them.

Here's the cheatcode for knoppix; at the boot: prompt type

knoppix screen=800x600

Here's the cheatcodes list:

http://download.linuxtag.org/knoppix/knoppix-che at codes.txt

You might try copying the Knoppix image to the HD and boot from it (see the cheatcodes; you can do this from a floppy, too). Might help with the cdrom errors. It's also possible the cdrom can't read the disks you burned properly - I've run into that with older cdroms before on certain cd media, especially silverback disks (blueback disks seem to work much better)

A side note: I have a small partition on my laptop with Damn Small Linux on it; DSL is a cut-down knoppix, so I know it works. I did a network install of it, tho (booted a floppy with the network drivers and copied the image)

The loopback errors are strange. Are you sure all the RAM is good? *scratches head* what was the relevant error text? I'm not a guru but I might have an idea (with Google's help :)

Gentoo is a little more limited, and I don't have the hard drive space to attempt it (shame, I like Gentoo+distcc :). Mandrake I don't even bother to try with older laptops. Debian, of course, installs on *everything* *grin* and although it's a PITA to configure devices, you would definitely be able to install it (it's what's on mine).

Alas, cs4236 didn't work for me...not sure if I'm setting it up correctly, or not. Going to take another shot at it tonite or tomorrow (damn work anyway :)

Hope this helps.

Cheers
SB

Re:backports

[Spacejock]

Hi,

Yes, I use fb800x600 with knoppix, also installed DSL to the hard drive yesterday. (Funnily enough, I accidentally installed it to the 384mb swap partition. They really do mean 'damn small' ;-)

When I installed knoppix to HD the console was 1024x768 on reboot, and when X came up it was a fuzzy grey screen with a white circle for a cursor. Used ctrl-alt-F1 to get to a console, stopped X, changed res in XF86Config, restarted X, still the same. (I want to use icewm on this, kde is too big.)

I have 5 gentoo machines in the house, networked. I know how to set it up so it makes sense to use it, but booting from the gentoo Live CD with 'nofb noX' (so I can read the screen!) gives me a 'CD not found' error. I burned a CD at 4x speed, just in case, but got the same error. (The laptop cdrom is a 24x, so I thought it would be new enough to handle the CDR media)

I have an external 40gb harddrive, I'll try a stage 3 install from there next.

Cheers
Simon

Re:backports

[shadowbearer]

(Funnily enough, I accidentally installed it to the 384mb swap partition. They really do mean 'damn small' ;-)

Heh. They sure do. I used the 50mb image to run linux on my old 486 laptop (put the image within the fat partition and booted via the floppy :)
DSL is sure useful. BTW, you don't actually need to install it, just having the "Knoppix" image on the drive you can boot it using floppy or cdrom. 50mb. Heh :)
WRT to knoppix and X, try changing the fb resolution. I'm not sure how knoppix sets up it's X resolutions, as on the main machine install of knoppix I've not had any luck adding resolutions to the XF86Config-4 file either. It's a minor annoyance, so I've tabled it for the time being. *Frustration* I want my 1280x1024 :) Suspect it's somewhere in all the config files, but know not where.

booting from the gentoo Live CD with 'nofb noX' (so I can read the screen!) gives me a 'CD not found' error

Ouch. That does sound like a bug. I've not tried it that way. I'd suggest posting that and any relevant info to the Gentoo install forum. Might be someone there who can suggest something or confirm a bug.
Have you tried just "nofb"?

Some 24x cdroms (I ran into a few) couldn't handle burned CDs. It's been several years, but I distinctly remember having at least two of them (trashed once I bought newer ones). Sigh. IIRC it was firmware problems.
Not sure who makes the cdrom pack that comes with the HiNote. Probably be damned hard to find out. It may or may not have similar probs. IIRC burning at lower speed didn't solve things, it was the media I was using (cheap 100 disc packs; using a better quality cd helped, sometimes...). Of course the 100disc packs I was getting wholesale had from 3-8 CDs or so turn out defective, so YMMV :)

Good luck with the install. I only have the 2GB HD it came with until I can afford another one, and that's not enough space for a stage3 compile. Sigh. Debian works OK on the laptop (with some tweaking) but I'd rather have Gentoo to complement my other machine. Then again, I hosed my Gentoo install last week so I'm running Debian on the big box until I find the motivation to fix it *grin*. One thing that ticks me off, however, is that making distcc work with debian/gentoo is a major PITA.

Good Luck. Let me know how it went. I believe my email is elsewhere in this thread, if you like.

You must have gotten off work about the same time I did :) MST, by any chance?

Cheers
SB

Re:backports

[Spacejock]

"You must have gotten off work about the same time I did :) MST, by any chance?"

WST - Perth, Western Australia

I'll try the gentoo forums, they're a very helpful bunch.

Cheers
Simon

Re:backports

[shadowbearer]

Good morning :) (I assume :)

Yup, the forums there are great.

Cheers
SB

Re:backports

[True Grit]

Hmmm, under 2.6 /proc/pci is now a legacy interface.

There is some info under /proc/bus/pci, but alsa's info (is that what you're using?) should be accessible via /proc/asound.

Also, everyone should take note of the new sysfs that 2.6 introduces which moves some of the info normally in /proc to a new location (wherever you choose to mount it, normally /sys). This will get more useful as software gets rewritten to use it.

Re:Where's the distros

[aardvarkjoe]

It was REALLY bad press for folks who COULDN'T UNMOUNT DRIVES safely.

The only people who had problems with that were the ones who decided to upgrade their systems within a couple hours of the new (bad) kernel being released. Yeah, sure, it was a mistake, but not nearly as bad as people made it out to be. If you put a three-hour-old kernel on a critical machine, you're taking your chances.

Forgot DMA on the 2.4.23 box?

[anti11es]

For instance, a simple read of a 500MB file during a streaming write with a 1MB block size on my Xeon-based test system took 37 seconds with v2.4.23, and 3.9 seconds with v2.6.

That is a pretty huge difference, could it be that he didn't have DMA off on the 2.4.23 test box while it was on for the 2.6 box? Can anybody confirm this?

Re:Forgot DMA on the 2.4.23 box?

[Nothinman]

I don't think they mentioned the hardware used, but I doubt there were any IDE disks involved.

Re:Forgot DMA on the 2.4.23 box?

[nmos]

AFAIK there arn't even any hard drives in existance that can actually pull data off their platters that fast. Unless this was a RAID setup something is wrong with the test.

Benchmark against vendor kernels?

[salimma]

No self-respecting large Linux distributor would ship a vanilla 2.4 kernel (well, SuSE does, but as an option, and not pre-compiled). The O(1) scheduler which AFAICT makes quite a large performance difference, has been available since RH9, IIRC, and presumably its contemporaries adopted it back then as well.

So it's not such a big leap for real users. Mind you, still a big improvement - especially for interactive use, and also considering that there are so many patches for 2.4 that are now integrated into 2.6, lessening compatibility worries (try patching Red Hat's pre-FC1 2.4 kernel source).

Re:Benchmark against vendor kernels?

[petabyte]

I ran the -ck patches to 2.4 (they have a backported pre-empt, O(1), and low latency) before moving to 2.6. The -ck patches were noticably faster than vanilla 2.4 but 2.6 still crushes them on boot. 2.6 also seems a hair more responsive than 2.4-ck but I expected that.

And I've never used a vendor kernel beyond booting the system and downloading/compiling my own.

Re:Benchmark against vendor kernels?

[salimma]

I expected 2.6 to be faster than patched 2.4 kernels too, but the point is that you run a custom kernel, while the benchmark was comparing vanilla 2.4.23 against vanilla 2.6.0. Which not many people probably use if they care about performance..

Re:Benchmark against vendor kernels?

[MobileC]

Slackware always ships vanilla kernels.
If there is a vital patch then then supply it as well and have instructions on how to patch the kernel.

just wait

[Anonymous Coward]

I can't wait until the year 2007 when 2.6 finally moves to Debian stable.

Re:just wait

[Ziviyr]

Wow, what optimism!

Re:just wait

[Unominous Coward]

Give it time. Debian/stable wasn't built in a day.

Re:just wait

[rawg]

I finally got so sick of Debian that I moved to FreeBSD.

Slashdotted

[chendo]

The site's timing out for me. Anyone have a mirror up or the text of the article?

Re:Slashdotted

[Zork the Almighty]

Maybe they were in the process of switching to 2.6.

2.6 just didn't wow me

[rsilvergun]

on low end hardware at least. I've got an old Dell with a p200 and 64 megs of ram running Slackware 9.1. The processor could keep up but there's just not enough ram, even running a lightweight window manager like blackbox or XFCE. I'm not exactly trying to run uber-apps here (Abiword and Firebird mostly). Switching to 2.6 from 2.4 helped some, but not nearly enough. Funny thing is, I ran RH 6.2 on simularily configured I guess it bugs me because I was looking forward to showing my friends how I've got a modern desktop and applications running on an old P200. hardware. When did Linux become such a memeory hog? And to think I laughed when I saw Lindows' min sys requirements where a PIII 800.

Re:2.6 just didn't wow me

[reiggin]

I had a similar problem and, being a linux newbie, I did some research. The problem isn't just the memory. It's also slow because many parts of a "newer" linux distro are compiled for 686 (Pentium III and higher) processors. This means your older 586 (Pentium) has to do more work. It can really eat into your window manager and X windows server. At any rate, it made my box (PII 350mhz) unusable and I opted for an upgraded box rather than recompile everything and troubleshoot.

Again, I'm a newbie and I'm just trying to vaguely recall the info I discovered that caused me to abandon much hope for my older box.

Re:2.6 just didn't wow me

[cxvx]

I had a similar problem and, being a linux newbie, I did some research. The problem isn't just the memory. It's also slow because many parts of a "newer" linux distro are compiled for 686 (Pentium III and higher) processors. This means your older 586 (Pentium) has to do more work. It can really eat into your window manager and X windows server. At any rate, it made my box (PII 350mhz) unusable and I opted for an upgraded box rather than recompile everything and troubleshoot.

You're not totally correct here:

686 is everything from PentiumII / K6-2 and up. So whatever caused your PII to be slow, it wasn't the compiler optimizations. I suspect it was just low on memory (I'm typing this on a Celeron 300 with 320MB ram, running KDE 3.2 and kernel 2.6.1. It runs just fine).

Re:2.6 just didn't wow me

[xenocide2]

There's a difference between compiled for 686 and optimized for 686. If your software is compiled for 686, there's a good chance that once in a while you'll run into an invalid instruction and the program will halt. If its optimized, I'm told that it may mean more work for a processor not 686 based but I'm not familiar with these platform independent optimizations.

That said, there's plenty to be gained indeed from newer hardware. MMX, sse, sse2 can all be used quite well in mpeg decoding, and general bit blitting. I recall reading about a 16 instuction matrix multiply routine in a Hugi magazine not too long ago, so the extra opcodes can help if present and your compiler knows how to use them effectively.

But there's certainly plenty to be gained from compiling one's own kernel and libraries.

Re:2.6 just didn't wow me

[Nothinman]

A PII is a 686, unless there was some upgrade chip I had never heard of.

Re:2.6 just didn't wow me

[reiggin]

Correct, I typed too quickly. Mine is an AMD K6, not a PII. Don't know what I was thinking. And I'm not sure what the original K6 is. Anyone?

Re:2.6 just didn't wow me

[reiggin]

I made a mistake in the categorization of my processor. It's actually a K6 (original, not K6-2). Don't know why I said PII.

Re:2.6 just didn't wow me

[chunkwhite86]

You're not totally correct here: 686 is everything from PentiumII / K6-2 and up.

Neither are you sir. Ahem, i686 is everything from Pentium Pro and up. I'm typing this from a PPro 200 (with a huge 1MB L2 cache) and 128MB running an i686 mozilla build.

Re:2.6 just didn't wow me

[chunkwhite86]

Correct, I typed too quickly. Mine is an AMD K6, not a PII. Don't know what I was thinking. And I'm not sure what the original K6 is. Anyone?

Original K6 is Pentium II equivalent and enjoyed speeds up to 300 Mhz.

But you forgot about the Pentium Pro. It is also i686. It came in 150, 166, 180, and 200 Mhz. I'm typing this from a 200Mhz PPro with 1MB cache and 128MB RAM using i686 build of moz.

Re:2.6 just didn't wow me

[Anonymous Coward]

Don't blame the kernel for that fact that some of your applications are memory hogs. I would be willing to bet money that your problem is a lack of RAM. In which case, the only involvement the kernel has is that a better paging algorithm could be the difference between very atrocious performance and very very atrocious performance. There is no way that any kernel is going to wow you; it's like having four flat tires on your car and driving around on the rims, then saying that you weren't impressed by the performance improvement you got from installing a new carburetor.

I too, by the way, once had an old Dell with a Pentium 200 and 64 MB of RAM. One thing you'll find is that you can add more RAM, but if you have the same crappy Intel chipset I did, anything more than 64 MB will not be cached, and the limit for total memory is probably 128 MB. (Sad, huh?)

Re:2.6 just didn't wow me

[Anonymous Coward]

The original K6 was missing a couple P6 instructions, so you are probably using a i586 compiled kernel.

Re:2.6 just didn't wow me

[DigiShaman]

You will only have that 64MB cache limit issue if you are using the 430VX chipset. The 430TX addressed this issue I believe. The 430TX is also the last chipset Intel made for the Socket 7 chips such as the Pentium MMX 233Mhz.

Re:2.6 just didn't wow me

[fucksl4shd0t]

You can rock it like Sir Sisyphus, But even in it's genesis, It's really quite ridiculous.

Hey, is that Rolf Harris, or at least a song that he did at some point?

Re:2.6 just didn't wow me

[Malc]

You call that low end? My server for my home LAN and web/email for my domain is a P75 with 32MB. It serves my needs well and I really shouldn't fiddle... but I'm tempted by backports.org to get this 2.6 kernel and updated versions of Spamassasin/Apache/bind/Samba. Not sure how these newer apps will perform with so little memory (i.e. if they require more memory than the versions currently shipping in Debian stable).

Re:2.6 just didn't wow me

[evilviper]

I'm not exactly trying to run uber-apps here (Abiword and Firebird mostly).

Firebird (or mozilla) is most definately an "uber-app"... It uses absolutely massive ammounts of CPU power, and is not all that light on RAM usage or disk I/O either.

Real world vs. fanboy fantasies

[Anonymous Coward]

I am what most people would consider a highly trained technical professional. Unlike most people who spout off at this site, I have the certificates to prove this, and furthermore they're issued by the biggest software company in existence.

I know how to tell facts from marketing fluff. Now, here are the facts as they're found by SEVERAL INDEPENDENT RESEARCH INSTITUTES:

Expenses for file-server workloads under Windows, compared to LinuxOS:

• Staffing expenses were 33.5% better.
  
• Training costs were 32.3% better.
  

They compared Microsofts IIS to the Linux 7.0 webserver. For Windows, the cost was only:

• $40.25 per megabit of throughput per second.
  
• $1.79 per peak request per second.
  

Application development and support costs for Windows compared to an opensores solution like J2EE:

• 28.2% less for large enterprises.
  
• 25.0% less for medium organizations.
  

A full Windows installation, compared to installing Linux, on an Enterprise Server boxen:

• Is nearly three hours faster.
  
• Requires 77% fewer steps.
  

Compared to the best known opensores webserver "Red Hat", Microsoft IIS:

• Has 276% better peak performance for static transactions.
  
• Has 63% better peak performance for dynamic content.
  

These are hard numbers and 100% FACTS! There are several more where these came from.

Who do you think we professionals trust more?
Reliable companies with tried and tested products, or that bedroom coder Thorwalds who publicly admits that he is in fact A HACKER???

--
Copyright (c) 2004 Mike Bouma, MCSE, MCDST, MS Office Specialist

Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.2
or any later version published by the Free Software Foundation;
with no Invariant Sections, no Front-Cover Texts, and no Back-Cover
Texts. A copy of the license is included in the section entitled "GNU
Free Documentation License".

Re:Real world vs. fanboy fantasies

[Anonymous Coward]

And who paid for that research, likely under the stipulation that they wouldn't pay for it if it didn't show results that Microsoft liked?

Microsoft.

You're living in a fantasy land if you think IIS is cheaper than or faster than apache 2.0.x

Re:Real world vs. fanboy fantasies

[mabinogi]

Its my experience that "highly trained technical professionals" would have enough real world experience to know the sort of crap code that can come out of any company...no matter how much money you throw at it, and that there's nothing inherently better about a product that costs money. In fact, it's been my experience that the more you pay for something, the crappier it is *cough*legato*cough*

Also, professional you may be (as in you get payed to do it), but you have very little professionalism if you have to resort to using terms like "open sores", not to mention the fact that you haven't got a clue what you're talking about due to your use of terms like "the Linux 7.0 webserver", "LinuxOS", and reference to RedHat as a web server, as well as referring to J2EE as an open source product, instead of the specification implemented by many proprietry application servers by companies such as IBM, Oracle, Borland and BEA.

I'm not even sure why I'm replying to this...it might as well have been generated by a script...

Re:Real world vs. fanboy fantasies

[ScrewMaster]

Fortunately, whoever moderated you "+2 Funny" actually has a sense of humor. The rest of the anti-Bill kneejerkers apparently don't. "Opensores solution" and "bedroom-coder Thorwalds", indeed. Good one.

Re:Real world vs. fanboy fantasies

[Anonymous Coward]

> I'm not even sure why I'm replying to this...it might as well have been generated by a script...

...or a clever troll (winkwink-nodnod)

Re:Real world vs. fanboy fantasies

[JPriest]

It was a joke that went way over your head.

For Windows, the cost was only:
$40.25 per megabit of throughput per second.
$1.79 per peak request per second.

Re:Real world vs. fanboy fantasies

[WhiteDeath]

A full Windows installation, compared to installing Linux, on an Enterprise Server boxen:

* Is nearly three hours faster.
* Requires 77% fewer steps.

Three hours faster on enterprise hardware?
Well compared to the complete install of slackware 9 (with everything) I did yesterday that would mean you can get a fully operational windows box connected to the internet, running proxy server, sharing postscript printers, talking to both MacOS and Windows 98, serving web documents, remotely configurable and ready for you to walk out the door in about zero minutes. I'm impressed!.

Mind you there were only a few actual user accounts to set up, and it was just an old pentium 3 500Mhz with 128M RAM and a 20G drive - not exactly your current "enterprise class" machine.
Something modern would likey have taken half the time.

Actually, I had allocated most of the day just in case - so I spent a few hours upgrading and tweaking a few things.

Re:Real world vs. fanboy fantasies

[tasinet]

sorry for double posting but i think it is VERY WORTH NOTICING of "..SEVERAL INDEPENDENT RESEARCH INSTITUTES:.." in this post and "A 2002 Microsoft-sponsored study of total costs of ownership over five years for working corporate infrastructure in North America shows that lower staffing expenses are a large part of an 11-22% cost advantage for Windows. For file-server workloads in particular: " in this link if its still a joke then i is gat no humour. sorry boys'n'girls.

Re:Real world vs. fanboy fantasies

[rossz]

Unlike most people who spout off at this site, I have the certificates to prove this, and furthermore they're issued by the biggest software company in existence.

Mike Bouma, MCSE, MCDST, MS Office Specialist

You left off the C&C so you owe me a new keyboard.

Re:Real world vs. fanboy fantasies

[rdean400]

Certification equates to neither knowledge nor wisdom. Certification equates to knowing enough about the test (subject matter and/or methods) to get a passing score. It doesn't say you know it, at all...it suggests that you knew it at the time you took the test.

You can go on and on about these supposedly independent research reports, but they're all tainted by their funding source...Microsoft. Apparently your certifications didn't teach you the importance of objectivity in research. No one should trust a research report funded by the beneficiary of that research. It's asinine to do so.

In the case of the Forrester study comparing J2EE development on Linux to Windows .net development, the findings were tainted by an extremely small sample size and a critical methodology errors: relying solely on interviews to come up with the findings. The report would have been more meaningful if it had taken several teams of programmers flent in their respective paradigms and had them come up with the solution to the problem.

As far as installation goes, any research report that suggests that is evidently not using a recent (within last 2 years) version of Linux. 3 hours faster and 77% fewer steps must have been comparing Windows 2k3 server on a brand new P4 chip versus installing Mandrake 5.2 on a P2-233.

Just because you're certified doesn't make you an expert. Just because these firms are "independent" doesn't make them objective.

Re:Real world vs. fanboy fantasies

[rdean400]

Unfortunately, "independent" != "objective".

Microsoft funded the research. Microsoft is the beneficiary of the results of the research. Conflict of interest? Duh!

Re:Real world vs. fanboy fantasies

[garbagedisposal]

_No_one_ believes propaganda paid for monopoly$oft.

If _you_ were are _such_ a professional who is _so_ confident of your conclusions why are you so desperate to push them here?

Because _YOU_ are a fanboy watching your precious monopoly$soft retreat in the face of people power.

The most hated & unjust dictatorships have fallen when people reject propaganda and think fpr themselves.

Re:Real world vs. fanboy fantasies

[Anonymous Coward]

You forgot billions of dollars lost due to code red infection... Linux=0. How many other IIS virii? Don't go spouting of the usual crap about Microsoft being the bigger target. Apache has double their market share and it's climbing while IIS is falling, and yet MS has double the vulnerabilities in their product.

Number of times my company has been shut down or had a virus replicate through my groupwise system. Zero. Number of companies I personally know people at who'se e-mail servers were shut down due to virus overload... too many to count.

I run 100% MS on the desktop. Linux isn't there yet, and doesn't pretend to be. I have both MS SQL and mysql servers in production. Novell runs my file/print/e-mail. Apache/Linux runs my web servers.

I have a couple of MS Sharepoint 2.0 sites running by special request from one group. It's actually not too bad of a product once you get over the glaring bugs it has.

The fact that you believe MIcrosoft paid for studies is pathetic. I wouldn't believe a study proving linux superior if it was paid for by Red Hat.

FYI I know dozens of MCSE's who did nothing other than study cram exams online and pass the tests. I know of 2 that did it just to say they could, and neither of them can install windows.

I've seen Novell CNE's that couldn't properly set up a Netware box if their life depended on it, and don't even get me started on the Lotus notes people...

People who try and list as many certs as they can in their sig's are about the same as people who drive flashy expensive sports cars... They are usually trying to make up for some other inadequacy.

Re:Real world vs. fanboy fantasies

[Felinoid]

I am what most people would consider a highly trained technical professional.
People have similer delusions of me as well.
"Oh my goddess your so smart" (After giggling the mouse after the screen saver has turnned the screen off)

Unlike most people who spout off at this site, I have the certificates to prove this, and furthermore they're issued by the biggest software company in existence.

I also have certificates and unlike yours mine is issued by someone who is actually conserned about my ability to preform the job and not intrested in pumping me full of hype.
(This is as much an issue with Red Hat and Sun as it is with Microsoft.)

I've found the installation time of Linux and Windows depends entirely on the trainning and skill of the person doing the installation.

I remember back when Slashdot sponsored a Linux installation race. You had to install Linux on a laptop get online and view Slashdot.
(See Geeks in Space for more info)

Who do you think we professionals trust more?
I'm sure you'd think the answer is "Windows" but a lot of Slashdotters would say Linux.

Actually it's nither...
Who do professionsl trust?
Ab-so-freaking-nobody.

MS Office Specialist
Your joking... That makes you a gloifyed secretary not a technology expert.

I mean with all your ranting of being a profesional I expected a System admin or maybe a programmer at the very least someone in tech support or maybe a technology consultent.

Me I'm a certifed Linux admin...
Actually my cert expired and I'm retrainning to get multipul certifications this time.

Something everyone should pay specal note not only to the qualification level of myself and this guy but to something far far more important...

I am a LINUX Admin... he is a Microsoft cert.
What dose that mean exactly?
Every time someone installs Open Office he loses a potental employer.
Every time someone installs Windows Server eddition I lose a potental employer.

I think you'll find this runs pritty much for any given cert.. The ones who have multicerts in multiplatforms (not multi-certs in one platform) are the ones you can trust to tell you the real deal..

Me I'm "my wallet" bies and so is everyone else.

Re:Real world vs. fanboy fantasies

[uarch]

"...highly trained technical professional"
"I have the certificates to prove this"

Using the phrase highly trained professional in conjunction with an MS cert is nearly worth +5 funny by itself. ;)

Re:Real world vs. fanboy fantasies

[Felinoid]

Because _YOU_ are a fanboy watching your precious monopoly$soft retreat in the face of people power.

No he's a Microsoft Office expert who is seeing himself out of work soon if companys continue to addopt Open Office.

Re:Real world vs. fanboy fantasies

[nmos]

I am what most people would consider a highly trained technical professional. Unlike most people who spout off at this site, I have the certificates to prove this, and furthermore they're issued by the biggest software company in existence.

And with such an impressive resume the best you could do was to copy one of MS's marketing pages. Did you get your certs via copy/paste too?

I know how to tell facts from marketing fluff. Now, here are the facts as they're found by SEVERAL INDEPENDENT RESEARCH INSTITUTES:

Who were PAID BY MS for their work.

Who do you think we professionals trust more?

Probably not someone who pulls some marketing tripe off a MS web page and tries to present it as independant fact.

Re:Real world vs. fanboy fantasies

[Felinoid]

It's ok.. It's totally a prospective thing trust me on this.
Your thinking with your wallet and your prospective is bent becouse you spend your time with remarkably impressive Microsoft professionals. Well impressive to you anyway.

I'm going to guess that your work has been directly effected by the addoption of open office and/or Linux.
From your experence this addoption of Linux seams utterly rediculous. Your what many people consider to be a computer expert so your thinking "how could I be wrong?".
What your not aware of is that most Slashdoters have a considerable amount of experence in Linux and Windows. Most of us dual boot and that means we have gone through the installation process for both Linux and Windows.

Your talking about something you know nothing about to a croud who knows better flashing credentals that have nothing to do with the matter at hand.

Your a scientist who's never seen daylight telling a bunch of people who work under the hot sun all day that the sky is yellow.
It's not only funny it's absolutly hillareous.

PS. I'm only guessing your the same person becouse your appologising for dubble posting..

Re:Real world vs. fanboy fantasies

[sg_oneill]

Who do you think we professionals trust more?
Reliable companies with tried and tested products, or that bedroom coder Thorwalds who publicly admits that he is in fact A HACKER???


Hats off to you good sir! The funniest troll I've seen in a while. Actually had me about ready to engage in a shouting fit till that last line. Hilarious.

Putting the post under a GPL, genius!

Re:Real world vs. fanboy fantasies

[stfvon007]

Microsoft had an "independant" tester do some tests, and even they found the total costs for a web server were LESS for linux. (other services were however found to cost more in total cost however)

Re:Real world vs. fanboy fantasies

[ogre57]

Must .. resist ..

A full Windows installation, compared to installing Linux ..

• Is nearly three hours faster.

Please stay calm. Some nice men in white coats will be here aaanny minute now. They're going to give you a pretty jacket with extra-long sleeves, take your for a niice relaxing ride. Okay? ("nearly three hours faster" .. sure. Full Linux server install I did last week took maybe 30 minutes. Otoh maybe there's some special Dr Who version. :)) )

Re:Real world vs. fanboy fantasies

[Cobron]

Sarcasm just goes WAY over your head, doesn't it?

Re:Real world vs. fanboy fantasies

[SEE]

In my experience, "highly trained technical professionals" would have enough real world experience to know that it was a troll.

Re:Real world vs. fanboy fantasies

[mabinogi]

probably...that's what I get for skim reading ;)

I knew it was trollish, hence my "I don't know why I'm replying" comment, but re-reading showed just how ridiculous it was...oh well...can't take it back now.

Re:Real world vs. fanboy fantasies

[kubrick]

Actually, Mike Bouma has a Slashdot account already.

I know this is just a stupid troll... just wondering why the AC picked this user in particular. Are those bitter, twisted post-Amiga politics coming into the picture?

Re:Real world vs. fanboy fantasies

[Nexx]

And just because you read slashdot, doesn't mean you have a sense of humour? :)

Re:Real world vs. fanboy fantasies

[tasinet]

PS. I'm only guessing your the same person becouse your appologising for dubble posting..

are you drunk?
from the main page of this article:

"A copy of the license is included in the section entitled "GNU
Free Documentation License".

[ Reply to This | Parent ]

did I hear JOKE? by tasinet (Score:1) Sunday February 01, @04:14AM
1 reply beneath your current threshold.
Re:Real world vs. fanboy fantasies by tasinet (Score:1) Sunday February 01, @04:16AM "

I double posted info on the link from microsoft.com because on the second post i wanted to emphasize on the "independent research" vs. "Microsoft-sponsored study".
cool?

Re:Real world vs. fanboy fantasies

[SoTuA]

shame on you... I mean, a M$ "cerified" highly-yadayada fanboi posting under a GNU license? :D

Slashdot: I came for the news, stayed for the merciless finger-pointing when one screws up.

Re:Real world vs. fanboy fantasies

[Dausha]

Compared to the best known opensores webserver "Red Hat", Microsoft IIS: * Has 276% better peak performance for static transactions.

But, IIRC, this is because MS cheats. When IE browsers confront an IIS server, they skip the usual handshake standards and immediately go to work. I believe the difference is: "Hello. How are you? Do you have this page? Thanks." compared with "Just give me the damn page!"

Could it be that these independent research organizations were hoping for post-research investment by Billy Gates?

Re:Real world vs. fanboy fantasies

[rdean400]

Trolls love to see people blow gaskets...I didn't.

Re:Real world vs. fanboy fantasies

[Linker3000]

Fair point perhaps, but..

My PERL code for a Web-based testing engine:

Win 2K/IIS: Uptime = approx 3 days before IIS hangs CPU with 100% utilisation.
Red Hat 9 + Apache: Uptime = approx 3 months now.

Oh, you want real-world benchmarks with numbers eh? OK...

MS Solution: NIL
Linux Solution: ONE

Re:Real world vs. fanboy fantasies

[DJPaddy]

Quoting: "Copyright (c) 2004 Mike Bouma, MCSE, MCDST, MS Office Specialist" MS Office Specialist ? Isn't this man ashamed to put THAT on his signature after stating that he is 'a highly trained technical professional' ? Hell, I will put this on a tag and call my secretary with the M.O.U.S. certification 'highly trained professional' as well !!! (Still Can't believe the nerve of those guys)

Re:Real world vs. fanboy fantasies

[Da Masta]

Trolls love to see people blow gaskets...I didn't.

Which says nothing of your ability to spot trolls, only that you weren't the best catch this troll baited. :-p

Re:Real world vs. fanboy fantasies

[TheJMan]

You sir are a TOOL of the highest order. Considering a FULL install on a halfway decent box takes about 1 hour (including me forgetting about it and not switching CD's) I would have to say that any operating system that installs itself 2 hours *BEFORE* I start is probably some sort of VIRUS.

Re:Real world vs. fanboy fantasies

[rofthorax]

Who has more to lose if linux is right..

You, so why should we believe you..

We don't..

And why are you concerned..

Furthermore where is the source of your statistics, what tests did they do? You could take any marketing group in the world and find someone who would give you favorable results.. But if you don't give references, its not very useful.. The reason so many people are led astray by Microsoft is they are too gullable, they are led to believe in simple advertising ploys.. Like the one above..

Furthermore you sound like a paid evangelist, a postion that would not have existed without Microsoft.. Paid actors..

Maybe some big companies don't recognize the difference, but those who do are on the road to independence.. And your special knowledge about such is obsoleted.. You stand to lose.. Its obvious..

Re:Real world vs. fanboy fantasies

[ACPosterChild]

Why can't we just shoot some people, like the numb-fucking-nuts who modded this Insightful?

But then, I guess a good troll is one indestinguishable from true stupidity.

Re:Real world vs. fanboy fantasies

[LA_Samurai]

I also am a "highly trained technical professional" with the certs that he's got. I, OTOH, have the experience to know that those so-called independent research institutes findings are far from reality. My IIS4 server crashed every time there was a virus outbreak--and it HAD auto updating anti-virus by a renowned AV company AND I had kept it abreast of MS updates. After I switched to Apache on Redhat 7.3, it's been running constantly without manual intervention for the last two years or so (there were planned maintenances). Three-hour Linux installation??? I've installed RedHat 5.2 thru 9 and Fedora Core 1, Mandrake 9.2 and SUSE 9 Pro. On a re-/de-cent machine (P4 1.7Ghz/AMD XP 1700), none of these distro takes more than 30 minutes to install. Windows (2000/XP), OTOH, takes over 45 minutes, no matter what. Out-of-the-box security is way better on Linux than Windows, too. In my role as CIO, I've interviewed and hired a number of sys admins and other techs. One of the questions I ask is the SMTP port number. Many MCSE's couldn't tell me. It's obvious that so many people are paper MCSE's. As an MCSE myself, I find it unconscionable that MS has allowed the de-valuation of their top cert. As certs go, I really believe that exam-based certs have no value. They should institute a more professional certification process where a candidate should follow a prescribed set of training AND work experience before being allowed to take the exams. As it is, a twelve year old can take and pass the MCSE exams without too much trouble. BTW, MCSE is a misnomer. System Engineer? Come on. Would I allow an MCSE to put his "engineering" skills on a system? No way. I don't know of any MCSE (including myself) who can "engineer" his way out of a downed system. I had to laugh when I read the article. It's hard to believe that there still are computer "experts" who spew out such junks.

Re:Who cares? OS X kicks both.

[be-fan]

Actually the OS X kernel is about half as fast in lmbench (UNIX benchmark) as Linux 2.4. The OS X kernel is really antiquated. Much of it is 4.4BSD and Mach code. The GUI is modern, but the guts are ancient.

This can mean two things...

[SharpFang]

Maybe one of them, maybe both...

1) The new kernel is really very good.

2) The old kernel is really very bad.

Really, if such huge increase was possible, there must have been a lot of room for it. If you face a really well written program, you have a hard time to speed it up by 5%. If you can speed it up by 50% without loss in other domains, it must have been seriously flawed.

Yeah, mod me flamebait. But first think if I'm really wrong.

Re:This can mean two things...

[Anonymous Coward]

You might be right, but lets take the analogy a bit further.

1) the old kernel was comparable to other OS

2) other OS kernels are equally bad as 2.4

I don't doubt there were serious issues with linux in the past, but it was comparable to other operating systems. This would indicate other systems can now learn from linux and improve their performance.

Re:This can mean two things...

[natmsincome.com]

Most of it has to do with the algorithms. For example I can write a really optimized bubble search but if you write an un optimized quick sort it will still be faster than my optimized bubble search. That being said any newbie programmer can do a bubble search but you have to know what you're doing to do a quick/merge sort.

So while 2.4 wasn't using the best solutions it was better than nothing. It's always better to have bad working code than great code that doesn't work. Hurd is a great example. It may be batter but it doesn't work (well enough for me anyway) yet so who cares.

IDE is another example. If I remember correctly 2.2 didn't have DMA support but it worked adding DMA makes it much faster but it would have made it more unstable if they added it at the beginning.

The last thing that you have to remember is that lots of the changes were taking advantage of features in the newer hardware. If you ran the same test on 486 you wouldn't get the same results as you'd have different bottlenecks. In another 10 years we'll get the same thing again. The might make it so that the bus to the memory is as faster as the level 2 cache on the CPU. If they do that they'll have to make big changes to the OS to get rid of the new bottle necks and you'd increase the speed by another 50% or maybe even 100%

Anyway that's enough ranting.

Re:This can mean two things...

[be-fan]

A kernel is much more complex than a single program. A program usually does one thing, and once you've optimized that, that's all there really is. In contrast, a kernel does all sorts of things. 2.2 was good for small-scale servers. 2.4 was good for mid-range servers. 2.6 is good for larger servers and desktop machines. 2.8 is supposed to get improvements to make it better on desktops and on huge NUMA machines. Linux has always been a fast kernel for what it did, its just that its doing a lot more today than it did a few years ago.

Re:This can mean two things...

[iabervon]

The old kernel had a lot of room for improvement on the systems they tested; but that's primarily due to the fact that the systems are substantially newer than the 2.4 series. A 2 GB dual Xeon system running 2.4 isn't going to use the processors efficiently (hyperthreading, imprecise locks), and isn't going to deal with the memory effectively. It was in part to take advantage of the availablity of such systems that the changes for 2.6 were made.

New conditions require new optimizations and new designs; a good program optimized for a set of inputs which are common at one time may be really inefficient at handling inputs that become common later. Sure, you can make a program that's good for both sorts of input, but it doesn't make sense to try to do so until someone actually has such an input to test with.

Re:This can mean two things...

[bonch]

BSD and other people were saying it for years, and you point it out. But, now that 2.6 has sped things up finally, people will pretend things were never bad before and call you a "troll" if you point out that for a long, long time, Linux kernel performance really WASN'T all that it was hyped up to be by Slashdotters. Selective memory is a plague around these parts...

Re:This can mean two things...

[ealar dlanvuli]

I suppose I should specify what I meant, that was a little light on details and I don't know what your knowledge base is.

I should preface this with it's been a while since I studied this, so if any numbers are wrong please correct me. I shall attempt to be as accurate as possible.

A proper implementation of qsort that chooses a bad pivot value will achieve O(n^2) performance. It is effectively a bubble sort (I believe, it might be an insertion sort... can't remember). If you have flaws in your pivot value selection this is the fastest way to guarntee a bad qsort implementation.

A programmer who doesn't get recursion well can easily make far too many function calls. This is a flaw with all "unoptimized" recursive functions unless the programmer knows what he is doing. Qsort can be turned into a loop trivially, and most people opt to do this (it's generally faster on modern cpu's anyway), but a programmer working without documentation who doesn't do any optimization could easily choose a bad loop exit condition (though it's not as likely as messing up a recursive implementation) leading to wasted recursions.

To wrap up: the biggest problem with an unoptimized qsort implementation is going to be bad piviot value selection; however, due to the complexity of the algorithm a bad programmer can easily make it even worse than a bad piviot selection by itself.

And I'm too lazy to go back and edit my grammer/spelling mistakes.

Re:This can mean two things...

[kasperd]

If I remember correctly 2.2 didn't have DMA support

I actually have some 2.2 kernels running with DMA on IDE drives. It didn't get enabled automatically though. I had to change some configuration files so hdparm would be used to enable DMA during boot. In one of those machines we just installed a new harddisk a few days ago. I couldn't get DMA to work on the new harddisk eventhough it worked on the old disk on the same controler. I tried booting a 2.4 kernel, then I could do DMA on both disks. Enabling DMA on the new disk improved throughput from 2MB/s to 26MB/s.

Re:This can mean two things...

[stor]

for a long, long time, Linux kernel performance really WASN'T all that it was hyped up to be by Slashdotters.

*jaw drops* No... way!

Cheers
Stor

Naked eye test

[Zordas]

Yesterday I started a new Gentoo install with the 2.6.1 kernel. I used GCC 3.3.2 and glibc 2.3.2 with NPTL support. I have to admit, the naked eye can see a major diferance with the new kernel. With my XP computer and the new gentoo install (exact same computers .. P4 512MB) I ran a simple boot up and lanch a web browser test. And supprise supprise, Gnome is screamming fast. I had already booted and opened up mozilla 1.6 befor xp was even done booting! Also, simple stuff like opening up email, browsing, etc. is all noticable faster than XP. Soo... before I get slammed by the XP folks.. my XP box was also a clean install. (yes, I have no life!) I am happy to say I am one step closer to completely weening myself off of windows XP.

Re:Naked eye test

[boudie]

And on Gentoo, you can also use udev, which takes a little effort, but works pretty good. This will really improve the integration of any hot-pluggable devices on Linux. Things are really happening with Linux these days.

Re:Naked eye test

[Zordas]

Agreed.. Currently I'm using devfs,pty and hotplug. I did read the doc's on udev and was a little miffed about how to actually install it. But since the Gentoo forums and pretty good, I'll more than likely installl it in the next few day's. This (according to the doc's, will speed up my computer even faster).

Re:Naked eye test

[akb]

A boot up test tells me zero about the overall performance improvements in the kernel.

Re:Naked eye test

[Zordas]

Agreed. The actual (tecnical tests) were preformed by the link provided by the parent poster. I was mearly conducting a naked eye test on booting, accessing email, web browsing, etc. This is what the "Normal computer user" cares about. Let's face it. even though you and myself are techies, most computer users are not. All they care about is how fast a computer responds to commands. That is what I was doing. Although today I'll be setting up a large database using db 4.x.x and compairing it to access XP. If you like I can I can email my results to you.

Re:Where's the distros

[myowntrueself]

"If you're running stable, you shouldn't be running a 2.6 kernel anyway"

If you're running (debian) stable, you shouldn't be running anything thats not like 100 years old *anyway* :-P

(I keep getting caught out by debian stable packages that are *way* out of date -- should be called 'debian *conservative*').

Article text

[Anonymous Coward]

Raw c&p, please forgive random spam.

Linux v2.6 scales the enterprise
Bigger, stronger kernel sizzles in our performance tests

By Paul Venezia January 30, 2004

If commercial Unix vendors weren't already worried about Linux, they should be now. Linux has seen wide deployment in datacenters, generally as a Web server or a file server, or to handle network tasks such as DNS and DHCP, but not as a platform for running mission-critical enterprise applications. Solaris, AIX, or HP/UX typically get the nod when an application demands the highest levels of performance and scalability. The recent release of a new Linux kernel, v2.6, promises to change that.

ADVERTISEMENT

DOWNLOAD PDF

Click here to download the InfoWorld Special Report on the 25 most influential CTO's of 2003.

Brought to you by Qualys

DOWNLOAD PDF

Click here to download InfoWorld's special report Linux looms large

RELATED LINKS
Compiling the v2.6 kernel
How the Linux kernel gets built - Infoworld Staff
Platforms RSS feed
>> MORE

IDG ENTERPRISE NETWORK
Sun preps second version of Java desktop (ComputerWorld)
Draft EC ruling finds Microsoft guilty in antitrust case (ComputerWorld)
EC circulating final draft of antitrust ruling in Microsoft case (ComputerWorld)

TOP STORIES
Oracle to cut price on lower-end database
Oracle's U.S. midmarket apps bundle delayed
PeopleSoft pushes forward meeting for board vote
Top News RSS feed
>> MORE

TOP SITE REFERRALS

Microsoft: change to IE will block some Web URLs.
(Secure IT)

Changes in IE's URL handling
(Information Management Weblog)

SIGN UP FOR THE MAGAZINE
FREE EMAIL NEWSLETTERS
IT SOLUTION SEARCH

The v2.6 kernel ushers in a new era of support for big iron with big workloads, opening the door for Linux to handle the most demanding tasks that are currently handled by Solaris, AIX, or HP/UX. The new kernel not only supports greater amounts of RAM and a higher processor count, but the core of device management has changed. Previous to this kernel there were limits within the kernel that could constrain large systems, such as a 65,536 process limit before rollover, and 256 devices per chain. The v2.6 kernel moves well beyond these limitations, and it includes support for some of the largest server architectures around.

Will the new Linux really perform in the same league as the big boys? To find out, I put the v2.6.0 kernel through several real-world performance tests, comparing its file server, database server, and Web server performance with a recent v2.4 series kernel, v2.4.23.

Linux Meets Big Iron

A primary focus of the v2.6 kernel is large server architectures. Support for up to 64GB RAM in paged mode, the ability to address file systems larger than 2TB, and support for 64 CPUs in x86-based SMP systems brings this kernel and Linux into the more rarified air of truly mission-critical systems. The included support for NUMA (Non-Uniform Memory Access) systems; a next-generation SMP architecture; and PAE (Physical Address Extensions), providing support for up to 64GB of RAM on 32-bit systems, is also new.

There is much more to v2.6 than just bigger numbers in processor and RAM counts, however. This kernel breaks apart some of the artificial limitations that have been present in Linux from the beginning, such as the number of addressable devices and total available PIDs (Processor Identifiers). The v2.4 kernel supported 255 major devices with 255 minor numbers. (For example, a volume on a SCSI disk located at /dev/sda3 has a major number of 8, since it's a SCSI device, and a minor number of 3.) On servers with a large number of real or virtual devices, device allocation can become problematic. The v2.6 kernel addresses these issues in a big way, moving to 4,096 major devices with more than one million subdevices per

Re:Where's the distros

[salimma]

mass-distribution before we even hit 2.6.2 might be a BIG mistake

Won't be a long wait; 2.6.2-rc3 is out now.

FC2 test1 should be out next week with a post-2.6.1 kernel as default. With SELinux to boot, though it's recommended to disable it at boot time for test1. Mandrake 10.0 beta1 has 2.6.1 too.

ahem...

[Apreche]

2.4 is the old and busted
2.6 is the new hotness

Re:Who cares? OS X kicks both.

[Anonymous Coward]

Oh yeah? Well, you're locking yourself into commodity hardware and a cycle of free software upgrades!

Hmm... that comeback doesn't seem as potent as I thought it would be. Oh, wait, I know! You're locking yourself into an eternity of vigilance.

That word means something else

[dbIII]

OS X is far better than either "kernel". With OS X you get a better GUI

Gnome, KDE etc is not part of the kernel - the kernel talks to the hardware, then programs talk to it. Making comparisons like this is like comparing an entire ferrari to a spark plug - it makes no sense at all.

OSX is very nice - but you don't buy it for speed.

It's the tool for the job - palm pilots run at a low speed with low memory and low screen resolution, but do the job very well. After playing with OSX for a while, I would recommend a mac for anyone that hasn't already used windows for years, but for servers on PC hardware or my own use I use linux.

Re:Where's the distros

[shadowbearer]

The only place I personally noticed the umount problem was with cdroms (part of the new ATA cdrom code?). I could umount any drive but cdroms, and the kernel would hang on reboot (not shutdown, weird) while trying to umount a mounted cdrom filesystem.

Otherwise everything else seemed to umount ok.

Otherwise I've been running 2.6 since -test3, and other than a few minor annoyances (framebuffer, cdrom umount, cd burner recognition) it's been fabulously stable and very fast.

SB

Re:Where's the distros

[MikeCapone]

The only really good distro supporting 2.6 soon is Fedora.

Slackare 9.1, f.ex, is "2.6.x ready".

Just download the source, compile and voila! No need to fiddle with the new modutils and such.

Why wait for distros to do it for you?

Re:interesting hardware comp

[Rob Riggs]

does it seem that the opteron smokes the itanium2 and xeon? this can't bode well for intel.

The comparisons won't make much real-world sense until the evaluation is done using Intel's compiler for the Itanium tests. The GNU compiler is just not up to snuff at optimizing for Itanium's EPIC instruction set.

I would like to see Intel contribute Itanium optimizations to GNU, but I doubt this will happen since they sell a competing product.

Re:Where's the distros

[Akai]

The 2.6 official was not available at the last release cycle of the major distros (For example Mandrake 9.2 includes support for a 2.6pre kernel for those who wanted to live on the edge, but defaulted to 2.4)

Mandrake 10, which is in beta phase 1 right now is 2.6, I imagine Suse will follow as well.

How does this compare to the BSDs?

[FFFish]

THIS IS NOT AN INVITATION TO A FLAME WAR.

Does anyone have factual comparisons of a reasonably-tweaked Linux (2.6 kernel) with a reasonably-tweaked [x]BSD (whatever kernel)?

Re:How does this compare to the BSDs?

[Atzanteol]

I would love to see this. To the best of my knowledge, nobody ever really does performance comparisons between *BSD's and Linux. It should be reasonably easy to put them on an even ground (all the same tools/apps are available for each).

Re:How does this compare to the BSDs?

[Josh Booth]

An older /. article did directly compare *BSD and Linux, but it really had only benchmarks of specific system calls, not much in the way of real application performance. However, it did show that, while Linux 2.6 was mostly O(1) and at most O(n), the BSDs were mostly O(n) or more, only occasionally being O(1). I think NetBSD was the worst offender. But then again it had few or no real world benchmarks.

Re:How does this compare to the BSDs?

[tarm]

Try this link.

It's an excellent comparison.

Re:How does this compare to the BSDs?

[Argon]

This one IS an invitation to a flame war.

Come on. Who are you kidding ;-).

An idea...

[MikeCapone]

I haven't noticed *any* differences on the desktop.

Either you have a crazy uber-fast box or just adapted and got used to it too fast; personally I feel the difference quite a bit on my old K6-2 450mhz.

Maybe you should try booting into your old kernel (you did keep it for backup purposes, right?); maybe you'll notice the difference.

Singing to the Choir.

[tealover]

I wonder what the response would be if someone posted similar numbers about Microsoft's next OS. I'm sure they'd find creative ways to diminish the results.

One thing I've learned is not to take tech writers too seriously. Most of them are writers for a reason.

MOD PARENT UP

[chendo]

Parent post contains article text.

Re:interesting hardware comp

[jarich]

I bought a dual Opteron early on and this numbers are exactly what I have seen when benching against a Xeons... I haven't run against the Itaniums though.

I gotta point out the difference in price between the platforms. Opterons are faster and cheaper.

Mandrake

[Anonymous Coward]

I have tried out 2.6 on Mandrake 9.2 and I must say that for desktop use I can't really tell any difference between 2.6 and the 2.4 version that Mandrake is shipping.
I don't know if Mandrake is patching the 2.4 series, but it seems pretty snappy as it is.

Re:Where's the distros

[John Courtland]

You gotta change that link to "Obligatory Litigious Bastards Link". Google has managed to put some sort of stop on the google bombing of SCO. And maybe even Caldera I guess.

Wish we coulda seen RHEL 2.4 kernel in action

[Anonymous Coward]

The reviewer used RHEL but compiled vanilla 2.4 and 2.6 kernels for the test, not Redhat's. I certainly understand his reasons for doing so but I wish I could have seen how Redhat's custom halfbreed kernel would have performed. Has RedHat already squeezed most of 2.6's performance enhances into 2.4? I'd love to know as I run a number of RHEL boxes, maybe I'll start testing 2.6 on them to see what kind of performance differences there are.

Re:Wish we coulda seen RHEL 2.4 kernel in action

[inode_buddha]

I use RH with homebrew and "stock" kernels. Not sure how it compares in absolute numbers. My gut feeling (just desktop use) is that the -mm (Andrew Morton) kernels are currently a tad quicker, but RH patches their kernels like crazy for better HW support and misc bugfixes.

I'd like to see how much of -mm ends up in the RH kernels in a year or so. I did try 2.4.20 with the -ck patches; 2.6 still felt better. Dunno if the info helps at all, but figured I'd mention it.

FWIW, 2 machines here:
Dual PPro at 200 MHz, 1Mb caches, 128 megs EDO
Dual P3 Coppermines at 1GHz, 1024 megs PC133.

Re:fuck debian

[slurpburp]

yeah, I know. Debian is great and all, but long live Gentoo.

Re:Where's the distros

[shadowbearer]

Thanks, yeah, I do. Still works with www.caldera.com, tho I doubt that that will last much longer.

SB

Re:Where's the distros

[chill]

All the issues with the early 2.4 kernels were taken to heart and ended up spawing projects like the Linux Stability Project (http://www.osdl.org/projects/26lnxstblztn/results /).

Because of all this attention, 2.6-pre were quite a bit more stable than the 2.4-pre ever was.

-Charles Hill

RPM for Fedora (Yum) ??

[justanyone]

So, I like the idea. Thus I'm looking for an RPM for Fedora Core 1 on Athalon, and/or a Yum repository-site that I can use for same.

Any hints out there? 2.6 sounds nice, I'd like to use it...

Re:RPM for Fedora (Yum) ??

[inode_buddha]

ftp ://ftp.dc.aleron.net/pub/linux/fedora/linux/core/d evelopment

Re:RPM for Fedora (Yum) ??

[Anonymous Coward]

http://people.redhat.com/arjanv/2.6/readme.txt

your sig

[morgajel]

...in response to your signature:
Actually, one of the original storylines for the simpsons was supposed to be about a boy who completely disrespected his father yet idolized a clown who looked exactly like his father.
I heard this in an interview on NPR with Matt Groening(sp?).
They later decided to drop that storyline, but the remnants are still there- the looks, the attitude towards krusty and homer.
If you ever get a chance, I highly suggest listening to that bit- you can probably find it on NPR's website.

(note to mods: don't bother voting this overrated or offtopic- it is what it is: informative. I didn't turn on my Karma Bonus so spend your mod points elsewhere.)

did I hear JOKE?

[tasinet]

i dont know if it's been posted already but check this. Aint no joke mister.

Re:did I hear JOKE?

[ogre57]

Well .. the claims on that page are among the most laughable absurdities on the web right now. Otoh, hand me enough millions and I too will sort thru thousands of companies to find the half-dozen or so needed to "prove" whatever "facts" you want.

Re:interesting hardware comp

[Anonymous Coward]

GCC is crap and there's no point contributing to it. Instead, Intel are doing the right thing by making Linux more portable: it now (2.6) builds with the Intel compiler, but it is now much easier to build the kernel with other compilers too, such as IBM's VisualAge C.

Speaking of a jump to conclusions mat...

[Thirty]

I'm not entirely sure that InfoWorld is generally a pro-Linux forum. If the article was written for /. (and not just posted here) then I might be inclined to agree with your skepticism. If the numbers were regarding MS's next OS, no one would have a goddamn clue if the numbers were accurate because no one would have access to the OS or the code (except maybe the Gartner group or some other bastion of testing integrity). These results appear to be pretty solid. If you think they aren't, you should be able to replicate the tests reasonably easily and see for yourself (the description of how he performed his testing was relatively clear). That's probably one of the coolest things about Linux and the rest of the Open Source movement. Anyone lying about it or exaggerating can be refuted pretty easily (unless you have magical asshole-SCO fairy dust in which case being refuted has no effect). -- Once I was in therapy for anger management. Then I realized that I liked being angry.

The root of all evil

[clasher]

Donald Knuth said "Premature optimization is the root of all evil"

Therefore I assume #1 is much more likely than #2.

It would seem as though the 2.4 focussed on getting a number of feature in the kernel while the 2.6 series allowed the developers to work towards making those new feature faster. Programming a new feature from scratch while also aiming to optimize it for speed can often lead to buggy code. Optimized code is rarely as straightforward and easy to debug as a more general (but slower) algorithm. When it comes to something as important as a kernel I'd much rather have clean, clear code which can later be optimized than a confusing kludge meant to squeeze out the last little bit of processing power.

Just my humble opinion

Re:The root of all evil

[GooberToo]

This guy is riding high on the clue train!

Toooot! Tooooot!

Welcome aboard! :)

It may be faster

[Molina the Bofh]

It may be faster, it may be smoother, but I bet it still won't be able to handle the full power of the terrible hordes of slashdotters.

Requirements?

[chunkwhite86]

Does anyone know the Linux system requirements for 2.6? I.e. what gcc, libc, etc. are required to run it?

Re:Requirements?

[The One KEA]

Check the Changes file in the kernel source. Basically any modern distro like Fedora Core 1, Mandrake 9.2, SuSE 9, Gentoo, Debian-unstable or whatever can all compile and boot 2.6 kernels without any real problems.

Re:Requirements?

[incuso]

In the ~linux-src/Documentation/Changes you can find all the tools that need to be upgraded (most of them only if you are actually using them). M.

Infoworld 2.6 test

[AhBeeDoi]

That article makes me want to run out and get a dual Opteron system. :^)

or whatever comes first

[dsb]

from trying to emerge -U world on my gentoo

Re:Where's the distros

[shaitand]

I'm trying out gentoo from stage1 now myself and it already includes a 2.6.1 version kernel. That among other things has me significantly impressed with gentoo.

Although I'm not sure I know anyone I'd recommend it to, the wait is only bearable to me after having built lfs systems ;)

Try IceWM

[LightStruk]

I had a very similar Gateway 2000 system (p 200, 64 MB RAM, STB Virge/VX video... blech) and I remember installing RH7 on it, and watching the computer constantly thrash the hard disk, trying to swap pages in and out.
One simple thing fixed it: I started using IceWM. Suddenly, Linux stopped thrashing the disk. Bliss.

Lest you think I'm an IceWM fanboy, I don't run it anymore; nowadays I'm a KDE 3 guy with lots of eyecandy, on a much more powerful system.

Re:Who cares? OS X kicks both.

[BandwidthHog]

What interests me is that both Linux and OS X are delivering serious *increases* in performance on the same hardware, most notably on older hardware. A certain other OS vendor who shall remain unnamed is not exactly known for that. If this keeps up for a few revisions of each OS, imagine what that'll do to the conventional wisdom.

Re:interesting hardware comp

[Hoser McMoose]

The hardware specifications weren't very complete, but from what I can see from IBM's x335 configuration they were using the no-L3 cache Xeons. A 3.2GHz Xeon with 1MB of L3 cache could easily boost the performance 10-20% over a 3.06GHz Xeon with no L3. Of course, the Opteron could still end up leading in a lot of the tests. What's more, the Opteron seems to really come into it's own in 4-processor configurations, where the Xeon scales poorly. In short, the Opteron is a heck of a good chip.

Where this really looks bad for Intel though is with their Itanium systems. Assuming that those 1.5GHz I2 processors are of the 6MB L3 cache variety, this is Intel's top-end chips. The servers probably won't have the performance of HP or SGI's I2 servers (IBM doesn't care much for the Itanium so they don't invest nearly as much time and effort in the designs as HP or SGI do), the chip still looks pretty weak.

Intel's saving grace here may be that the Itanium line of chips are VERY dependant on a good compiler, and chances are that these applications were compiled with GCC. Using Intel's ICC instead probably would boost performance by a noticeable margin, though a number of applications still won't compile with ICC from what I understand.

Very interesting link

[arevos]

Linux 2.6.0 looks pretty nice. :D

Re:Who cares? OS X kicks both.

[Anonymous Coward]

How is that informative? The kernel is Mach and modern *BSD, mostly FreeBSD from what I can tell - not ancient 4.4BSD. While in the latest benchmarks I've seen show Linux leading OS X in server benchmarks on the same hardware, the margin is smaller then the 2.4 > 2.6 margin in the above mentioned article. That is to say, OS X hardly sucks. It can easily keep pace, but not yet surpase. Maybe it never will, but to be even close with a Mach-based kernel is quite impressive. The Mac OS X kernel is called Xnu, and its a melding of of BSD and Mach - its no longer a pure micro-kernel and a lot of the weaknesses of Mach are offset by having many important bits, like TCP/IP, in kernel space rather then some floating server. I've seen those lmbench scores before, and the last ones I saw was comparing a rather old release of OS X to Linux. Try Panther, its noticably faster across the board.

"Redundant?"

[bonch]

How on earth was this post redundant? I guess it's okay to have endless "I installed 2.6, and things 'seem' faster!" stories, but one guy points out a lack of speed increase for him, and suddenly there's some sort of redundancy.

Devices no longer renamed....

[sakshale]

For instance, the first-seen SCSI device will remain as device sda, using the serial number of the device as an identifier regardless of the order in which it's found during a later boot.

Sweet! This was my one pet peeve when using Linux on a production server, such as a tape backup systems.

Re:Who cares? OS X kicks both.

[be-fan]

Read the OpenDarwin FAQ: here.

The code is mostly 4.4BSD-Lite2. There is a good thread on OSNews about OS X that goes into comparing the source code, and it supports the idea that, while there are many changes, most parts of the BSD subsystem is still plain 4.4BSD-Lite2 code.

Re:No, no, no... supports HT != HT-aware

[anthonyrcalgary]

ULE has specific optimizations for HT and NUMA.

http://www.chesapeake.net/~jroberson/ULE.pdf

Why don't these get modded offtopic

[bangular]

I may be the only one to feel this way... but haven't the completely unrelated SCO jokes gotten a little old? They are completely offtopic 99% of the time and most of them are just "Where do I send my 699 dollars".

Re:Who cares? OS X kicks both.

[nbdy]

The new panther is based on FreeBSD 5.0?

naked ear test

[anthonyrcalgary]

My music used to skip (2.4). My music no longer skips (2.6). Next question, please.

Re:interesting hardware comp

[bangular]

They would make perfect real world sense because almost no Linux distro is compiled with the Intel cc. They are almost all compiled with GCC.

Re:Where's the distros

[chickenmonger]

Beta 1 of Mandrake 10.0 gives you the choice of kernel 2.6.1 or 2.4.25pre6.

Re:HELP: Where can I browse Linux kernel sources?

[kasperd]

I'd love to browse around sources without having to download a tarball using only my web browser.

That is possible. The best place I know is The Linux Cross-Reference.

Re:Who cares? OS X kicks both.

[ealar dlanvuli]

That's why you log in as >console. No Quartz is fun.

What's with all these debian jokes?

[OverlordQ]

I mean it has nearly 8000 different packages. Each one has to be rigously tested for stable, and is for all sakes and purposes nearly 100% bugfree. ALl relavent security fixes are backported to stable.

If you want something more current try 'testing'. Testing means that things should not break as bad as in unstable or experimental distributions, because packages are allowed to enter this distribution only after a certain period of time has passed, and when they don't have any release-critical bugs filed against them. You can track these here. (More info here).

Finally if you want even more up-to-date, try unstable. Most of the development work that is done in Debian, is uploaded to this distribution. This distribution will never get released; instead, packages from it will propagate into testing and then into a real release. "sid" (codename for unstable) is subject to massive changes and in-place library updates. This can result in a very "unstable" system which contains packages that cannot be installed due to missing libraries, dependencies that cannot be fulfilled etc. Use it at your own risk

from the bang-bang dept.

[HalliS]

This is what the 2.4. kernel had to say about this:

Bang bang it shot me down
Bang bang I hit the ground
Bang bang that awful sound
Bang bang my brother shot me down

I was 2.4 and he was 2.6
We rode on horses made of sticks
He wore black and I wore white
He could always win the fight
...

Yeah, I know it's pretty crappy, but it's past my bedtime and I'm tired ^_^

Switching to the new kernel I lost 50% performance

[incuso]

I am using a P4 hyperthread processor. Back in the 2.4.x days /proc/cpuinfo showed me two processors. Now it shows me only one! Where is the lost one? Linus, give it me back! Maybe this is the reason his mozilla is faster now? Anyway, the PCs seems as fast as before, that means that the 2.6.x kernel is able to gain a 100% performance with only one processor ? :) M.

Re:what desktop users want to know

[grolschie]

Will Quake/2/3, RTCW/ET, UT, etc, get more frames per second?

prelink

[Nicolas MONNET]

Prelinking will probably help much more for what you mention.

Porting 2.4 code to 2.6?

[daybyter]

Anyone here, who has already ported 2.4 modules to 2.6? I'd like to get my notebook softmodem going, and have some binary, that has to interface with the new 2.6 PCI methods. And to be honest, I'm somewhat stuck with it at the moment... :-(

2.6 on server!

[greppling]

The fact is, 2.6 is considered extremely stable on servers by those who use it. (It's a little more problematic on the desktop where a big variety of hardware needs to be supported, and one of the drivers you need might not be so well-tested yet.)

In fact, some of the more server-oriented developers were so content with the stability of 2.5 early on that they started making mild pushes towards a 2.6.0 release almost a year ago.

Why not -aa tree?

[greppling]

Most people in the know seem to assume that Andrea Arcangeli's tree is a clear winner over 2.4 vanilla in these kind of performance tests. Would have been much more interesting to see a comparison of 2.4.23-aa with 2.6.

(Any SuSE or United Linux 2.4 kernel is based on 2.4-aa.)

Re:Switching to the new kernel I lost 50% performa

[Majix]

If you're not seeing two logical processors in cpuinfo with 2.6, hyperthreading is not enabled for your processor. Most likely this is because you forgot to enable some kernel option, for example, you must enable ACPI device enumeration for HT to work. The kernel must also be SMP capable, even though you have an uniprocessor machine.

Re:interesting hardware comp

[RoLi]

The comparisons won't make much real-world sense until the evaluation is done using Intel's compiler for the Itanium tests. The GNU compiler is just not up to snuff at optimizing for Itanium's EPIC instruction set.

Actually the tests will make a lot of sense because in real life all programs in some Linux distribution are compiled with gcc. But of course nobody stops Intel from putting their tricks into gcc, which would help them move some Itaniums, too.

Please elaborate .......

[Taco Cowboy]

You sez:

"FreeBSD can do some really nifty stuff that Linux can't. Like jails."

Other than "jails", what other "nifties" FreeBSD has ?

I am really interested in knowing, so please elaborate.

Thanks !

Re:Please elaborate .......

[bluGill]

The BSD RC system, not the SysV one (I know that some linux distributions use a BSD style RC).\

CAM on freeBSD is a lot better than Scsi for linux (As of 2.4 kernels, I've not looked at 2.6). As a user you might not notice, but as a programer who deals directly with scsi devices I care.

Seriously though, linux as a lot of cool things. BSD has a lot of cool things. Some get ported back and forth, some do not. (Linus has commited code to freeBSD in the past) I like the BSD way of doing things because it feels right. Others disagree, and that is their right.

Re:Please elaborate .......

[anthonyrcalgary]

Documentation is an example that comes to mind. I don't mean "man ls". I mean, the base system plus the handbook have enough documentation between them to do just about anything needed. I need to hit Google for anything hard on Linux, and Gentoo has better documentation and a more intuitive setup than most.

I often ssh to my OpenBSD machine to find out how to do something on Linux because otherwise I'll be sifting through newsgroups for an hour.

Dual Xeon is slower. Can't disable Hyper Threading

[jasonhamilton]

I've got a dual xeon that ran on 2.4 with HT disabled using "noht" append option in lilo.conf.

I upgraded to 2.6 in the hopes that it would be able to use HT better and thus improve performance.

The problem is, for MySQL applications, it still ran slower for me with HT enabled. And... "noht" option doesn't disable HT anymore.

Anyone know how to disable HT in 2.6?

TIA

Re:This will save me some money

[kill-1]

How could this be moderated as Troll? It's funny. Also notice the low uid.

VMWare

[Peter Cooper]

Would 2.6 result in a performance increase if you were using something like VMWare to run Linux? For example, I've found Redhat 9 to be UBER slow in VMWare, but SuSE 8 is quite fast. And Redhat 9 has certain 2.6 features backported... which is why I worry.

Doesn't it work from the commandline?

[roystgnr]

I don't see any X11 libraries (or any widget libraries) in that list. Gnomevfs is just a "virtual filesystem" library, not part of their GUI code. I'm pretty surprised to see ORBit in there, but even that isn't X-specific, IIRC.

glibc, kernel headers

[drinkypoo]

I have an almost new (1.4) gentoo install with which I am using a 2.6.1 kernel. However I have some 2.4 kernel headers, which my glibc was built with. I noticed that there are now 2.6.0 kernel headers in portage.

Does anyone have any idea if it's worth it to rebuild glibc with the new headers? If I do this, will I have to rebuild my toolchain? (I would guess yes since the bootstrap builds the toolchain, then glibc, then the toolchain again.) And will I have to rebuild world as well?

Only the sheduler?

[SlashDotAgent]

Everybody are talking about the increased performance in 2.6 due to the new O(1) sheduler, but what other differences are there from 2.4 which also make it faster, if any?

jumpy mouse pointer

[mrm677]

Not to be a troll, but I sure hope 2.6 makes X more responsive.

Whenever my compile reaches a large link point, interactive response is terrible...my mouse pointer jumps all over the place.

I know that I could improve things by tweaking the kernel and such, but this box is managed by our IT department (it is RH9) and I don't even have root.

By comparison, if I compile the same app in Windows XP, the system is smooth as butter. Nothing jumps, no MP3s skip, etc.

I don't understand InfoWorld

[Mr. Piddle]

Right below the article is a link to a Microsoft white paper about "third party" research regarding Linux. Note that the white paper link is not put under "sponsored links". Apparently, InfoWorld has been bought by Microsoft's "Get the Facts" double-speak campaign. Also, the author of the Linux article is a "contributing editor", so he is probably outside the inner core of InfoWorld Microsoft payees, unless his article is taken as a direct attack against Microsoft's competitors, Sun and IBM ("Linux meets Big Iron").

Unbiased journalism -- oxymoron of the 21st century.

Depends on what they compiled...

[ShadowRage]

if they added pre-emptive suport to 2.6.. then with server apps, of course 2.6 would be slower (I forget the reasoning why, someone was telling me about why that happens)

pre-emptive support benefits desktop stuff, not server.

besides, 2.6 is still a baby kernel compared to 2.4.. which has been around for a bit and has tried and true features.. the 2.6 is messing with new things..
compile both kernels to almost exactness, then do a speed test.

Plus....

[mcraw]

... Rhythmbox inexplicably stops skipping like a stone on the 2.6 kernel. Hooray! ... I think...

Not ready

[Grayswan]

2.6.1 always freezes on me. Within ~10 minutest of boot. I'm not talking oops or panic. The only thing that gets a response is the reset button. Always during heavy disk access. 2.6.0-pretests did not have this problem and Win2k is rock solid and memtest shows no errors so its not hardware. Athlon 2500 nforce2 asus board.

YMMV.

Re:Not ready

[Garen]

I have that same exact setup and see that same problem. It mysteriously goes away if I disable all the audio stuff. Fucked for sure.

Re:Not ready

[Amadeus0]

I found that disabling ACPI in the kernel stopped this from happening. I also disabled APIC, and the pre-empt kernel options. I'm going to try to re-enable pre-empt, and APIC to see if those features can be used. Shuttle nforce2 MB with IGP.

Re:This will save me some money

[Sabalon]

thanks for the support.

Guess some people just don't get it. It would be painful...last time I ran Linux on a 386 was around 96 and kernel compilation was about a day I think!

Don't need to

[Graelin]

RedHat's ES/AS kernels include the scheduler and VM enhancments from 2.6.

Oranges or Apples? Please specify.

[axxackall]

How does 2.6 compare to Free or Open BSD & how do they compare to windows 2003 server doing the same job?

Doing what job? You didn't specify it.

I use both Linux and various Windows on a daily basis. I can tell then one or another is faster in different tasks. Linux can share files way faster. But J2EE runs on Windows much faster. Running multiple small tasks is better scheduled in Linux. But openning some multimedia files in players is better on Windows.

I don't work with Free or Open BSD on a daily basis. But developers who write the code of PostgreSQL told in mail-lists that their database server runs faster on FreeBSD than on Linux. Well, perhaps because they prefer to spend more time to optimize their code for BSD as they religiously against Linux.

My point is: don't try to compare oranges againsta apples unless you really specify the comparisson.