OS's that support twin CPUs

Tim_K · Post by **Tim_K** » Thu Dec 09, 2004 11:52 pm

From what I've been able to find here, it seems like Windows XP Pro, NT, and 2000 support dual CPU's, but 95, 98, 98SE, and ME do not.

Do all versions of Linux support dual CPUs? How about Knoppix 3.6?

I have access to Knoppix 3.6, 98, XP Home, and XP Pro right now. I wanted to get a copy of 2000, but they keep getting bid over $30.00 on eBay, and some don't even have the COA and Key along with the CD. Yes, I'm cheap.

I'm getting a BP6 and will overclock a pair of 533 Celerons in it and use it for folding@home. I'd like to use an OS that will not bog the system down. I don't have a lot of experience with Linux variants yet. Knoppix 3.6 runs but I don't really want to use it because it won't run Windows Media Player easily. Not that I'd use WMP on it often, but would like to have the ability to.

How about Damn Small Linux, or other Linux variants? Can they be configured to look differently, and how?

purrkur · Post by **purrkur** » Fri Dec 10, 2004 1:14 am

Tim_K wrote:From what I've been able to find here, it seems like Windows XP Pro, NT, and 2000 support dual CPU's, but 95, 98, 98SE, and ME do not.

True.

Tim_K wrote:Do all versions of Linux support dual CPUs? How about Knoppix 3.6?

All versions of Linux support dual CPU's. Knoppix should as well if the kernel that Knoppix uses is set up for SMP. I haven't tested Knoppix on a dual cpu system yet but I am 99% sure that it does.

Tim_K wrote:I'd like to use an OS that will not bog the system down.

I am still trying to figure out what you mean by this one. Yes, a more recent OS will work slower on older hardware but if you are only doing a single thing with this machine then the OS will be out of your way once it is booted and the software you are going to run is loaded as well.

Tim_K wrote:How about Damn Small Linux, or other Linux variants? Can they be configured to look differently, and how?

If you have no experience whatsoever in Linux and you want results fast then I would suggest using other distro's that are not based on KDE. I suggest you try Ubuntu Linux which is supposedly the one of the hottest thing around atm. www.ubuntu.com. It is free and it is a desktop-centric distro based on Gnome (which is the other big point-n-click system in Linux). See if you like it better.

Usually distros such as Damn Small Linux etc are harder to use because user friendlyness does not come in a tight package. If you want to test a Live-CD Linux distro (like Knoppix) based on Gnome then try something like Gnoppix (http://www.gnoppix.org/).

Dave Rave · Post by **Dave Rave** » Fri Dec 10, 2004 6:05 am

I think he means overhead...

a 500mhz in seti takes x hours
a 1ghz doesn't take x/2 hours
a 2ghz doesn't take x/4 hours

it's more like
a 500 takes x-20 hours
the 1ghz takes x/2-20

no wait, arghh...............
uhm,,,...

a 500 in win98 takes x hours
in win2k it takes a slightly different time
in linux it's much slower, but possibly because it's a completely different .exe file

there, is that clear as mud can be ?

purrkur · Post by **purrkur** » Fri Dec 10, 2004 7:18 am

Dave Rave wrote:there, is that clear as mud can be ?

Yessir!

Tim_K · Post by **Tim_K** » Fri Dec 10, 2004 8:53 am

I forgot to mention XP Home. From what I've read XP Home won't support dual CPU's either.

In Knoppix 3.6 the Linux kernel is 2.4.27.

By not bogging the system down, what I mean is that I've heard that XP uses a lot of system resources. Since I only plan to run 128 or 256 MB of memory to keep costs down, I don't want the computer to be a slow processor of folding work units because the OS takes up more of the clock cycles than it absolutely needs to.

Like I said before, I want to get a copy of Windows 2000, but every one I see on eBay gets bid up over $30 by the auction end, which is the most I'm willing to pay right now. Anyone got a good one with the codes and certificates and keys?

hugoc · Post by **hugoc** » Fri Dec 10, 2004 12:51 pm

I've run Knoppix on an SMP system and it works fine. Slax works too, both are great live distributions (you can try before you partition).

As to reducing overhead, I've found that XP gets bogged down on a PII system. It really needs a PIII or better. You can turn off all the eye-candy and it is better, however, then you're basically running Windows 2000, so what is the point? Sure, XP gives you a "firewall", if you could call it that...

If you want low overhead from a Windows OS, just install NT4. Your typical BP6 hardware will be blisteringly fast for a 10-year-old OS. Out of the box you won't get FAT32 or USB support, your best DirectX version is 3, you can't run any MediaPlayer version over 6.x, nor will you get any PnP. But those are the breaks.

Tim_K · Post by **Tim_K** » Sat Dec 25, 2004 2:05 am

I found a copy of Windows 2000 Pro. I put it on my Dell Latitude 366 Mhz P2 laptop in place of XP Home, and the laptop runs much better and smoother now.

I've got a hard disk all prepared with 2000 Pro for when my BP6 gets here.

Billl · Post by **Billl** » Sat Dec 25, 2004 1:33 pm

Tim_K wrote:I'd like to use an OS that will not bog the system down.

purrkur wrote:I am still trying to figure out what you mean by this one. Yes, a more recent OS will work slower on older hardware but if you are only doing a single thing with this machine then the OS will be out of your way once it is booted and the software you are going to run is loaded as well.

I have to respectively disagree with you on this one Purrkur. XP does indeed slow you down. I'll give you an example. I had a celeron 566 OC'ed to 860. I installed XP pro and SETI. Work unit times were 14 - 16 Hrs. This was after I had eliminated all unnecessary processes. I later reformatted the drive put 98SE on and work unit times were 6-8 Hrs. My guess is any fairly recent OS would do the same since we all know how programers like to bloat code. Personally I think he would be best off with an older linux or windows distribution. 2K sounds just fine for what he wants.

Billl

davd_bob · Post by **davd_bob** » Thu Dec 30, 2004 3:44 am

Im with Billl al the way.
UNLESS a software takes advantage of newer technology built into equipment, AND that equipment is available, it is better to use older software and OS if it does the same job. I guess its best to match the software to the equipment.

For example, WIN95 will run on a 386DX but it is reaaaaaaally slow whereas DOS 5.0 is ok. But DOS 5 on a P-ii 400 would fly...but NT4 or WIN95 would be a more versital(sp) choice.

purrkur · Post by **purrkur** » Thu Dec 30, 2004 5:04 am

I am still not convinced.
<rant mode>
The only things that I have seen slow down with a modern OS compared to something like Windows 98 is boot times and responsiveness because of general bloat (larger programs take longer to load and get running and larger code means you are manipulating more data which is time consuming) which is the result of machines getting faster all the time and developers writing sloppy code.

Billl, what you are saying is that Windows 98SE is more than twice as fast at running a single process which is mainly CPU specific than XP? I just can't believe it. There must have been something else that you are not aware of that was affecting your runs, or if it is possible that the code was maybe specific to the old Windows OS's and that it wasn't really built to run on XP?

I have been working with various OS's on varying computers throughout my computer-related life and I have never experienced a machine being only half as fast by only changing the OS.

And Billl, if you are right then this article is pretty much untrue. It is a comparison between Windows ME and XP as a gaming platform on a "modern" computer. It shows you that they are pretty much as fast and that there is no difference between the two (cpu intensive task). If you are right then this article is wrong because you stated that XP by definition is twice as slow as Win98SE.

And Davd_bob, I don't see your point. You compare running DOS and Windows95 on a 386? Think about it. We both know that Win95 loads a lot more code than DOS does and therefore it will be slower because your 386 won't have the memory to load the OS fully so it will start swapping and become painfully slow. Even if you had the memory to load the OS then it would still be slower because you are loading (and working with) a lot more code. If you have a car that can reach 120mph and you go flat out for a mile then it will take you a certain time. If you then to twice the distance then it will take you longer of course. Well, your computer is your car and the code bloat is your distance. It is very simple.

Again, I am not saying that you won't feel a difference because I know all about moving computers from NT to XP. But XP is doing a lot more, loading more stuff and the code itself is probably close to being twice as large. Just do a clean install of NT and check the memory footprint and do the same with XP! I know that on a modern machine XP will take close to 100 megs of memory without anything else loaded than the OS itself. I can't specifically remember the numbers for NT but I think they are closer to 50-60 megs. I have also seen computer habits change as a result of XP. Folks load more services and programs and the computer is forced to do more at once which will definitely make XP slower.

But if a computer is told to run a specific task under two different OS's and it does nothing else and it has the resources to do so (under both OS's) then my opinion is that the speed difference won't be so large as you say. This goes especially for processes that are CPU intensive. The small difference will be explainable by the fact that a newer OS is keeping tabs on more services and background programs that is expected of a modern OS.

I have seen this time and time again under Linux. As soon as a new stable kernel is released, it is always compared speedwise to its older syblings. For example, the 2.6. kernel was extensively compared to both version 2.2 and 2.4. The differences found (and they were never extreme) were always explainable. I can also say that the 2.6 kernel is a lot faster than for example the 2.2 kernel in many things and it is way more efficient.

I want to finish off this rant with a mention of overclocking. When I bought a super7 motherboard and a 350MHz K6-2 I went straight into overclocking it. It was a very good overclocker and I was able to run it successfully at 450MHz without stability problems. However, despite the faster speed the machine felt a lot slower for some reason. I could never put my finger on it but it felt like the machine was missing a beat every once in a while. I never found out what it was but when I clocked it down to 400MHz then this performance "problem" went away. I ended up running it at 400MHz instead because 450MHz, although stable, never felt right. After this incident which was many years ago, if I am working on an overclocked computer (and almost all of mine are) and I run into an issue I can't explain then I always try things again running at standard speeds to see if the problem occurs again.

So if you want to say that XP causes a computer running a CPU intensive task at half the speed as the same machine running Win98SE then I will have to see more proof of exactly what you are saying. I have been working on computers and different OS's since 1983 and I have never experienced nothing like it, nor have I seen any tests elsewhere that have confirmed such a fact.
</rant mode>

davd_bob · Post by **davd_bob** » Thu Dec 30, 2004 5:53 am

Jeez,
Ok, I was not clear in my last post. I just ment I think hardware and software designed for each other is the best combo for each. DOS does not compare to any graphical interface OS. I agree that a CPU should do comparable output if on WIN98SE or XP other then the startup and such(just like Purrkur said) HOWEVER I feel that you are better running XP on a GHz machine with DDR.

I ran WIN95 on my super7 with a K6-III/400. It was sweet. The K6-III wouldn't OC at all, but I ran 100FSB and that system matched pretty well with a P2/350. When I hit shutdown the system would be powering off before I got my hand off the mouse. Games and SiS Sandra2001 gave about the same results whether on WIN95 or 98 or W2K. I never tried linux.
Now I must seem fickle.

Billl · Post by **Billl** » Thu Dec 30, 2004 4:32 pm

purrkur wrote:I am still not convinced.
<rant mode>
The only things that I have seen slow down with a modern OS compared to something like Windows 98 is boot times and responsiveness because of general bloat (larger programs take longer to load and get running and larger code means you are manipulating more data which is time consuming) which is the result of machines getting faster all the time and developers writing sloppy code.

Billl, what you are saying is that Windows 98SE is more than twice as fast at running a single process which is mainly CPU specific than XP? I just can't believe it. There must have been something else that you are not aware of that was affecting your runs, or if it is possible that the code was maybe specific to the old Windows OS's and that it wasn't really built to run on XP?

Well this is exactly what I am saying. The code was identical between loads. Seti command line is the same no matter what Windows OS it is running on. And I'm betting the developers would be willing to argue with you about how well their code was written

purrkur wrote:I have been working with various OS's on varying computers throughout my computer-related life and I have never experienced a machine being only half as fast by only changing the OS.

And Billl, if you are right then this article is pretty much untrue. It is a comparison between Windows ME and XP as a gaming platform on a "modern" computer. It shows you that they are pretty much as fast and that there is no difference between the two (cpu intensive task). If you are right then this article is wrong because you stated that XP by definition is twice as slow as Win98SE.

I would never go so far as to say I can't be wrong, but I do know what I have seen. This wasn't a quick test and unload either. I ran this for quite some time in XP and then even more so in 98SE. And remember this was after I had shut off all the extra processes that XP loads by default. One thing to keep in mind running Seti is vastly different then running a game. For instance who would think that a Celeron would be faster at Seti then a PIII? I never would have thought so but it is. Case in point. Celeron 533 running at 824 on a BE6 M/B versus a dell PIII 733. Now note the 733 is usiing a 133 MHZ bus speed the Celeron is running at 103 MHZ. less then 100 MHZ seperates them yet the 733 does 9 - 10 hr work units and the 533 does 7 - 8 hr work units. The Celeron has more Memory 384 versus 256 on the PIII. Both running 98SE. With that much memory in each machine ram isn't part of the equation. Windows XP would be effected but not 98. Bus speed to the memory seems to be more important then actual CPU cashe size. These results are after running many hundreds of work units. So explain that one then?

purrkur wrote:And Davd_bob, I don't see your point. You compare running DOS and Windows95 on a 386? Think about it. We both know that Win95 loads a lot more code than DOS does and therefore it will be slower because your 386 won't have the memory to load the OS fully so it will start swapping and become painfully slow. Even if you had the memory to load the OS then it would still be slower because you are loading (and working with) a lot more code. If you have a car that can reach 120mph and you go flat out for a mile then it will take you a certain time. If you then to twice the distance then it will take you longer of course. Well, your computer is your car and the code bloat is your distance. It is very simple.

Again, I am not saying that you won't feel a difference because I know all about moving computers from NT to XP. But XP is doing a lot more, loading more stuff and the code itself is probably close to being twice as large. Just do a clean install of NT and check the memory footprint and do the same with XP! I know that on a modern machine XP will take close to 100 megs of memory without anything else loaded than the OS itself. I can't specifically remember the numbers for NT but I think they are closer to 50-60 megs. I have also seen computer habits change as a result of XP. Folks load more services and programs and the computer is forced to do more at once which will definitely make XP slower.

I do agree with you on this point.

purrkur wrote:But if a computer is told to run a specific task under two different OS's and it does nothing else and it has the resources to do so (under both OS's) then my opinion is that the speed difference won't be so large as you say. This goes especially for processes that are CPU intensive. The small difference will be explainable by the fact that a newer OS is keeping tabs on more services and background programs that is expected of a modern OS.

Well I have to say from my experience your wrong. At least when it comes to Seti. As anyone knows looking at the stat page I've crunched a work unit or two

purrkur wrote:I have seen this time and time again under Linux. As soon as a new stable kernel is released, it is always compared speedwise to its older syblings. For example, the 2.6. kernel was extensively compared to both version 2.2 and 2.4. The differences found (and they were never extreme) were always explainable. I can also say that the 2.6 kernel is a lot faster than for example the 2.2 kernel in many things and it is way more efficient.

Well it's well known (at least among women) how we love to see who is bigger lol

purrkur wrote:I want to finish off this rant with a mention of overclocking. When I bought a super7 motherboard and a 350MHz K6-2 I went straight into overclocking it. It was a very good overclocker and I was able to run it successfully at 450MHz without stability problems. However, despite the faster speed the machine felt a lot slower for some reason. I could never put my finger on it but it felt like the machine was missing a beat every once in a while. I never found out what it was but when I clocked it down to 400MHz then this performance "problem" went away. I ended up running it at 400MHz instead because 450MHz, although stable, never felt right. After this incident which was many years ago, if I am working on an overclocked computer (and almost all of mine are) and I run into an issue I can't explain then I always try things again running at standard speeds to see if the problem occurs again.

I've never experienced that at all. Any machine I've overclocked has always felt much more responsive. I've overclocked quite a number of them.

purrkur wrote:So if you want to say that XP causes a computer running a CPU intensive task at half the speed as the same machine running Win98SE then I will have to see more proof of exactly what you are saying. I have been working on computers and different OS's since 1983 and I have never experienced nothing like it, nor have I seen any tests elsewhere that have confirmed such a fact.
</rant mode>

I've been working with computers myself since 1983. So I have an idea or two about the things I've told you here. I don't know what evidence I could offer you to convince you of what I'm saying?

Billl

purrkur · Post by **purrkur** » Thu Dec 30, 2004 6:12 pm

Billl wrote:Well this is exactly what I am saying. The code was identical between loads. Seti command line is the same no matter what Windows OS it is running on. And I'm betting the developers would be willing to argue with you about how well their code was written

If it works so badly on one OS then I would like to say that it is badly written code

Billl wrote:One thing to keep in mind running Seti is vastly different then running a game. For instance who would think that a Celeron would be faster at Seti then a PIII? I never would have thought so but it is. Case in point. Celeron 533 running at 824 on a BE6 M/B versus a dell PIII 733. Now note the 733 is usiing a 133 MHZ bus speed the Celeron is running at 103 MHZ. less then 100 MHZ seperates them yet the 733 does 9 - 10 hr work units and the 533 does 7 - 8 hr work units. The Celeron has more Memory 384 versus 256 on the PIII. Both running 98SE. With that much memory in each machine ram isn't part of the equation. Windows XP would be effected but not 98. Bus speed to the memory seems to be more important then actual CPU cashe size. These results are after running many hundreds of work units. So explain that one then?

OK. I know that certain jobs rely on raw CPU power and thats it. What you just said above would have me think that seti doesn't use SSE, MMX or anything else that would be specific to a certain CPU architecture. What seti needs is pure MHz and your overclocked Celeron has more of them. In Linux we have something of a useless cpu "benchmark" called Bogomips. We describe that as a test of how good the processor is at doing nothing. In any case, that test will give you figures based purely on cpu speed and nothing else. So a 500MHz K6-2 will show similar bogomips figures to a XEON P3 running at the same speed. Case in point is my dual Celeron running at 576MHz has higher bogomips figures than my dual XEON 550MHz machine at work.

I'll gladly admit though, I am totally useless at seti. I have no idea of how this thing is built. Maybe I should check into it for fun.

Billl wrote:I've been working with computers myself since 1983. So I have an idea or two about the things I've told you here. I don't know what evidence I could offer you to convince you of what I'm saying?

Hmmm. I dunno. Somebody else must have experienced this because it is highly remarkable in my opinion. Maybe I'll google around a bit since I know so many hardware nuts are crazy about seti and the crunching they do there. I'll post if I find something of interest....

davd_bob · Post by **davd_bob** » Thu Dec 30, 2004 11:38 pm

Purrkur and Billl,

Sometimes percpetion is everything.

My example is like comparing apples and oranges but my first taste of DUALing was a Pentium PRO 200 and my first attempt at Linux was Mandrake. Both left me feeling dissapointed. Trying again with a BP6 and Red Hat left me much more satisfied.

purrkur · Post by **purrkur** » Sat Jan 01, 2005 3:26 pm

Billl: I did do some research on SETI and how good it is as a benchmark. Apparently it doesn't serve that purpose very well.

I had a hard time finding decent material on the subject (strangely enough) but I did find some comparisons, especially between Linux and Windows and let me tell you that every other comparison had Linux winning while the others had Windows winning. Almost all I saw actually said that NT and 2000 were slightly faster than 98 or ME. But I didn't think that the comparisons being made were good enough to be quoted here so I won't bother.

I had a hard time finding information on the usage of SETI being used as a CPU benchmarking tool but I did find one interesting article here. The article is long and not really about SETI specifically but about benchmarking and how worthless they can be

He wrote this:

Since Celerons acquired built in cache, the trade press has often referred to them as price performance bargains; on the same version of OpenBSD, I found 533Mhz Celerons to run 2.5 to 3 times slower than a PIII 500 running the identical SETI application. Treat all performance comparisons with skepticism. The only performance that matters are your production systems, which are not likely to directly correlate closely to any standard benchmarks.

Because of that extreme difference I became sceptical and so did the author. It seems now that the BOINC system is built entirely different so it won't be a possible tool for comparing systems because powerful machines will get a heavier workload while less powerful computers will get a lower workload. I am sort of surprised that they would do that because I think one of the sweet factors of SETI for geeks is just how much data crunching their computers can manage.

Billl · Post by **Billl** » Mon Jan 03, 2005 3:16 am

Well all I know is what I've personally experienced over a number of years working with Seti. As to it being much a benchmark, probably not. Unless you go download the standard work unit used for benchmarking, and run that on all the machines. But over time you see a lot of work unit times and you can begin toi draw conclusions. As to XP being faster then 98, I suppose in some situations it may well be. But in my experience with Seti it hasn't been.

Billl

BP6.Com

OS's that support twin CPUs

OS's that support twin CPUs

Re: OS's that support twin CPUs

Re: OS's that support twin CPUs