SMP. Worth it for gaming?

[AlphaWolf]

toms hardware is a bs site, and thats a fact.

Wrong, that is an oppinion.

VR-ZONE may be less known to YOU, but their results are far more accurate than toms Tom's hardware is BS.

Apparently not.

where was that quote taken from, because it isnt on the games benchmark page...

Its in there, if you don't want to read the article, then try using your browsers search function (ctrl-f if you haven't figured it out yet.)

If the minimum frame-rate was higher, then the average frame rate would also be higher wouldnt it.

In games, usually that is the case, but it doesn't have to be. I'll give you time to think about why that is for a minute. It's a simple math concept, but if you give up, let me know and I'll help you.

A little SMP/HT lesson.

SMP stands for Symetrical Multi Processing.

Ding ding ding! Thats the first right answer you've had all day. However, the rest of what you said was wrong, as you do not have to have multiple physical processors for SMP in the case of hyperthreading. Just think of it this way: when you have hyperthreading enabled, both the OS and the software think you have two CPUs installed and they treat it as such, when in fact you only have one. (why else do you think you can't use hyperthreading with windows 9X or windows XP home edition?)

 

[james.miller]

you are wrong on so many counts.

1) check out what i said about toms, it's all correct. do some research and prove me wrong, otherwise YOU are wrong. While your at it, ask the various hardware forums what they think about toms.

2) that quote is not anywhere in that article that is relevant. I.E, GAMING. therefore - doesnt matter.

3) doesnt have to be? whatever. Let's see, what could the answer be? AH, I KNOW. The average is lower, because the maximum frame-rate is lower.......so that makes it better does it?

In YOUR case mabey, but not for anybody else it would seem.

4) SMP IS NOT HYPERTHREADING. Do some research as do as both a favour. Certain operating systems Treat HT cpu's as SMP, that much is true. BUT, that is because they do not recongise HT cpu's correctly.

5)DING DING DING, as you say. you are wrong. WinXP home CAN use hyper-threading, but not SMP. Why? because they are different.

http://www.tech-report.com/reviews/2002q4/pentium4-3.06/index.x?pg=2

Next, you'll need a Hyper-Threading-aware operating system. To date, only Windows XP and Linux (kernel versions 2.4.18 and higher) are HT-aware. Some multithreaded operating systems, like Windows 2000, will run fine with multiple logical processors, but they don't offer the performance benefits of an HT-aware OS. Microsoft has produced an interesting little white paper on HT support in WinXP, which explains the kernel tweaks needed for best performance. For instance, WinXP more aggressively executes the HLT command on an unused logical CPU. (The HLT command exists to tell processors to take a break for a while.) Doing so frees up shared resources for the other logical processor. In case you were wondering, while the Home edition of WinXP supports only one physical processor, it will support HT on that processor.

some more proof that you are wrong. Now show me yours

Alphawolf it has become blatently obvious that you opinions are based on nothing more than rumours, and what youve heard from other people. I asked at the beginning of this thread for PROOF to back up people opinions. You have provided none, and don't seem to want to. Either stop posting in this thread, or start backing up your claims.

 

[AlphaWolf]

you are wrong on so many counts.

1) check out what i said about toms, it's all correct. do some research and proive me wrong, otherwise YOU are wrong. While your at it, ask the various hardware forums what they think aobut toms.

Some of it may very well be correct; I doubt there is a hardware reviewing site in existence that hasn't made similar mistakes. I myself am not the biggest fan of thomshardware either, however, it is the premier review site, and other hardware forums will probably hate it simply because it is a competitor that is more well known than they are. That aside though, I really don't care to argue about which review site is the best because that is off of the topic of this thread, and I will not care to answer anymore arguments to that effect.

2) that quote is not anywhere in that article that is relevant. I.E, GAMING. therefore - doesnt matter.

That quote was referring specifically to quake3, and trust me, its there.

3) doesnt have to be? whatever. Let's see, what could the answer be? AH, I KNOW. The average is lower, because the maximum frame-rate is lower.......so that makes it better does it?

Bingo! Think about that the next time you run into a firefight, and your framerate suddenly drops at a critical point. Suddenly your half extra frame per second wont matter anymore.

Certain operating systems Treat HT cpu's as SMP, that much is true. BUT, that is because they do not recongise HT cpu's correctly.

Wrong: ALL SMP supported operating systems detect a single hyperthreaded CPU as being two logical CPUs.

As for the article that you are quoting, you totally misread it.

It is saying that if you have a hyperthreading aware OS, it's kernel is more optimized for hyperthreading to perform even better than a standard SMP kernel would otherwise perform.

Some multithreaded operating systems, like Windows 2000, will run fine with multiple logical processors, but they don't offer the performance benefits of an HT-aware OS.

This means that windows 2000 can get a performance increase out of hyperthreading, but, it would get an EVEN BETTER performance increase if it was HT aware. (theres a reason for this, I'll explain it if you want)

As for windows XP home not supporting hyperthreading: microsoft may or may not have added it since I got my P4, however, with the p4 package I have, they sent along a free copy of windows XP professional, along with instructions to replace it with any other copy of windows, as they said that windows XP home wouldn't allow you to use hyperthreading, whereas pro at least has SMP support.

EDIT: As for this

Alphawolf it has become blatently obvious that you opinions are based on nothing more than rumours, and what youve heard from other people. I asked at the beginning of this thread for PROOF to back up people opinions. You have provided none, and don't seem to want to. Either stop posting in this thread, or start backing up your claims.

You mean like the article I quoted that you refuse to acknowledge?

 

[james.miller]

Some of it may very well be correct; I doubt there is a hardware reviewing site in existence that hasn't made similar mistakes. I myself am not the biggest fan of thomshardware either, however, it is the premier review site, and other hardware forums will probably hate it simply because it is a competitor that is more well known than they are. That aside though, I really don't care to argue about which review site is the best because that is off of the topic of this thread, and I will not care to answer anymore arguments to that effect.

You dont like it when you are wrong, do you. Again, you argue with nothing more than your personal opinion. I'm talking about skewing results, im talking about lieing. Im tlaking about flat out denying What they have said/written and editing their results to fit whatever needs they have at the time. Think back to where [H]ard Altered their results and got caught out - Toms have done the same thing. Tom's skew all their results towards intel. Want to know why? here's a hint - Their offices are VERY clocse to one another.

That quote was referring specifically to quake3, and trust me, its there.

oh right.

Quake III and RTCW can be SMP enabled but it is totally horrible when it is turn on even on dual processor system therefore it might as well be considered as a single threaded application. Anyway, P4 3.06 with HT only register a very slight decrease in performance therefore you can always run your RTCW or Q3 with HT turn on. Pentium 4 processors does maintain a good lead against AMD Athlon XP 2800+ here.

That says nothing of the sort. Infact, it says the complete opposite.

Bingo! Think about that the next time you run into a firefight, and your framerate suddenly drops at a critical point. Suddenly your half extra frame per second wont matter anymore.

think about what exactly? While that is obvious the most desireable results, it is not the results that everybody BUT you seems to be experiancing. How come you are different to everybody else?

Wrong: ALL SMP supported operating systems detect a single hyperthreaded CPU as being two logical CPUs.

rolleyes

posted by me
4) SMP IS NOT HYPERTHREADING. Do some research as do as both a favour. Certain operating systems Treat HT cpu's as SMP, that much is true. BUT, that is because they do not recongise HT cpu's correctly.

Yes, all SMP capable operating systems recongnise HT cpu's as being two cpu's *of some type*. However, only some operating system actually recognise HT cpu's as BEING HT cpus, and not SMP.

As for the article that you are quoting, you totally misread it.

It is saying that if you have a hyperthreading aware OS, it's kernel is more optimized for hyperthreading to perform even better than a standard SMP kernel would otherwise perform.

This means that windows 2000 can get a performance increase out of hyperthreading, but, it would get an EVEN BETTER performance increase if it was HT aware.

when did i say otherwise?

As for windows XP home not supporting hyperthreading: microsoft may or may not have added it since I got my P4, however, with the p4 package I have, they sent along a free copy of windows XP professional, along with instructions to replace it with any other copy of windows, as they said that windows XP home wouldn't allow you to use hyperthreading.

well, it does. I've seen convirmation from both microsoft AND intel on this matter. let's keep this on topic, eh? :satisfied

 

[AlphaWolf]

That says nothing of the sort. Infact, it says the complete opposite.

/me sighs

Try looking harder.

think about what exactly? While that is obvious the most desireable results, it is not the results that everybody BUT you seems to be experiancing. How come you are different to everybody else?

Show me something that contradicts my claim here.

Yes - THAT IS WHAT I JUST SAID:

No, what you said is that only certain hyperthreading enabled operating systems detect it, not SMP operating systems. In fact, you were specifying one logical CPU earlier. There is nothing wrong with the OS detecting HT as dual CPUs; the only thing is that HT aware OSes can add optimizations to HT.

Yes, all SMP capable operating systems recongnise HT cpu's as being two cpu's *of some type*. However, only some perating system actually recognise HT cpu's as BEING HT cpus, and not SMP.

Now you are trying to eat your previous words as an attempt to disacknowledge what you were saying earlier.

Lets just clear this up for your sake: If hyperthreading is not SMP, then dual CPUs is not SMP. SMP only describes how the software sends its instruction to the CPU(s). SMP is not the only way that you can thread instructions across multiple CPUs mind you.

 

[Eagle]

Carefull guys, your approaching the line here, the discussion is fine but lets knock out the personal flames plz.

 

[james.miller]

/me sighs

Apparently you are either blind or just plain stupid.

show me then

Show me something that contradicts my claim here.

Sure, try the first page of this thread......

Having 2 SMP machines and having played numerous games , I'll toss in my opinions

Quake III - don't bother, SMP almost but not quite slowed down my machine. If I could get it to run in SMP without crashing

Alice - see Quake III (same game engine)


Return to Castle Wolfenstein - see Quake III (same game engine)

Clear enough?

No, what you said is that only certain hyperthreading enabled operating systems detect it, not SMP operating systems. In fact, you were specifying one logical CPU earlier. There is nothing wrong with the OS detecting HT as dual CPUs; the only thing is that HT aware OSes can add optimizations to HT.

The first thing i said about it:

4) SMP IS NOT HYPERTHREADING. Do some research as do as both a favour. Certain operating systems Treat HT cpu's as SMP, that much is true. BUT, that is because they do not recongise HT cpu's correctly.

"Certain operating systems" being all SMP systems that dont otherwise treat them as HT processors. It's not hard to figure out what i ment

Now you are trying to eat your previous words as an attempt to disacknowledge what you were saying earlier.

No im not lol. Everything i have said has been backed up by evidence. You have STILL done nothing. Ive asked you for proof - nothing. Ive asked you for benchmarks - nothing.

you quoted me as saying "yes - thats what i said earlier". Well im sorry - that was a mistake on my part. that wasnt what i said at all.

Lets just clear this up for your sake: If hyperthreading is not SMP, then dual CPUs is not SMP. SMP only describes how the software sends its instruction to the CPU(s). SMP is not the only way that you can thread instructions across multiple CPUs mind you.

Sounds like you are ust making assuptions again.

SMP = Symetrical MULTI Processing. It does exactly hat it says on the tin - symetrical processing using more than one (ie - multi) processor.

from google:

------------------------------------
What is smp?

SMP - An abbreviation for symmetric multiprocessing, implemented in Windows NT, which distributes tasks among CPUs using a load-sharing methodology. Applications must be multithreaded to take advantage of SMP. See asymmetric multiprocessing.

------------------------------------

HT = hyperthreading. From intels website http://www.intel.com/technology/hyperthread/

------------------------------------
Hyper-Threading Technology provides thread-level-parallelism (TLP) on each processor resulting in increased utilization of processor execution resources. As a result, resource utilization yields higher processing throughput. Hyper-Threading Technology is a form of simultaneous multi-threading technology (SMT) where multiple threads of software applications can be run simultaneously on one processor.
------------------------------------

Notice the mention of SMT, but no mention of smp. Again, because they are different.
Aplhawolf dont try and insult me. dont cal me stupid, and dont try and pick holes in everything i say. Again, just show me some proof. Is it really that hard?

 

[AlphaWolf]

show me then

I did show you, its in the webpage. Try copying and pasting what I quoted into your find box.

Sure, try the first page of this thread......

Which does not answer my comment about framerate drops.

"Certain operating systems" being all SMP systems that dont otherwise treat them as HT processors. It's not hard to figure out what i ment

Ok, linux and windows XP. Care to name any other of these "certain" operating systems?

No im not lol. Everything i have said has been backed up by evidence. You have STILL done nothing. Ive asked you for proof - nothing. Ive asked you for benchmarks - nothing.

I've given you proof twice. You just refuse to look at it.

Sounds like you are ust making assuptions again.

SMP = Symetrical MULTI Processing. It does exactly hat it says on the tin - symetrical processing using more than one (ie - multi) processor.

HT = hyperthreading. From intels website http://www.intel.com/technology/hyperthread/

Right, explain to me this much then: Why doesn't hyperthreading give any performance increase without SMP support? (I'll give you a hint, its the same reason that dual CPUs don't give better performance without SMP support.)

 

[james.miller]

you really are unbelieveable. Nothing you have said is backed up. I show you everything you want and you argue with all of it. You pick wholes on my statements because you have nothing relevant to argue. What is the point of you even posting? whatever, i wont be responding to your posts anymore......

.....that is unless you show me some PROOF for the last time.

Does anybody else have some proof, one way or the other?

 

[AlphaWolf]

Nothing you have said is backed up.

I gave you two URLs, and you didn't read either of them. I at least gave you the benefit of the doubt by reading yours. And I didn't poke the holes in your argument; they were already there (namely you don't have a proper understanding of what exactly SMP is.)

Since he isn't reading my posts anymore, heres a little comment I might add for everybody else:

About the more constant framerates in UT2k3, I think its possible that the reason it is this way is because when hyperthreading is enabled, tasks that need to be performed at the same time during normal gameplay are accomplished simultaneously during the time that a framerate drop would otherwise occur. E.g, a linear thread would only do one thing at a time, as each task has to wait on the other one to be performed before it is done, and that waiting is where the frame drops are. The loss in the otherwise constant framerate speeds would be explained by the fact that since these tasks are all being performed at once, they perform slower overall.

Think of it like if you play e.g. half life, and set its thread priority to realtime. During CPU intensive operations, the user controls will freeze up while the game keeps going if you do this, because there isn't any spare processing time being allocated to them since the game is using it all up.

This is only a guess though...the reason could possibly just be something specific to my system configuration, I don't know honestly. If anybody has any articles that prove or disprove my theory, let me know.

 

[zAlbee]

james.miller, you are as biased as ever. i can't believe how fast you turned on AlphaWolf just for suggesting a different opinion.

It seems completely reasonable to me that HT would provide smoother framerates during high-action areas, and yet still register the same average framerate. Sounds similar to Kaos' review of SMP where playing a game by itself did not improve, but doing multiple other tasks (burning CD, scanning for viruses) at the same time as playing a game was much smoother. HT has the same ability - it was shown in the VR-Zone article james mentioned. Most likely, during high-action play the burden of running the game at the same time as other mundane tasks (such as the OS, I/O etc) finally becomes apparent as the non-HT (or non-SMP) processor suffers, whilst the HT/SMP one excels. I think that's pretty much the same idea that Alpha has.

As for the HT == SMP issue, I really think you guys are nitpicking on whether the name "multi processing" can only be applied to multiple physical processors, or also sub/logic/virtual/imaginary processors as well. i mean, who cares

 

[james.miller]

how on earth am i biased? throughout this thread i have asked for proof or evidence either way. how is that concidered biased? Everything i have posted has been from various sites, it is not my opinion.

You want my opinion? Ideally, i WOULD have an SMP system. The benefits it brings for running different tasks at the same time is a big plus for me, but honestly that has nothing to do with gaming. Now how is that biased?

I have no problem with opinion - as long as that opinion is not forced as being correct unless there is evidence to support it. as for coming down on him, if you saw the PM's he has sent me, you'd do the same.

It seems completely reasonable to me that HT would provide smoother framerates during high-action areas, and yet still register the same average framerate. Sounds similar to Kaos' review of SMP where playing a game by itself did not improve, but doing multiple other tasks (burning CD, scanning for viruses) at the same time as playing a game was much smoother. HT has the same ability - it was shown in the VR-Zone article james mentioned. Most likely, during high-action play the burden of running the game at the same time as other mundane tasks (such as the OS, I/O etc) finally becomes apparent as the non-HT (or non-SMP) processor suffers, whilst the HT/SMP one excels. I think that's pretty much the same idea that Alpha has.

yes and i totally agree with that, but the operating system *shouldnt* need to do any high cpu-usage tasks if you keep on top of it - i know mine doesn't.

Now, for the sake of this debate, i have asked the question in various different forums. The results thus far from ALL forums has been the same (except from one person - alphawolf). everybody who has an SMP or HT system has noticed no difference, or due to SMP/HT has actually seen a drop in performance.

one such person:

It doesn't do anything for gaming. In fact, my Dual would be slower (only ever so slightly) than an equivalent single cpu system, due to the memory overhead. I think there is only one or two games out there that can use two cpu's (or HT), and I think one is the quake 3 engine. But I am not sure if it is enabled in the games that use that engine. It would be great if some games used dual cpus, like Champ Manager, or something like BF1942 where you crate your own bots game, and have the other cpu controlling the bots AI.

So that settles it

Sorry for no benchmarks, but I have done benchmarks before, and they are always slightly slower, but technically not noticable. Plus right now I am trying to defrag a drive that has NEVER been defragged (we're talking over a year here) so I need to get that done! LOL! 24 hours to do 4%...

What alphawolf has been saying is theoretically possible and i never said otherwise. The problem, is that he is the only person who seems to see that difference . What does that say? Everybody else is wrong? Or is he wrong?

 

[AlphaWolf]

As for the HT == SMP issue, I really think you guys are nitpicking on whether the name "multi processing" can only be applied to multiple physical processors, or also sub/logic/virtual/imaginary processors as well. i mean, who cares

Well, the thing here, is I am saying that SMP is a method and not a physical setup. E.g, if software doesn't use SMP, then neither hyperthreading nor dual CPUs will bring any benefit, whereas james.miller believes that if you have two CPUs, its automatically called SMP, which simply isn't true.

What aplhawolf has been saying is theoretically possible i i never said othersie. the problem is, he is the only person who seems to see any difference. What does that say? Everybody else is wrong? Or is he wrong?

I showed you an article that also mentioned a difference, yet for some reason or another you keep saying that you can't find the text in there. Not only that, but toms hardware was quoting a games developer, and I don't care how biased you think toms hardware is, the games developer was accurate.

 

[zAlbee]

how on earth am i biased? throughout this thread i have asked for proof or evidence either way. how is that concidered biased? Everything i have posted has been from various sites, it is not my opinion.

well, just in the way you wrote during your long debate with him, you came off as being biased. your points might still be valid, but it was the writing style that got me. i mean, if one day the revelation hits me that you're right, i might think "wow, james miller was right! He's still biased though."

 

[james.miller]

nevermind. This isnt about my opinion, and if it comes across that way then i am sorry no skin off my nose.

I showed you an article that also mentioned a difference, yet for some reason or another you keep saying that you can't find the text in there. Not only that, but toms hardware was quoting a games developer, and I don't care how biased you think toms hardware is, the games developer was accurate.

An article from an unknown person with NO benchmark results on ZDnet. Need i say more?

tomshardware? this is what thier article had to say:

Game developers say they will add more complex threads to future gaming applications if and when HT becomes more prevalent.

article dated dec. 2002

If....and...WHEN.....

ok then, that was their report before developers actually did anything, so where are the benchamrk to show they ARE doing anything? That article says nothing about what Ht or SMP is actually doing for games. It is nothing but pure speculation and hype from over a year ago.

 

[AlphaWolf]

An article from an unknown person with NO benchmark results on ZDnet. Need i say more?

I personaly don't care about numbers, because most of the time they are meaningless, and at best they only give you an index (for example, the UT2k3 framerate discussion we had earlier.) Nonetheless, that article is every bit as relavent as your vr-whatever one.

tomshardware? this is what thier article had to say:

Evidently you didn't read the bit about galciv at all. Oh, and by the way, where was galciv on your SMP games list? If I recall correctly, galciv is the best example of an SMP game there is, yet that list didn't mention it....

 

[james.miller]

I personaly don't care about numbers, because most of the time they are meaningless, and at best they only give you an index (for example, the UT2k3 framerate discussion we had earlier.) Nonetheless, that article is every bit as relavent as your vr-whatever one.

of course it is - he provided actual results and charts, didnt he?

Evidently you didn't read the bit about galciv at all. Oh, and by the way, where was galciv on your SMP games list? If I recall correctly, galciv is the best example of an SMP game there is, yet that list didn't mention it....

evidently you are, again, wrong. galciv may well be the best thing since sliced bread as far as HT is concerned, but since when does one game from one developer making use of HtT make HT a viable solution?

one game. Does that make HT/SMP worth it? would you purchace an SMP system solely on the basis that it will play one game mabey 10% faster than a non SMP/HT system of similar spec would?

I dont think so *in my opinion*. I certainly don't know anybody that would. As for the list, can't you just drop that? AGAIN, it is more than you provided. At the bottom of that list is an email address, and he askes that you provide him with any updates you find. Prehaps you should do that?

On the subject of GalCiv, i'm having trouble finding any benchamrks comparing HT on and off. Could you find some for me?

 

[AlphaWolf]

of course it is - he provided actual results and charts, didnt he?

Like I said, useless.

evidently you are, again, wrong. galciv may well be the best thing since sliced bread as far as HT is concerned

Galciv is older than hyperthreading.

Wait...didn't you say you weren't going to read anymore of my posts or something?

 

[james.miller]

and again you pick whole in my posts becuase you have nothing better to do. please, concerntrate on the matter at hand.

Once more,since you rate GalCiv so highly on the HT/SMP list, can you please help me find some benchmarks? or are you going to make useless attacks on my posts again?

 

[AlphaWolf]

Once more,since you rate GalCiv so highly on the HT/SMP list, can you please help me find some benchmarks? or are you going to make useless attacks on my posts again?

*shrug* use google? I don't know what you are trying to prove by finding a benchmark. The game is just coded well for SMP support...what more is needed?