Sign in with
Sign up | Sign in
Your question

3.06 P4 with HT benchmarks

Tags:
  • CPUs
  • Performance
Last response: in CPUs
Share
November 11, 2002 7:23:41 AM

The french site x86-secrets.com has released the <A HREF="http://babelfish.altavista.com/babelfish/urltrurl?lp=fr..." target="_new">following benchmarks</A>. Apparantly the way Intel solved the performance loss problem was in software and it really wasn't Intel that solved it. It is up to the OS to handle threads sent to logical processors so that there is no resource conflict. In WinXP, and Linux with Kernel 2.4.1 (I think) and above, logical processor support is enabled. In Win2k and previous versions of Linux, there can be a significant performance loss in single-threaded applications.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

More about : benchmarks

November 11, 2002 12:57:38 PM

Heheh, I can only laugh at those who kept hyping it so much.
While SMP benches benefit, so far I haven't seen anything of home use that would benefit us. And I was surprised when I saw the tests on Win2K and how it affects it!

Anyways, another laugh by me to those rabid zealots who thought HT would be the life threat of AMD!

--
The worst of enemies shall be prone to later be the best of friends. -Eden
November 11, 2002 2:43:47 PM

Eden: what are you talking about? By just watching the game benchs (all the other benchies don't interest me because I only care about real-life performance, although the sandra did also very well) I can only see increases in performances when using HT! Those performance enhancements will proportional increase with higher clocked P4s coming out the next months and maybe their will be even patches to further optimize HT performance on games and apps... just watch x86-secrets overclocking results.

The reason why HT performed so bad under win2000 and q3a with r_smp enabled is quite obvious. Both apps don't support HT and therefor enabling SMP (not SMT which HT is) can only degrade performance.

Now, someone should make a extensive list with all most-used games and applications to see the real benefits or maybe disadvantages of HT P4s. It's too early to judge HT based on a few synthetic benchies posted by one website.

I'm still wondering how good Prescot with HT2 will perform...
Related resources
November 11, 2002 3:16:21 PM

what does it matter that a pIV 3.06 underperforms in win 2000 if you've got the cash for that CPU you've got the cash for win XP (which fully supports HT) aswell

or not
November 11, 2002 5:13:29 PM

I never said it didn't have performance increases, I just said it was not impressive, rather disappointing in how people kept hyping it.
Sure, SMP synthetic benchmarks showed quite some improvement, but that's what the name is, they're SYNTHETIC as well. And I did notice some others have some sizeable boosts, it yet remains however to test more real-world apps. Where are the MPEG compression and encoding benchmarks? What about LAME MP3? AutoCAD or anything that is 3d Rendering?
These are what interest me, as well as the games benchmarks, which so far show that even SMP supported games don't deal with much improvement. I suspect Commanche will, since it's so CPU dependant.

BTW I am not sure on the proportional claim, they did overclock the CPU, and I might've skipped a place where they said it was multiplier unlocked, but if I didn't, then that means the boost came from a higher bus.

I knew about the Win2K reason, I simply outlined that the loss of performance was much more astronomical than I expected. Most people thought it loses not more than 10% on incompatible programs, but geez!

Quote:
I'm still wondering how good Prescot with HT2 will perform...

Why does everyone keep saying HT2? THIS IS HT2! Why should it be any different on Prescott? The boosts on it come from a higher bus and memory, and a higher cache so far. Unless you refer to the 3.06GHZ HT as HT2, then it doesn't matter, I just outlined why it shouldn't really make any difference.

Quote:
It's too early to judge HT based on a few synthetic benchies posted by one website.

Reports are the 3.06GHZ is out, if not in 3 days from today, I don't see how it's too early.


--
The worst of enemies shall be prone to later be the best of friends. -Eden
November 11, 2002 6:08:15 PM

Quote:
These are what interest me, as well as the games benchmarks, which so far show that even SMP supported games don't deal with much improvement. I suspect Commanche will, since it's so CPU dependant.

Well I don't think any game yet is optimized for SMP. Perhaps Quake and UT2003 have a couple threads, but the code is still designed for single processor systems. This may change as the installed userbase of HT P4s increases.

Just like with SSE2, it will be some time before HT improves IPC in everyday use. However, HT is cooler than SSE2 in one way: it offers some IPC benefit right now, today, in existing apps!

Ritesh


<P ID="edit"><FONT SIZE=-1><EM>Edited by ritesh_laud on 11/11/02 02:15 PM.</EM></FONT></P>
November 11, 2002 6:42:14 PM

Yes you're right. However for current use, I think some people should stop giving so much hype over the 3.06GHZ, claiming as if it will rape the competition up, especially later on against Hammers.

Nothing is known to date, AMD could as well develop a multithreading technique.

--
The worst of enemies shall be prone to later be the best of friends. -Eden
November 11, 2002 7:48:36 PM

Quote:
Most people thought it loses not more than 10% on incompatible programs, but geez!

You are confusing some things here. The performance decrease (down to -10%) you are referring take place because the HT optimizations don't work on some rare applications under HT aware OSes! The more important speed penalties under Win2000 observed by x86-secret are only due because win2k doesn't know how to run HT P4s and therefore trying it to run as a SMP CPU which it doesn't is, of course slows down everything. Again, the slowdowns under win2k have nothing to do with some rare speed decreases in real HT environments (XP as .DOT). So please don't mix up those two different things even if you are pro AMD and if you try to turn everything to your advantage.

Quote from the x86-secret review : "Sous Windows XP, la nouvelle technologie d'Intel ne semble en aucun cas provoquer une dégradation de performances lorsqu'elle est activée"

Summary: the reviewers couldn't find any speed degradation with the 3.06 GHz P4 under WinXP!


Quote:
Why does everyone keep saying HT2? THIS IS HT2! Why should it be any different on Prescott?

The later Xeon CPUs as also the HT P4s include the same HT technology (since they are the same core). This is version "1" of HT or "HT1". Intel is working in the meantime on a enhanced version of HT for their upcoming Prescott CPUs which is HT2. They hope on further increasing performance by making better use out of the L1 cache and by "simulating" 4 CPUs instead of 2 like with Xeon and P4 HT CPUs. Just read the last paragraph on the last page of x86-secret's 3.06 GHz review to verify this. The Inquirer also wrote about this but I can't find a link for the time being. Again, even if you are pro AMD, don't tell lies about the competition but try to be fair...


Quote:
Where are the MPEG compression and encoding benchmarks? What about LAME MP3? AutoCAD or anything that is 3d Rendering?
---------
Reports are the 3.06GHZ is out, if not in 3 days from today, I don't see how it's too early.

You are contradicting yourself. First you ask where all the real-world application benchies are but then you say that it's not too early to make a final judgement about the 3.06 P4 although there doesn't exist any comprehensive benchmark list. This doesn't make any sense. Concerning lake of complete benchmarks, after the NDA we will know more about the real performance of 3.06 P4s...


Quote:
These are what interest me, as well as the games benchmarks, which so far show that even SMP supported games don't deal with much improvement.

For the last time: "SMP" is NOT "SMT". So, you can't expect much improvements on SMP optimized games (Q3A and DOOM3 being the only two as far as I know) like you could expect out of HT optimized games which don't exist yet. Although SMP and SMT (or HT for the matter) increase performance through the use out of multithreading, the MT is implemented differently on both technologies.


Quote:
Nothing is known to date, AMD could as well develop a multithreading technique.

HT is not an extension set like SSE which can be added at any time to any cores if wanted. SMT is only possible on CPUs which have been designed from the beginning on to use this technology. Athlon XPs in any way are not able to do it and I doubt that the Hammer CPU was designed with SMT in mind...
November 11, 2002 10:31:00 PM

Quote:
things even if you are pro AMD and if you try to turn everything to your advantage.

I'll say this only once dude, don't ever start calling me pro-AMD. I may been that in the past, but not anymore, and I could right now dig up so many of my posts dating 1-2 months ago and pummel you with them until you CHOKE for what you claimed you think I am.
I'm not in the mood for this name-calling, and I know what I said. It ticks me off how some people just predicted how the 3.06GHZ was the howl of doom by Intel, and yet we see that day to day applications are not showing the extraordinary boosts. Yes it's a good increase, but please oh please just can it with the "Hyperthreading will kill AMD" crap.

Quote:
Intel is working in the meantime on a enhanced version of HT for their upcoming Prescott CPUs which is HT2.

Please back that up with some official proof. As far as I am concerned, Intel made sure HT is what it should be, starting 3.06GHZ.

Quote:
You are contradicting yourself. First you ask where all the real-world application benchies are but then you say that it's not too early to make a final judgement about the 3.06 P4 although there doesn't exist any comprehensive benchmark list

No I am not, Fugger himself said the 3.06GHZ is out, I know Spud even has them in his backstore, if not he should get them soon. You say it's too early to judge, but I say it isn't, BECAUSE apparently it is out, just that retailers will not display it until the official date of 14th Nov, otherwise why would a website be allowed to test it? Why on earth would Intel go let a french website benchmark a couple of tests, with LIMITS of NDA to what they test, when the website even used their own written benchmark?
Or are you saying they faked it?

Quote:
HT is not an extension set like SSE which can be added at any time to any cores if wanted. SMT is only possible on CPUs which have been designed from the beginning on to use this technology. Athlon XPs in any way are not able to do it and I doubt that the Hammer CPU was designed with SMT in mind...

From what I gather, HT takes 5% of the CPU die. Nobody ever said it wasn't possible, AMD probably would not use HT, but would have to come up with an SMT algorithm themselves, however you can't say it wasn't designed for SMT. If anything, as many had explained, K7 is more suited for HT than P4, due to the 9 IPCs.

My point out of all this, is that if this website is credible, then the tests are as well. And if the launch of the 3.06GHZ has the following performance about CURRENT APPLICATIONS, then I am right about one thing, people who said the 3.06GHZ would set such a huge gap, are dead wrong. We'll see in 3 days how everything turns out, however I just do not like it when there are some Intel fans who scream out the next CPU as the holy killer. Some names span to mind though I won't divulge them.
It's not like I am against it dude, if anything I ain't buying no AMD until a long time, not at least until I see a CPU from them that can be bought for a reasonable amount and has at least a 10% lead or more with some features like the P4 has for future application optimizations (so we're talking about some very very modified ClawHammers or K9s), otherwise I am all for an Intel system now, but I swear you say once more I am pro-AMD, and you'll have another thing coming at you.

--
The worst of enemies shall be prone to later be the best of friends. -Eden<P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 11/11/02 07:58 PM.</EM></FONT></P>
November 11, 2002 11:54:21 PM

Quote:
<i>Written by Eden</i>
I'll say this only once dude, don't ever start calling me pro-AMD. I may been that in the past, but not anymore, and I could right now dig up so many of my posts dating 1-2 months ago and pummel you with them until you CHOKE for what you claimed you think I am.
I'm not in the mood for this name-calling, and I know what I said. It ticks me off how some people just predicted how the 3.06GHZ was the howl of doom by Intel, and yet we see that day to day applications are not showing the extraordinary boosts. Yes it's a good increase, but please oh please just can it with the "Hyperthreading will kill AMD" crap.

Calm damn man. Nobody ever said that you'd see a 20% performance increase with HT in things like Microsoft Word, or Internet Explorer. "everyday" applications won't see a big benefit because they are not designed to stress the CPU alot. Some programs like photoshop may see a performance improvement, but again, it depends on whether programs like photoshop use alot of multiple threads. I said hyperthreading was going to be good, I never said it would "kill AMD". Think about it, if you're running linux 2.4 or WinXP, and you have a fully capable HT mobo, then you WILL see a performance benefit in applications that use more than one thread. on the 3.06 P4, HT is doing exactly what it was originally intended to do (on the Williamette).

Quote:
<i>Written by Eden</i>
Please back that up with some official proof. As far as I am concerned, Intel made sure HT is what it should be, starting 3.06GHZ.

By "official", I assume you mean Intel actually stating this. The fact is, Intel hasn't said a thing "officially" about Prescott. Most of the enthusiast sites speak of HT2 as being in Prescott. Anyways, I doubt it's a rumour and it can't be completely proved, but it's quite probable that Prescott will have HT2. As to what HT2 is, you are in fact quite wrong in saying that HT2 is on the P4 3.06 Ghz. On the 3.06 P4, it's a "revised" version of HT, NOT HT2. HT2 is what what previously known as PNI (Prescott New Instructions). Many sources several months ago said that
Prescott would feature PNI. It's also quite probable that Tejas will have the rumoured TNI (Tejas New Instructions). PNI was though to be SSE3, and was rumoured to boost performance in speech recognition software. Apparently, the PNI are now called HT2. They are supposed to improve HT performance in some way. Also, as <b>vacs</b> stated, Prescott will feature an improved L1 cache that will be improved in some way. Prescott was already rumoured to feature a modified L1 cache. The L1 in Prescott is supposed to co-operate alot better with HT than the current implementation. Prescott also has more L2 cache, and is rumoured to have several other micro-architechtural enhancements that should (in some way) complement the HT2 instructions.

Quote:
<i>Written by Eden</i>
No I am not, Fugger himself said the 3.06GHZ is out, I know Spud even has them in his backstore, if not he should get them soon. You say it's too early to judge, but I say it isn't, BECAUSE apparently it is out, just that retailers will not display it until the official date of 14th Nov, otherwise why would a website be allowed to test it? Why on earth would Intel go let a french website benchmark a couple of tests, with LIMITS of NDA to what they test, when the website even used their own written benchmark?
Or are you saying they faked it?

Even if the 3.06 is out, most sites still have not reviewed it, so there isn't enough benches/evidence to judge the improve HT gives.

Quote:
<i>Written by Eden</i>
From what I gather, HT takes 5% of the CPU die. Nobody ever said it wasn't possible, AMD probably would not use HT, but would have to come up with an SMT algorithm themselves, however you can't say it wasn't designed for SMT. If anything, as many had explained, K7 is more suited for HT than P4, due to the 9 IPCs.

Actually, the P7 Williamette <i>was</i> designed with HT <b>in mind</b>. All Williamettes have the original HT version on the die.

BTW, Intel is very serious about SMT. Nehalem, Intel's upcoming P8 core, is rumoured to feature sophisticated power management a la Banias, and to "greatly expand" on SMT (HT).






- - -
<font color=green>All good things must come to an end … so they can be replaced by better things! :wink: </font color=green>
November 12, 2002 12:05:01 AM

well buddy, I know that the 3.06 is available since last week, why are you trying to convince me? I was just trying to say that one shouldn't make a final verdict about a product after reading the first review of it and having looked at the results of a handful benchmarks. It's just not fair unless you own one and have already done some comprehensive tests with that CPU.

There's certainly more about HT then those few benchmarks but of course it won't be the "Hyperthreading will kill AMD" thing you were talking about. HT is just a nice performance boost to some applications/games nothing more, nothing less. I doubt that x86-secret has revealed it all although I have no doubt about the accuracy and correctness of this review!

BTW. If you don't want people to call you "pro AMD" then you should better say from time to time something good about Intel, otherwise other readers who didn't read any of your previous posts might believe that you are "pro AMD"...
November 12, 2002 1:56:01 AM

Quote:
BTW. If you don't want people to call you "pro AMD" then you should better say from time to time something good about Intel, otherwise other readers who didn't read any of your previous posts might believe that you are "pro AMD"...

This is exactly what I said by pummeling and by all means, either you or I, go dig up my posts in the past month, and when you finally weigh in my comments, please, your honor, give me the verdict. I shall enjoy laughing.
You've got NO evidence that I am pro-AMD simply by looking at this thread.
Dark is pro-Intel and is easily backed by almost any post he has, but you have NO PROOF by simply limiting yourself to this thread and my comments here. And I seriously mean it, when you find out about my posts and reach a verdict, IF I am not proven pro-AMD, you will be apologizing BIG TIME as I do not tolerate such comments without substantial proof.

Quote:
well buddy, I know that the 3.06 is available since last week, why are you trying to convince me? I was just trying to say that one shouldn't make a final verdict about a product after reading the first review of it and having looked at the results of a handful benchmarks. It's just not fair unless you own one and have already done some comprehensive tests with that CPU.


I agree, however I was commenting on the content that article had shown. Showing gaming benchmarks including a synthetic one like 3dMark 2001 which so many seem to venerate, as well as ScienceMark and Scientific benchmarking, does speak volumes to me. However yes I am not fully convinced, you are right about that, and that I shouldn't just judge things. However these benchmarks are quite credible in my eyes, and it kind of paints a small portion of the big picture. I'll wait for the 14th, and then I'll gladly retract what I said, SHOULD the HT 3.06GHZ have an average of 15% IPC improvement.

Quote:
HT is just a nice performance boost to some applications/games nothing more, nothing less.

And I agree, however I simply did not like it how it seemed like this thing is a be-all end-all technology.
I can rant on reciprocatly about how annoying those idiots who post "OMG HAMMER IS GONNA DESTROY INTEL DOWN, AMD OwnS j00!!!" just so you'd know I am not just angry at Intel zealots, but AMD fanboys annoy me as much if not more. I can withstand a good part of Meltdown than those who are so close minded about the P4.


--
*You can do anything you set your mind to man. -Eminem
November 12, 2002 2:06:04 AM

Quote:
By "official", I assume you mean Intel actually stating this. The fact is, Intel hasn't said a thing "officially" about Prescott. Most of the enthusiast sites speak of HT2 as being in Prescott. Anyways, I doubt it's a rumour and it can't be completely proved, but it's quite probable that Prescott will have HT2. As to what HT2 is, you are in fact quite wrong in saying that HT2 is on the P4 3.06 Ghz. On the 3.06 P4, it's a "revised" version of HT, NOT HT2. HT2 is what what previously known as PNI (Prescott New Instructions). Many sources several months ago said that
Prescott would feature PNI. It's also quite probable that Tejas will have the rumoured TNI (Tejas New Instructions). PNI was though to be SSE3, and was rumoured to boost performance in speech recognition software. Apparently, the PNI are now called HT2. They are supposed to improve HT performance in some way. Also, as vacs stated, Prescott will feature an improved L1 cache that will be improved in some way. Prescott was already rumoured to feature a modified L1 cache. The L1 in Prescott is supposed to co-operate alot better with HT than the current implementation. Prescott also has more L2 cache, and is rumoured to have several other micro-architechtural enhancements that should (in some way) complement the HT2 instructions.

I could also start a rumor that Hammers will use SMT technology. I have no proof to back it up. You see where I am going?
I trust as far as official sources go to, and that includes Intel, and possibly credible HW websites. Anandtech heard Intel stating the Prescott uses a bigger cache, higher bus and small micro-architectural enhancements. Assuming Intel stated Prescott is a SMALL enhancement, those micro-architectural boosts are probably in the L1 cache, and shouldn't be as beneficial as like, adding an FPU.

Quote:
Actually, the P7 Williamette was designed with HT in mind. All Williamettes have the original HT version on the die.

I know that, but what exactly is your point?
My point was to say that the HT can be designed for K7 as well, and technically said, HT is better designed for K7 than P7, simply because one has more IPC. And don't counter that as you have been countered previously.

As I previously said, you are like Juin, you look at cores that are years away from now, that anything could occur to them anytime. I'm not saying it's wrong, but it sure as hell shows that you hype Intel's cores too much, that any core name in the next decade will fuel pump you.
T'other day juin mentioned the Bandera core, and I was like, where the hell did he know about THIS core? I even asked myself if we will ever hear about it in the next 3 years! :eek: 

--
*You can do anything you set your mind to man. -Eminem
November 12, 2002 2:09:05 AM

A few things worth mentioning. Games don't benefit from multithreading much period. There is no "optimization" method. The program is serial by nature. You are pretty much dealing with tons of dependent FP instructions in a single thread. I'm surprised UT2003 even noticed a 5% increase in performance with HT. It may suggest that WinXP is indeed employing this "psuedo-hyperthreading" feature.
Secondly, there were some mp3 encoding benchmarks in that review and the increase is quite impressive to say the least. I'm still waiting for the video encoding benches which Aces has promised and I'm sure both Anand and Tom will do but things are looking quite good.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
November 12, 2002 10:46:36 AM

There, that's all I wanted.
The video benches should speak much more. Although there were MP3 benchmarks, I never had heard of the program and it had a weird way of benching (reverse highest score wins), but I think I'll wait for LAME and some conversion benchies.

At least now there is a major reason for going to get WindowsXP for future enthusiasts. However how will this chip be used in systems who use Windows 2000 or below? As in, how will OEMs or simply users who get it, will disabled HT before it kind of sucks most of the performance off their applications?

--
*You can do anything you set your mind to man. -Eminem
November 12, 2002 4:53:08 PM

I already read that Xeon HT review when it first came out in September, but I now would like to see some more P4 3.06 GHz HT reviews instead of 2.0 or 2.4 GHz Xeon benchmarks in case Intel revised indeed HT as it is claimed with the newly released P4. Also one shouldn't forget that Xeons and P4s are not exactly the same CPUs (although they are based on the same core) because Xeons offer different sizes of Level 2 and 3 caches. Well, in two days we will know more...

I don't believe that many 3.06GHz owners will try to run the CPU on win2k because as <b>wschuerm</b> said: if you have the cash to buy a new 3.06 GHz P4 then you will also have the cash (and hopefully the knowledge) to buy WinXP.
November 12, 2002 5:18:20 PM

There are two kinds of Xeons: Xeon DP (dual processor) and Xeon MP (multi processor). Xeon DPs are just HT enabled P4s with more pins on the package. Same cache.

SMT is much like SMP - multithreaded applications benefit and others don't. As more and more apps become multithreaded, the more useful these chips will be. It's like SSE2. No one is going to optimize software to run on a chip that isn't built yet. No one will write DX10 games before DX10 video cards come out, even if the standard has been approved. In three years I'm sure most apps will be multithreaded and then we'll all look back and think to ourselves "wow, good thing intel started selling HT chips back in '02".

- JW
November 12, 2002 6:07:59 PM

All the Xeons have HT enabled. The Xeon MP's have a larger L3 cache and are capable of scaling to more processors while the Xeon DP's only go in dual processor solutions and have a lesser L3 cache. However, they are the same chip.
While the Xeons do have different pin configuration than the P4 it is indeed the same core design. Intel enabled the first generation of HT on the Xeons and the result was mixed. They will probably take the revised design that's on the P4 and implement it in future Xeon revisions probably in the form of a new stepping as well.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
November 12, 2002 6:55:34 PM

Quote:
Although there were MP3 benchmarks, I never had heard of the program and it had a weird way of benching (reverse highest score wins), but I think I'll wait for LAME and some conversion benchies.


I have this program. Based on earlier LAME codec, have MMX, 3D Now!, 3D Now! Extention, SSE and SMP oprimizations. Freeware, open source. Development of this project has stopped. Latest version 2.39c @ Dec31, 2001. This is the reason for not having SSE2 optimization is this codec.

This encoder is fast as hell. It encodes MP3 files at <b> 115x </b> (average) in my friends KT333 + AXP1800 + PC2100. In terms of sound quality, it sucks. This encoder produced worst sounding MP3 files I have ever listened to. The reason for producing bad sounding MP3 files may be using earlier verion of LAME. It may have sounded better if it was based on LAME 3.92.

Optimizations of this codec is very good. It boosted performance by 5 times with 3D Now! enabled in my older K6-2 450 MHz. Now it boosts performace by 2 times in My current Duron 1 GHz (by enabling 3D Now! Extention or SSE). SMP optimization of this codec may be as good as SIMD optimization.

Although it looks like real world benchmark, this is actually a synthetic benchmark. Nobody will want to use it for creating MP3 files in real life. I myself use it as a benchmark program only.

Let us know <A HREF="http://forumz.tomshardware.com/community/modules.php?na..." target="_new"> What File compression format you use? </A>
November 13, 2002 12:48:28 AM

Quote:
<i>Written by Eden</i>
This is exactly what I said by pummeling and by all means, either you or I, go dig up my posts in the past month, and when you finally weigh in my comments, please, your honor, give me the verdict. I shall enjoy laughing.
You've got NO evidence that I am pro-AMD simply by looking at this thread.
Dark is pro-Intel and is easily backed by almost any post he has, but you have NO PROOF by simply limiting yourself to this thread and my comments here. And I seriously mean it, when you find out about my posts and reach a verdict, IF I am not proven pro-AMD, you will be apologizing BIG TIME as I do not tolerate such comments without substantial proof.

Ok. Most people already know this. I favor Intel over AMD. There, I said it. You want to know why? Intel is the industry leader for a reason. I favour Intel because Intel does many things that AMD does not. Intel has many things that AMD does not. Whethere it's fair or not, biased or not, I don't want to talk about it. This is the truth, and nobody can change that. Yes, I feel sorry for AMD, yes I want AMD to be around for competition, and especially to keep pushing Intel to innovate due to competition. But, Intel makes well rounded chips. More so than AMD. Intel's chipsets are more stable than AMD's. P4's have less compatibility problems than Athlon XP's. Yes, I know many of them are software-related, and many AMD fans blame Microsoft for poor support, but the facts can't be changed. Athlons still have a few compatibility problems (even if they are only software). And you know what? Intel doesn't point fingers; Intel simply does an insane amount of software testing to ensure P4's are stable and compatible. AMd points fingers, and AMD fans blame everyone except AMD itself. I understand AMD is financially limited, but again, you cannot change the facts. P4's are more software compatible than Athlon XP's, whether you (or I) like it or not. Also, about the whole thermal protection debate. AMD blames their thermal protection problems on the mobo makers, saying that they did not properly implement the protection. Alright, that's understandable. Intel, though, took matters into their own hands and put on-die protection on the P4's. I know AMD may not be able to implement the protection on die for several reasons, but the fact is, Intel's P4s already have it on-die. As I have said before, the facts can't be changed. If you want to ramble all day about how AMD's protection is good <b>if</b> properly implemented, then fine. I have no problems with that. With AMD CPU's though, it's always alot of <b>if's</b>. What if you don't have the win2k Athlon patch? What if your games freeze running DirectX 8.1 on an Athlon XP? What if your Athlon's thermal protection doesn't operate properly because the mobo doesn't support it? If you buy a P4, there are very few if's, compared to an Athlon. Hopefully, anyone who's reading this understands now why I favor Intel.

Also, I understand that Intel's CPU's cost more than AMD's. This difference has recently become smaller due to Intel getting more competitive, and due to AMD actually adopting a <b>profititable</b> pricing structure, as opposed to it's previous pricing structures. Intel does more testing on their CPU's, and uses a more profitable pricing structure than AMD. Intel also spends more on R&D than AMD. Basically, that's why you pay more for an Intel CPU, compared to an AMD CPU. It's as simple as that.

Quote:
<i>Written by Eden</i>
I agree, however I was commenting on the content that article had shown. Showing gaming benchmarks including a synthetic one like 3dMark 2001 which so many seem to venerate, as well as ScienceMark and Scientific benchmarking, does speak volumes to me. However yes I am not fully convinced, you are right about that, and that I shouldn't just judge things. However these benchmarks are quite credible in my eyes, and it kind of paints a small portion of the big picture. I'll wait for the 14th, and then I'll gladly retract what I said, SHOULD the HT 3.06GHZ have an average of 15% IPC improvement.

Let me ask you something. Have YOU ever heard of sciencemark or CPUbench? Do you EVEN know what kind of benchmarks they are running? Personally, I have never seen sciencemark or CPUbench benchmarks before, and I'm not going to judge the performance of HT based on some strange synthetic benches that I'm seeing for the first time. See my point?

Quote:
<i>Written by Eden</i>
And I agree, however I simply did not like it how it seemed like this thing is a be-all end-all technology.
I can rant on reciprocatly about how annoying those idiots who post "OMG HAMMER IS GONNA DESTROY INTEL DOWN, AMD OwnS j00!!!" just so you'd know I am not just angry at Intel zealots, but AMD fanboys annoy me as much if not more. I can withstand a good part of Meltdown than those who are so close minded about the P4.

I'm sorry, but now you're ranting. Apart from Meltdown, can you <b>specifically</b> show me someone who was hyping HT to be the "be-all end-all technology"?

Quote:
<i>Written by Eden</i>
I could also start a rumor that Hammers will use SMT technology. I have no proof to back it up. You see where I am going?
I trust as far as official sources go to, and that includes Intel, and possibly credible HW websites. Anandtech heard Intel stating the Prescott uses a bigger cache, higher bus and small micro-architectural enhancements. Assuming Intel stated Prescott is a SMALL enhancement, those micro-architectural boosts are probably in the L1 cache, and shouldn't be as beneficial as like, adding an FPU.

It seems I made a mistake.

Correction: Intel HAS stated that Prescott will have HT2

Take a look at <A HREF="http://images.visualwebcaster.com/8064/pdf/highres.pdf" target="_new">this</A> link. <i>Note: this PDF may take a while to load.</i> FYI, this is an Intel PDF. Also, check out <A HREF="http://www.geek.com/news/geeknews/2002Oct/bch2002102501..." target="_new">*this*</A> link for info related to this PDF and HT2. Also, I have the following link for the Inquirer about HT2 and PNI:

<A HREF="http://www.theinquirer.net/?article=3308" target="_new">http://www.theinquirer.net/?article=3308&lt;/A>

On top of these links, I have 2 Aces forums links which feature some interesting discussions related to HT2, PNI, and a few other things. Here they are:

<A HREF="http://www.aceshardware.com/forum?read=80032905#TEXT" target="_new">http://www.aceshardware.com/forum?read=80032905#TEXT&lt;/A>
<A HREF="http://www.aceshardware.com/forum?read=80035846" target="_new">http://www.aceshardware.com/forum?read=80035846&lt;/A>

Also, Even though Intel did state that Prescott was going to be a small core change, Intel did say it would be a <b>bigger</b> change than northwood. And of course, Tejas is intented to be the biggest core change in the P7 family.

Quote:
<i>Written by Eden</i>
As I previously said, you are like Juin, you look at cores that are years away from now, that anything could occur to them anytime. I'm not saying it's wrong, but it sure as hell shows that you hype Intel's cores too much, that any core name in the next decade will fuel pump you.
T'other day juin mentioned the Bandera core, and I was like, where the hell did he know about THIS core? I even asked myself if we will ever hear about it in the next 3 years!

Ok, even if I did favor AMD, what is there to hype? 0.09 Hammers? Ohh wow, that sounds exciting. Also, I don't hype the same thing 100 times. Although I may have hyped Prescott a few times, it was about a different thing each time. I'm not going to hype Barton or Hammer, since both of them have been overhyped and all the details about them are pretty much known. Prescott, on the other hand, is a more mysterious beast than Hammer. Who knows what kind of improvements will Prescott exactly have?

And about the whole HT performance debate, the fact is, HT won't give a huge performance boost in a single application (unless it heavily uses multiple threads). Intel stated that you can get up to a 70% performance boost on HT, ar about 25% on average. The only way you'll get the maximum performance out of HT is if you multi-task with several programs that utilize multi-threading. HT shows it's true potential when you're running several programs/games at the same time. And HT is a dream come true for a user like me, since I multi-task very often.




- - -
<font color=green>All good things must come to an end … so they can be replaced by better things! :wink: </font color=green>
November 13, 2002 1:24:51 AM

Look dude, it doesn't matter what you think, you are hell bent when it comes to just favoring small things.
Like, seriously man, stability, I run a damn 1600+ at 55ºC, a VIA chipset, and SBLive. Sounds like the haz-mat call for some, because I never got any problems. Your problem is you're fragile to that, one of the main reasons for many pro-Intel users.
Well whatever, it's who you are. I was simply pointing out to VACS that what he claimed about me is in no way as provable as you. I know you're pro-Intel but I used you as a known example to show him why he's wrong to simply assume.

Quote:
Let me ask you something. Have YOU ever heard of sciencemark or CPUbench? Do you EVEN know what kind of benchmarks they are running? Personally, I have never seen sciencemark or CPUbench benchmarks before, and I'm not going to judge the performance of HT based on some strange synthetic benches that I'm seeing for the first time. See my point?

No I don't see your point Dark. I said myself it's ok, I'll wait for the big time benchmarks.


Quote:
I'm sorry, but now you're ranting. Apart from Meltdown, can you specifically show me someone who was hyping HT to be the "be-all end-all technology"?

No, I shan't, I don't feel like keeping this up. I will admit I was in quite the mood monday, for some odd reasons. I feel a bit better now, but whatever, I know I acted quite dumb with my remarks in the begininng and I lashed out.

Quote:
Ok, even if I did favor AMD, what is there to hype? 0.09 Hammers? Ohh wow, that sounds exciting. Also, I don't hype the same thing 100 times. Although I may have hyped Prescott a few times, it was about a different thing each time.

What do you think this is, a CPU priesting conversion session?
I can't believe you're trying to even implicitly make it show as if I am trying to tell you to FAVOR AMD. FAVOR is outlined to show you that I don't even bother to wonder why. I am neutral, period. You should be, unfortunatly you're not.
And btw when you say what's to hype about AMD, it just shows you are doing so for Intel. Wake up Dark, you look like an Intel marketting guy. We don't need those people here, we need neutral discussions about technologies, not a constant "Intel stated this, Intel even stated that, OMG now Intel is adding this...." that rings of the name Dark_Archonis.

--
*You can do anything you set your mind to man. -Eminem
November 13, 2002 2:28:07 AM

LOL yep. Hyperthreading or no hyperthreading you still only have one processor with a finite number of Alu/FPU/MMX/SSE registers.

THining about it though, hyperthreading would give the best benefits fro apps that need all the pipelines.

<b>LHGPooBaa + Evil Hamster Sidekick: Serving Toms Hardware community for 2 years as of the 11th of November</b>
November 13, 2002 2:48:44 AM

Quote:
Yes, I feel sorry for AMD, yes I want AMD to be around for competition, and especially to keep pushing Intel to innovate due to competition

So why you are not buying an AMD chip?

Quote:
Intel's chipsets are more stable than AMD's

AMD 750/760 chisets was stable as hell. Same thing applies for AMD's MP chipset. And I dare to tell you that nForce chipsets are not a bit less stable than Intel chipsets.

Quote:
P4's have less compatibility problems than Athlon XP's.

Find me a app that doesn't run in a Athlon and runs in P4. Are you telling about SSE2 compatibility? Ok, nobody claims that AXPs have SSE2.

Quote:
What if you don't have the win2k Athlon patch?

Is this Athlon patch costs $$$ and extremely hard to find? Who uses Win2000 now without Service Pack? I don't need to have the win2k Athlon patch. I have integrated Service Pack 3 with windows 2000 installation files.

Quote:
What if your games freeze running DirectX 8.1 on an Athlon XP?

I had 0 lockups in games w/ DX8.1 with my Duron 1 GHz + nForce. My friends AXP 1800+ w/KT333 also doesn't lock up in games.

Quote:
What if your Athlon's thermal protection doesn't operate properly because the mobo doesn't support it?

I don't care because I am still to find a guy who had a Athlon died thermal-death due to ineffective thermal protection.

Quote:
Apart from Meltdown, can you specifically show me someone who was hyping HT to be the "be-all end-all technology"?

You (and juin) were one of these people who loved to think that HT is going to kill the competition from AMD.

Quote:
Also, I don't hype the same thing 100 times

How many times you talked about Tejas? You told about Nehalem, which is similar to talking about K9.

Quote:
HT shows it's true potential when you're running several programs/games at the same time.

Ok, can you play two games at the same time? Or want to encode video while playing game? Video Encding will take same time when you are playing game or doing 3D rendering with 3D Studio Max?

Let us know <A HREF="http://forumz.tomshardware.com/community/modules.php?na..." target="_new"> What File compression format you use? </A>
November 13, 2002 3:07:21 AM

Oh no, and so it begins.........

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
November 13, 2002 9:27:48 AM

:lol: 
Did I mention that ATI sucks, Via is good and Fords are much better than general motor cars?


<b>All CDs will be protected and you are a filthy pirate! - <i>Bertelsmann Music Group (BMG)</i>
Serving THGC for over 2 years.</b>
November 13, 2002 10:37:37 AM

Ford? Are you joking? maybe in the US but here in Europe Ford s.x big time. They are still selling cars with a fixed rear axis. Where do you still find this crap? They don't even know what grip is...

I'm still wondering how Ford could win several prices for best car of the year. I believe bribe has something to do with it :) 
November 13, 2002 2:50:58 PM

were talking computers here not cars theyre to expensive to tweak or mess around with (altough computers are 2 when i think about it)

Remember the time You used 20Mb harddrives
November 13, 2002 8:11:03 PM

Quote:
They are still selling cars with a fixed rear axis.

Here in the U.S. big engines (>5 liters) are quite common especially in trucks and it's expensive to manufacture an independent rear axle strong enough to handle the torque! Hence solid axles... In Europe they shouldn't use solid axles because engines are smaller and don't generate as much torque. Not sure why Ford is doing it there.

Ritesh

<P ID="edit"><FONT SIZE=-1><EM>Edited by ritesh_laud on 11/13/02 04:37 PM.</EM></FONT></P>
November 13, 2002 9:00:13 PM

As for getting un-official, but direct stuff on Prestonia, I got a bunch of my questions answered by contacting Intel Pre-sales department. They were really helpful on issues like HT2, projected Xeon FSB releases ("soon"), &c.
November 13, 2002 10:38:17 PM

Yes i was joking. I was saying that with tonge so firmly implanted in cheek it looked as if i had eaten a snooker ball!

<b>All CDs will be protected and you are a filthy pirate! - <i>Bertelsmann Music Group (BMG)</i>
Serving THGC for over 2 years.</b>
November 14, 2002 12:42:07 AM

One thing that makes me very disappointed at Intel right now, is how they only put HT for the 3.06GHZ. Yes that means only those so stupid to shell out 700$ US for it (Spud in Canada claims he has it for 1100$ CDN at his backstore, say hello to the P3 1GHZ price days) will get it. And that plain bugs me. The worse has yet to come:
It takes a LONG time relatively to get into mid-range pricing which will still be a lot. And THEN it will still take more time for the FIRST of the serie so far, the 3.06GHZ to get to low end that by then, HT in my eyes is a dream far away, nearly forgotten. All I'm saying is, Intel HAS to release sub 3GHZ HT enabled CPUs, otherwise we're talking about released virtually inaccessible technology, which is downright sad!

--
*You can do anything you set your mind to man. -Eminem<P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 11/13/02 09:43 PM.</EM></FONT></P>
November 14, 2002 1:19:48 AM

I'm betting it's a C1 stepping thing. I think we're supposed to officially find out on the 14th.

and then there's this from <A HREF="http://www.vr-zone.com/#2742" target="_new">Virtual Zone</A> and this from <A HREF="http://www.overclockers.com/tips00184/" target="_new">OC'ers.</A>

To <font color=blue>be behind</font color=blue> in <font color=white>one's</font color=white> <font color=green>hindsight</font color=green> is <font color=blue>but</font color=blue> a <font color=red>retrospection</font color=red> in <font color=white>one's</font color=white> <font color=blue>backward</font color=blue> <font color=purple>vision.</font color=purple>
November 14, 2002 1:39:49 AM

In other words, there has been virtually nothing fixed in the HT 3.06GHZ right?

Why oh why do people seem to midguide us into thinking it has been revamped!

Still, if it can be enabled, that will make BIG + for my friend whom I'm helping get a system for Xmas. Though I hope that's the same HT, and that it can be tested for same performance boosts under WinXP.

--
*You can do anything you set your mind to man. -Eminem
November 14, 2002 1:54:14 AM

Yoohoo, we're days later now, have you read what happened since?

--
*You can do anything you set your mind to man. -Eminem
November 14, 2002 1:56:50 AM

For the HT topic only, you started from saying it is useless in terms of improving performance, to commenting it is not accessible to general public. God knows what you are gonna say tomorrow. Are you spreading typical AMD FUD?
November 14, 2002 2:00:18 AM

Another thing I found interesting, was how the temperatures were for the 3.06GHZ with HT and above.

It pretty much confirms my thought about HT's effect on temps. If it doesn't then it still shows that P4s do run hot at some point. XP2800 probably has as much temps.
What strikes me as odd is that for the 3.6GHZ which did not run stable, required 1.8V, and we're looking at 600MHZ of OCing. This isn't making any sense, for its micron AND compared to how far OCing was easy on 1.5xV earlier on. I could be wrong and this could be normal, though reacing 1.8V so fast in 600MHZ, and at 6xºC, I guess that's why Intel wants 0.09m.

We need Fugger's thoughts on this, he tweaks so often these CPUs and their speeds!
--
*You can do anything you set your mind to man. -Eminem<P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 11/13/02 11:02 PM.</EM></FONT></P>
November 14, 2002 2:01:44 AM

Run along...

--
*You can do anything you set your mind to man. -Eminem
November 14, 2002 2:05:38 AM

The reason why HT performed so bad under win2000 and q3a with r_smp enabled is quite obvious. Both apps don't support HT and therefor enabling SMP (not SMT which HT is) can only degrade performance.

yes win 2k dont help but for quake and most of the games Ht decrease overall bandwith from memory and quake is bandwith hungry.For me i all run SETI wimamp and my actual aps 1 or more so i have all treads tht i need to run.I was expecting more decrease of performance.

Personaly HT is a bad impletation of SMP but is the 1 of the group HT will be better HT 3 will be king around that AMd will do it rip-off.

Now what to do??
November 14, 2002 2:07:31 AM

You already look forward to HT version 3?! Holy crap...

--
*You can do anything you set your mind to man. -Eminem
November 14, 2002 2:07:31 AM

Judging the performance of a CPU from the nominal "C1" stepping given by some software is purely garbage. In real CPU manufacturing term, stepping is more related to the backend, frame change or the metal layer arrangement. Stepping change can improve speed. What related to HT is more the transistor arrangement. Designers keep updating it to fix the bug, in this case, to make HT function from disabled to working well. Stepping plus transistor rev change is what keeps pushing CPU performance forward. P4 with different transistor arrangement could have same "stepping" reported by CPUID.
November 14, 2002 2:10:43 AM

Only a few technical hours before THG releases or SHOULD release the CPU review. That'll finally be something to look at and drift this conversation away from the stuff I seemed to have started and the others throwing more mud at to make things worse.
I'm expecting some good MPEG performance boosts, but most of all when multitasking while encoding like the IDF video showed.


Hmm I forgot to check OCers.com, so it does seem like there should be ways to get the P4s to run HT quite easily. I certainly hope that's true, I really do. Hell, since 2.5GHZ 400MHZ FSB are confirmed C1s new generation, this would be a perfect match of performance+OCability+HT+Granite Bay. Now that's something to gaze at. I wouldn't expect 700$ CPUs to sell so easily, and "prepare" the homes for HT any soon, as contradictory as it is.
--
*You can do anything you set your mind to man. -Eminem<P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 11/13/02 11:15 PM.</EM></FONT></P>
November 14, 2002 10:12:27 AM

Intel has already stated that it will enable HT on all P4 2.4 GHz and upwards (could also be 2.0 GHz and upwards, I can't rememeber exactly). The question is just when this CPUs will be available.

Like VR-Zone.com demonstrates it, everyone can enable HT on their current 2.x GHz P4s. You just need to have an chipset supporting HT (all Intel 533MHz FSB chipsets support HT) and Windows XP. Of course you still need to know the address in the BIOS where the information whether HT is enabled or not is stored...
November 14, 2002 10:54:01 AM

Although it's still too early to make a final judgment about HT, now after having read (sometimes very quickly) over a dozen of reviews, I'm quite surprised how it turned out. Of course HT isn't a miracle which doubles performance but it isn't also the performance stopper some have feared.

Quote:
<A HREF="http://www.aceshardware.com/read.jsp?id=50000319" target="_new">Ace's Hardware</A>
It is quite remarkable how almost every single threaded benchmark still got a small performance boost from HyperThreading, between 1 and 5%. This shows that HyperThreading has matured as it almost never decreased performance, as it did in the first hyperthreaded Xeons.

Most reviewers seem to agree that the performance increase ranges from 5% to almost 30% on top of the 3.06GHz, that also singlethreaded application will profit from HT and that it should have a bright future since the HT increase in performance also increases with a higher clocked P4.

Quote:
<A HREF="http://www.hardocp.com/article.html?art=Mzg4" target="_new">HardOCP</A>
Honestly, we were shocked at the large performance gains with HT enabled in our normal suite of benchmarks

It's a certainly a good thing that Intel finally released HT enabled P4. Now software developpers can start optimizing their progies and when they get available, CPU prices will have dropped so that also "normal" users can afford HT P4s sometime later next year.
November 14, 2002 11:41:12 AM

You guys just wait until the Via C4 comes out with quantum threading.... That will definately be the intal/AMD KILLER!!!!!!! Muahahahahaha....

[rant]AMD, YOU ARE EVIL!!! GIVE US WHAT WE WANT!!! A STOCK, UNLOCKED CPU!!![/rant]
Anonymous
a b à CPUs
November 14, 2002 1:58:13 PM

> With AMD CPU's though, it's always alot of if's.

I think you put that entire paragraph explaining your preference for intel quite well. Congrats. Not that I agree with you however, since to me, these "IF" you refer to are extremely minor and easy to avoid. Just picking up a decent modern motherboard and installing XP or Linux gets rid of 99% of your ifs (and ironically, you need to do EXACTLY the same thing to run a 3 GHz P4 with HT enabled. An "old" and/or cheap P4 motherboard is likely not to meet the new P4s power requirements, and win2k doesnt properly support HT. And no patching or carefull HSF installing can get around those IFs.)

As for these incompatibilities.. Im tired of hearing about that. Im sure you'll be able to find one or two obscure pieces of software that check for the cpu some intel ID and bomb in a BSOD when it doesnt detect an intel cpu. In fact, one should be glad such a piece of junk coded software doesnt run on your PC (taking this over the top, one could say: "oh no, Norton Antivirus makes your PC completely incompatible ! Thousands and thousands of apps wont work anymore, all these nice litlle virusses, trojan horses, .." :) 

Besides, a small quote from THG 3 Ghz P4 review: "The 3.06 GHz P4 HT will not even run the 3D Mark 2000". I aint saying the P4 is "incompatible" (with what anyway ?) but it makes you wonder nevertheless...

>Let me ask you something. Have YOU ever heard of
>sciencemark or CPUbench? Do you EVEN know what kind of
>benchmarks they are running?

Ahem.. yes. They are quite common really, and if you'd read reviews on other websites besides THG, you'd know them. Go over at lost circuits, x-bit labs, Aces',.. besides.. Do you know exactly what Sysmark measures ? I dont.. in fact, no one does besides Bapco and its partners. If you want to know what ScienceMark measures, download the source, and check it out yourself.

>Intel does more testing on their CPU's

Can you give some proof of this ? Though it would be almost natural to assume this, the evidence doesnt exactly support this (eg caminogate, 1.13 Ghz P3, FDIV, etc, etc). Im not saying your statement is false, but I challenge you to back it up. I'd even dare say AMDs track record so far, has been better than Intels; the only issue I can think off, was some obscure JPEG problem on some K6s, though Im still not sure this has ever been confirmed. If AMD has a worse reputation than Intel on QC, they own it all to VIA, not to themselves.

= The views stated herein are my personal views, and not necessarily the views of my wife. =<P ID="edit"><FONT SIZE=-1><EM>Edited by bbaeyens on 11/14/02 11:07 AM.</EM></FONT></P>
November 14, 2002 3:12:52 PM

Damn guys...get over it. Its a cpu a friggin cpu. Its not your girlfriend, its not your granny. Its not personal! You guys are the same type of people that make my Monday morning group meetings suck. One word in the wrong place, and immediately its a pissing contest. I'm tired of all the WHINING!!! This is a forum, not a redneck bar (ford is better than chevy, is not...is too..is not...ect). I've often wished that I had total recall, but seeing how you guys use peoples past statements as weapons, I'm friggin glad that half of what I say goes unoticed and the other half gets forgotten. Its fun to debate issues, but not peoples character or motivation. Eveyone...breathe deep...take it down a notch... its all good...life's too short to argue over hyperthreading of all things...

[rant]AMD, YOU ARE EVIL!!! GIVE US WHAT WE WANT!!! A STOCK, UNLOCKED CPU!!![/rant]
Anonymous
a b à CPUs
November 14, 2002 3:20:07 PM

>[rant]AMD, YOU ARE EVIL!!! GIVE US WHAT WE WANT!!! A
>STOCK, UNLOCKED CPU!!![/rant]

YOu can buy them you know.. they're called Athlon MPs :) 

= The views stated herein are my personal views, and not necessarily the views of my wife. =
      • 1 / 3
      • 2
      • 3
      • Newest
!