When Tom's posted the first review on the Pentium 4, I had been happy to see Dr. Thomas Pabst barely even jibe at Intel. It had seemed as though he understood Intel was trying to establish a new standard in SSE2 with their P4 chip.

But now, after reading the update to the P4 review, I can only be disappointed with Dr. Thomas Pabst. His clearly obvious biased statements against Intel are saddening.

He completely fails to admit that the P4 is not only incorporating SSE2, but is DESIGNED around SSE2. And that without SSE2 compatible software, the P4's results are less than ideal.

He completely fails to inform the user that his benchmarks will be completely unimportant once software is compiled under SSE2. And that once SSE2 becomes a new standard, it will be incorporated into nearly all x86 software, making the Pentium 4's poor non-SSE2 FPU performance virtually unnoticeable because very little software will be using the non-SSE2 FPU.

And he completely fails to mention to the user the fact that at least some of the test results were intrisically biased against Intel because the software being run had optimizations for AMD's 3DNow, but had absolutely no optimizations for Intel's SSE2. So of course software optimized to run on an AMD chip and not on a Pentium4 chip is going to give better results to the AMD chip.

All of this disheartens me. If Dr. Thomas Pabst cannot even inform his readers that his Pentium 4 review is meaningless once SSE2 becomes a standard, then he is clearly biasing his opinions against Intel.

What seems obvious to me is that the Pentium 4 is engineered completely around SSE2 and that while running SSE2 optimized code, it will out perform any other chip, even if that chip has support for SSE2.

The Pentium 4 chip is designed for the market of any software that runs SSE2 optimized software. And it appears to me that the Pentium 4 will reign supreme in that market.

Which leaves AMD's chips to reign supreme in the non-SSE2 optimized market and low-cost market.

So it looks to me as though the Pentium 4 will be a valuable CPU in the future, and has a definite purpose, despite Dr. Thomas Pabst's articles.

- Anything can be fixed with duct tape, a swiss army knife, and WD-40.

No, you 'failed' to see he DID mention that the Pentium 4 will most likely do good with SSE2 optimized software... however there is NON out now, and there probably won't be for many months.

Therefore, like he said... don't buy a P4 now. In the future, it could be a great processor. But right now, it's just NOT worth the money it costs.

<i>Intel was trying to establish a new standard in SSE2 with their P4 chip. </i>

So basicly Intel needs to create a new instruction set to compete? Can they not compete based on hardware alone? Althon had no attached instruction set....

<i>And that once SSE2 becomes a new standard, it will be incorporated into nearly all x86 software</i>

Really? Right now, only the P4 supports SSE2 (yes, I realize that AMD has licensed it); but it might be quite sometime before SSE2 becomes wide-spread. And if P4 can't stand on its own legs without SSE2, why buy it? Catch-22.

<i>So of course software optimized to run on an AMD chip and not on a Pentium4 chip is going to give better results to the AMD chip.</i>

Ok smartass, look at the 3D Studio Max 2 benchmarks, which measure the 80-bit precision x87 instructions (FPU instructions). The AMD processors rate @ 150.0, the P4 1.4 Ghz at 73.5, 1.5 Ghz at 80.0. Now even if SSE2 <b>doubles</b> the FPU power (like hell!), then the 1.5 Ghz will <b>just barely</b> outpreform the Athlons.

As a final thought, here are data I collected from Pricewatch.com today; the price range shown was the first 10 or so prices.

<i>
From Pricewatch.com, 22 Nov, 2000.

Athlon Thunderbird 1Ghz: $247 - $263
128MB SDRAM, 133MHZ: $ 47 - $ 54
Total: $294 - $317

Athlon Thunderbird 1.2Ghz: $487 - $569
256MB SDRAM, 133MHZ: $ 89 - $131
Total: $576 - $700

PIV 1.5Ghz: $1097 - $1198
64MB RDRAM PC800 x2: $ 230 - $ 346
Total: $1327 - $1544
</i>

So the P4 set is 4.5 - 5.0 times more expensive than a 1Ghz Athlon w/ compareable memory, or around 2.25 times more expensive than a 1.2Ghz Athlon with twice the memory!

Is it really worth it? Even if the P4 was 20% better than the 1.2Ghz Athlon in <b>every</b> test?

I've got a Pentium 225Mhz MMX and I think nor Office nor Windows are MMX optimized, just a bunch of games, hope somebody compiles things for SSE2.

See things that way: The FPU is the cake (always needed, you should make it as good as you can) and SSE-x is the icing ( It shouldn't be absolutely necessary, but you can put some on your piece of cake). What Intel is trying to do is making you eat ONLY icing!=)

Seoman- Newbie at last

key words on your post is
"his benchmarks will be completely unimportant once software is compiled under SSE2"
Probably true but by that time who knows what AMD has up it's sleves. Sure we have all seen their roadmap but how many times has it been changed in the past? I do not see a near future on having ALL software SSE2 optimized until perhaps 2002 or so... and by then HAMMER TIME !!
I not for AMD or for Intel i just post my views. And i honestly think this seems kindda like the old k62 bashing over at the old forums "Amd cant compete with the K62 based solely on 3dNOW!" same goes for Intel.

Why would Intel make the same mistake AMD did in the past is beyond me but it did and will suffer the consecuences for it.

AMD CURRENTLY in our PRESENT KNOWN DATE has the BEST procesor on the Mark PERIOD! and till i see otherwise ill keep on saying it.
When and if only when the P4 becomes a viable CPU ill be first in line to but it. But for now it is not...


Why do I use LINUX ? Cause its the BEST OS
Why do I use Windows? Cause its the BEST Nintendo..

Jeepers.. It looks like Tom might have stepped on some intel fans toes this time. ;-).. Well I'ld buy intel if intel was better, it isn't and probably won't in a year either.. but what do I know.. Still the P4 and SSE2 and all that crapola is like: Hey this is a brand new car.. nice ? well err I'm sorry but U only get the motor.. the rest hasn't been developed yet.. Geeze.. .. Intel has just made another mistake, the launch of an insanly expensive proc. and a lousy one too.. sad sad sad.. I have allways bought intels, but my newest pc which I got a year ago, just had to be a K7, nothing could compete.. Eventhough P3 was cheaper then.. now things have changed.. New thinking right now won't sell.. perhaps in a few years, but ppl won't buy a crappy cpu when there's no software to utilize it yet..

-Hasse

If I buy high end software I want to know that if I upgrade my hardware to a new processor it will be faster. These days hardware is cheap compared to software. You have to pay for upgrades on high end software. Our current software does everything we need CAD/CAM, Rendering, etc. Most of our packages are ports from other operating systems and chips with code that stresses the FPU. The authors will not spend time rewritting the code to support sse2 in the short term or even the long term, just so that it can run at expected speed on a new processor. The whole reason X86 is popular is that it has become a standard.
SSE1 has been out for ages, how much software is optimized for it. How much software is optimized for MMX. Not that much. They are marketing instruction sets. Like a spoiler on a CAR. It does not make the car go faster in straight line (in fact top speed will be lower)it just means you can go round a corner at a faster speed, but it looks good.
So give Top a break. At the moment the P4 does not perform as expected/hoped which in most peoples eyes matters.

Frankly, I thought his original post was more than optimistic about P4's future -- perhaps more than it deserves. I'm not a big expert in hardware, but I perfectly got the idea, from both the original review and an update, that SSE2 is coming and it will increase P4's performance. It says so numerous times, in all objectivity, so quit attacking the poor guy
Again and again, I'm not buying a chip now because there will be a better version of it next year -- the chip I buy in the next few months, I'm stuck with -- no matter how good the 2GHz P4 is, I will have the 1.5GHz one, and it is just not good enough (it seems).
As far as for SSE2, I can't believe you guys are seriously suggesting that we'll all suddenly be running all SSE2 applications next year. First of all, the next Windows won't come for a few years probably, so your OS to start with won't use SSE2 that soon (Linux maybe will, who knows...)
Second of all, MMX has been around for ages, and not every application (or even most of them) I use is MMX optimized. It is unrealistic to believe we'll suddenly pay to update all our software when it becomes SSE2 optimized. Even if SSE2 becomes widely adopted, it will be quite a few years before we pass the 50% mark. By that time, who KNOWS what other things these guys will come up with?
It's not like the world stopped after MMX... after all those guys running around in multi coloured space suits on TV ads in '96, trying to tell us how great the MMX revolution is, who today bothers to figure out whether their software is MMX optimized or not? And has it seriously cramped AMD's efforts? I don't quite think so...

Do you know much about SSE-2 and what it is designed for, how it can be optimised, how often it could be used?

SSE-2 is not a complete replacement for the FPU. It's job is to do lots of repetitive calculations on double precision numbers. It is almost impossible to use on everyday programs.

1) double precision numbers are not very often used unless they really have to be as they typically take more time to process than single precision numbers.

2) Repetitive calculations on doubles are found in scientific work and some media apps. However IEEE compliant MPEG codecs cannot use SSE as the way it performs the calculations is not IEEE compliant. This potentially cuts out a major chunk of the programs where it might otherwise be used. In any case I think many use single precision numbers and so SSE-2 cannot help.

3) Creating a compiler which can automatically optimise loops is very difficult to make. The SPECfp and SPECint scores for the P4 are from a highly agressive optimiser which designed to make the chips look as good as possible. Also the SPEC scores are for applications which are not common but are in fact high end commercial apps which would benefit from the P4 architecture.

4) Much more useful to the P4 will be simple recompilation to cut out as many of the problematic instructions it can (e.g. shifts) and possibly insert branch hints which can have a significant impact on performance.

Please don't rant when you don't really know what you are talking about all it does is make you look silly especially when much of what was said is untrue.

L

I could be wrong but I think MMX (Multi Media Xtensions?)were a series of extra instructions to allow improved handling of 3d graphics (or at least in the main) and so MMX optimising would not really have place in MS Office or Windows bare OS.

OM(F)G its acutally someone here in the PIV argument that actually understands what SSE2 is! Good man!

Also note that neither Tom's or others' reviews I've read have said that the P4 is that bad. They're stating that without SSE2 being around at the moment the P4 just doens't have a lot going for it at the moment...except Quake III, Quake III and definately Quake III..oh did I mention..

Anyways, they also state that they will be watching the PIV's progress. If things do look up for this new beast Tom's reviews just as any of the others reviews will reflect that fact.

SSE2 is coming soon to Tom's arsenal of tools of the benchmarking kind but will it save the PIV? We'll just have to wait..the longer it takes, the longer the PIV will suffer and that's no ones fault but Intel..I don't take pleasure in that fact, I bet Tom doesn't either because no one really wants to see Intel continue on its bungling path any further.

The only thing you can be sure of though is that PIV buyers had better like Quake III

I didn't see any bias in Toms review update...well except for maybe that anti PIV picture but I'm certain that he stared at the image for a while knowing that he'd get tons of Intel fans protesting and shouting about..I'm glad he thought 'f*ck it and hit that upload button with a smile'

See the p4 for what it is, a stepping stone. Not unlike the amd k6-3 line of processors, which never met with a great deal of success, but had alot to do with the development of the athalon. There is a certain argument over the backwatds compatability issue with the p4 and sse2. Sure the p4 might perform extremly well when sse2 is finally fully implemented but was it neccesary to cripple it in the process as far as non sse2 software is concerned? Maybe it was, it is hard to say. I have a question that I have not seen much press on, is dx8 sse2 optimized?

The question is was the PIII, ever supposed to go upto 1ghz?
When the P4 design was started the PIII was to go up to about 600-800Mhz (If I remember the roadmap correctly, it was a while ago). If you were to compare the P4 to this then we would all be raving about it. AMD were the fly in the ointment. Considerable resource was required to push the PIII upto 1Ghz and beyond to compete. The 840i issue may well have had an impact on the 850i design and the CPU design.
If the Athlon had not been developers may well have taken Intel advise on SSE2 to heart and may have started development with SSE2 in mind. IMO with the strong FPU of the Athlon and PIII developers focused on the FPU thinking this would be the future.
I may get a P4 in the future, but they will have to stablize on long term pin grid and manufacturing process first and have good SSE2 support for the applications I use. I learnt the hard way with the PIII and Athlon were both switched to socket shortly after I got the slot versions.

This my personal opinion, wonder anyone agree with me..
A hardware review should dig as much weakness of the product as he can, as the vendor itself would not tell you the weakness but quality of the product. Infact whenever i read a review i look for the weakness of the product, the less weakness of a product the better it is and vice versa.

As in Tom's review, he is trying to show us the weakness of PIV and warn us not to make a purchase decision which we would regret later. To me, i think Dr Pabst did a great
job especially the follow up review which he warn ppl what he overlook in the first review.

We need reviewer that have guts to speak out the truth, to let ppl know what is happening.

For the Intel Vs AMD case , we need AMD to be there. As we dont want to let 1 vendor to decide for us everything like the old days( Ram, Price, Speed etc). We want more choice and value for money. If PIV beats TBird in every benchmark everyone will accept it as TBird cost half the price. With the price of PIV it should be a superior product but it is not. So Dr Pabst should advice us not to buy a PIV now(I will be disappointed if he dont ). Keep it up !! Dr Pabst.

Sorry for my poor English and Long winded post.