Home

News

Forums

Hardware

CPUs

Mainboards

Video

Guides

CPU Prices

Memory Prices

Shop



Sharky Extreme :

Search

Latest News


- Outdoor Life: Panasonic Puts 3G Wireless Into Rugged Notebooks
- Averatec Launches Lightweight Turion 64 X2 Laptop
- Acer Fires Up Two New Ferrari Notebooks
- Belkin Debuts Docking Station for ExpressCard-Equipped Notebooks
- Logitech 5.1 Speaker System Puts Your Ears At Eye Level
News Archives

Features

- SharkyExtreme.com: Interview with ATI's Terry Makedon
- SharkyExtreme.com: Interview with Seagate's Joni Clark
- Half-Life 2 Review
- DOOM 3 Review
- Unreal Tournament 2004 Review

Buyer's Guides

- September High-end Gaming PC Buyer's Guide
- September Value Gaming PC Buyer's Guide
- October Extreme Gaming PC Buyer's Guide

HARDWARE

  • CPUs


  • Motherboards

    - Gigabyte GA-965P-DS3 Motherboard Review
    - DFI LANPARTY UT nF4 Ultra-D Motherboard Review

  • Video Cards

    - Gigabyte GeForce 7600 GT 256MB Review
    - ASUS EN7900GT TOP 256MB Review
    - ASUS EN7600GT Silent 256MB Review
    - Biostar GeForce 7900 GT 256MB Review





  • SharkyForums.Com - Print: Anandtech Review of Pentium 4 1.8GHz

    Anandtech Review of Pentium 4 1.8GHz
    By Conrad Song July 02, 2001, 10:53 AM

    A couple of comments:

    1. The Pentium 4 scales considerably better than the Athlon (and presumably Athlon MP) on almost all benchmarks.
    2. The AthlonMP's hardware prefetch adds considerable performance increase to games and 3d, but adds nothing to the other benchmarks. (Hardware prefetch requires a very regular data set in order to work).

    AMD will have a hard time bringing in Palominos above 1.5GHz, to compete with 2.0GHz plus Pentium 4 processors. Considering the lower performance scaling over Pentium 4, a 1.5GHz AthlonMP may not compete. However, rumors have it that a 1.5GHz AthlonMP is to be released soon, which may mean a 1.6GHz in Q4.

    By Doward July 02, 2001, 11:29 AM

    Hehe... I'm not sure if AMD will have as hard a time as you say. They got +50% out of ONE STEPPING (Axia) Just let them work their magic over there.

    By Ahem July 02, 2001, 11:45 AM

    There sure was a whole lot of spankage going on! As far as mucho better scaling, I have to agree, although seeing as how their 2Ghz had the 133fsb, it's not a fair and direct scaling, so unless Intel does move to a 133fsb, I wouldn't expect their 2Ghz to perform quite as well. As far as the MP goes, I sure did like what I saw in terms of games. I know they advertised up to a 15% gain, but, honestly, I'd be quite happy with a consistant 5-7% gain, that'd really seem like success to me.

    By Arcadian July 02, 2001, 11:52 AM

    Hi guys,

    It's been a while, but I just read Anand's article, and I thought I'd drop by to see what people are saying.

    It does appear that prices are moving back up with this release. It may indicate that current low prices are simply an effect of a slower economy, and prices on all items are destined to go up later this year as things recover. I wouldn't have thought that the 1.8GHz Pentium 4 would be back up over $500 this soon, but that may speak for Intel's confidence in this part.

    As Anand showed with his overclocked 2.0GHz part, the Pentium 4 continues to scale well. And only 2-3 months away, we'll no doubt be seeing things run even faster. I have a feeling that AMD will welcome Intel's higher prices, since that gives them a little more headroom to be competitive, and raise prices of their own. It's been pretty apparent that AMD is berely close to meeting margins with prices so low, so I'm sure that they are looking forward to seeing things the way they used to be.

    This may impact the consumer a little, which is why I'm sure there will be some complaints here, but I remember computers being affordable before all these price drops, so I am not that upset that things may be going back the way they are. I'd rather see AMD and Intel do better as companies, since they have both been hit by the economy.

    But that's it for my little rant. Hope to see you guys soon.

    Arcadian.

    By Un4given July 02, 2001, 11:54 AM

    The Athlon is still being partially crippled by limited FSB. JEDEC has laid out the guidelines for PC2600 DDR memory, and since the EV6 bus is good to 200MHz (400MHz DDR) it is possible for the Palomino core to close the gap with more bandwidth. However, this will require AMD and chipset manufacturers to work together to bring this to the table.

    By Moridin July 02, 2001, 01:19 PM

    Just though I would throw in a quick scorecard. Of course any scorecard of this type can be misleading since you often see similar benchmarks won by the same processor.

    By benchmark.

    P4 1.8 has 7 wins
    Athlon 1.4 5 wins
    Athlon 1.2 MP 1 win
    And the over clocked P4 finished on top in all but one benchmark.


    If you break it down by wins per page:

    The 1.8 P4 has 2 wins and a tie
    The 1.4 GHz Athlon has 1 win and 2 ties
    The 1.2 GHz Athlon MP has 1 tie

    If the OC'd 2 GHz P4 is thrown into the mix it wins 4 pages and ties one.

    So it seams like the 1.8 GHz P4 has a slight edge over the best AMD has to offer at the moment. Of course the Athlon is still a lot cheaper.


    One test that struck me as interesting was the CSA office bench. It was interesting because of the way performance dropped on the P4 as the stress loading went up. The second interesting thing was that this drop-off was much less pronounced in the over clocked P4.

    In fact in the final test the OC'd P4 was running 11% faster but finished 20 % faster. The obvious reason for this is the faster FSB, but if bandwidth were the key for this bench why would the Athlon outperform the P4? Further more this is one of the benchmarks where the Athlon MP was faster then the Regular TBird.

    The answer seems to be that this bench does a lot of Data Base work. It has Access component as well as MS Exchange (which is built around the MS "Jet" database just like MS Access is). Database work often has a large number of difficult to predict reads and branches.

    The question is which is coming into play here. Since the FSB OC on the P4 makes such a big difference I would say it is performing a lot of "random" reads and the processor is stalling while it waits for this data. We already know from Anands dual processor review that the P4 doesn't do particularly well in database work.

    The OC'd FSB has 30% lower latency then the regular FSB if the chipset is running in lockstep. This accounts for the performance jump with the FSB overclock. The Athlon MP has HW prefetch, which is able to predict some (but not all) of these random reads and therefore outperforms the regular Athlon.

    I am speculating that this is one benchmark on which the DDR chipset for the P4 will have a significant performance edge over the current i850 and i860 chipsets. Anyway, enough rambling for now.

    By JabberJaw July 02, 2001, 01:59 PM

    Considering you can purchase a 1.7Ghz P4 and a good mobo for less money than the 1.8Ghz CPU alone, the 1.8 seems quite overpriced. It just doesn't make sense to raise prices like that in current economic conditions, it's a disincentive for people to buy. What is Intel trying to pull? I think with this first glimmer of hope that their processors can once again compete successfully with AMD, Intel is trying to reverse the market trend back to higher prices. I'm sure Intel execs are still pining for those halcyon days of old, of obscene profit margins and gouging their customers. (Remembering $1K+ for a 1Ghz PIII, less than a year ago...) Meanwhile, AMD prices continue to fall. The Athlon 1.2 MP is starting to show up for under $200.

    By Moridin July 02, 2001, 02:25 PM

    quote:Originally posted by JabberJaw:
    Considering you can purchase a 1.7Ghz P4 and a good mobo for less money than the 1.8Ghz CPU alone, the 1.8 seems quite overpriced. It just doesn't make sense to raise prices like that in current economic conditions, it's a disincentive for people to buy. What is Intel trying to pull? I think with this first glimmer of hope that their processors can once again compete successfully with AMD, Intel is trying to reverse the market trend back to higher prices. I'm sure Intel execs are still pining for those halcyon days of old, of obscene profit margins and gouging their customers. (Remembering $1K+ for a 1Ghz PIII, less than a year ago...) Meanwhile, AMD prices continue to fall. The Athlon 1.2 MP is starting to show up for under $200.
    If Intel can sell all the 1.8 GHz processors it can produce there is reason for them to charge less. That will only happen if they are having difficulty selling the processors that they are producing.

    One other thing that does happen is that if the can sell a lot of 1.8 GHz at a high price they can change the rest into 1.7 or 1.6 GHz and sell them at a reduced price without having to reduce the price on all the1.8 GHz processors they are producing. (This it the situation that overclockers love. It means you can buy a processor that normally would have been a 1.8 GHz for the price of a 1.6 Ghz)

    By Whíte July 02, 2001, 03:20 PM

    Strange how a review of an Intel processor contained about as much, if not more information about AMD's processors. Anyone else notice this?

    Does the netburst architexture use any form of data prefetching? Cause the Athlon MP 1.2 wins over the Athlon 1.4 in Q3A and they say it is the Data prefetch that does it. now there's been so much whining about if Q3A uses SSE/SSE2 optimisations and i thinkthe conclusion was it didn't so that would mean the only reason why the P4 wins in Q3A over the regular athlon (this is neglecting the frequency difference) would be the data prefetch.

    In which case the nForce chipset should indeed increase performance significantly for the Athlon, at least in Quake 3.

    I'm thinking we'll be seeing the real performance gains in memory and motherboard upgrades for the next gen CPU's and not so much the processors themselves.

    By Bolweval July 02, 2001, 04:04 PM

    Ya they got 50% out of one stepping with IMO an UN-expectable amount of HEAT.


    quote:Originally posted by Doward:
    Hehe... I'm not sure if AMD will have as hard a time as you say. They got +50% out of ONE STEPPING (Axia) Just let them work their magic over there.

    By Moridin July 02, 2001, 04:34 PM

    quote:Originally posted by Whíte:

    Does the netburst architexture use any form of data prefetching? Cause the Athlon MP 1.2 wins over the Athlon 1.4 in Q3A and they say it is the Data prefetch that does it. now there's been so much whining about if Q3A uses SSE/SSE2 optimisations and i thinkthe conclusion was it didn't so that would mean the only reason why the P4 wins in Q3A over the regular athlon (this is neglecting the frequency difference) would be the data prefetch.


    The P4 does have hardware prefetch. According to its creator QIII uses NO SSE instructions whatsoever. It has been pointed out that the Open GL driver QIII uses may have extensive SSE optimizations, but I don't see how this is a QIII only optimization.

    By Mandorallen of Molynnr July 02, 2001, 04:42 PM

    I didn't get much from the review except that a quad pumped 133 bus P4 is going to rock, even in Willamette style.

    The only thing that disappointed me with that review is that they didn't overclock the Athlons. Would have been better to see a 1.6GHz Athlon vs a 2 GHz P4.

    By Doward July 02, 2001, 04:54 PM

    Actually, they DID overclock the Athlons. The only problem was, the one they wanted to (the MP) didn't take too well to it

    The Palamino is still a little too fresh for my tastes... I can't wait to see an Axia version of MP

    By Predator2 July 02, 2001, 05:04 PM

    quote:Originally posted by Mandorallen of Molynnr:
    I didn't get much from the review except that a quad pumped 133 bus P4 is going to rock, even in Willamette style.

    The only thing that disappointed me with that review is that they didn't overclock the Athlons. Would have been better to see a 1.6GHz Athlon vs a 2 GHz P4.

    I was just thinking that.

    To compare an unOCed Athlon to an OC Intel is a bit unfair not to speek of the 600MHz headroom it had.

    By Whíte July 02, 2001, 05:05 PM

    quote:Originally posted by Moridin:
    The P4 does have hardware prefetch. According to its creator QIII uses NO SSE instructions whatsoever. It has been pointed out that the Open GL driver QIII uses may have extensive SSE optimizations, but I don't see how this is a QIII only optimization.

    In that case shouldn't the P4 be doing much better than the Athlon in ALL OGL based programs, cause from other reviews it seems to be only Quake III that really gives the P4 an edge over the athlon. Im lead to believe Quake III makes good use of hardware prefetch and that this is what nVidia is basing their Speculated 30% performance gain on (which is what the nForce should bring according to them, since it uses Hardware prefetch) think i'm onto something here? or am I just rambling?

    By dullard July 02, 2001, 05:09 PM

    quote:Originally posted by Arcadian:
    It does appear that prices are moving back up with this release. It may indicate that current low prices are simply an effect of a slower economy, and prices on all items are destined to go up later this year as things recover. I wouldn't have thought that the 1.8GHz Pentium 4 would be back up over $500 this soon, but that may speak for Intel's confidence in this part.

    When they introduced the 1.7 GHZ, Intel simply dropped their highest price level and instead sold the 1.7 at the 2nd highest price level. Now that the 1.8 is here, it fits nicely into that vacated highest price level. Thus, I don't feel this necessarily means that future prices will rise. When the 2.0 and 1.9 GHz models come out in a few months, then this "Expensive" 1.8 should fit right where the 1.6 now resides, about $260. I think that $260 sounds great for the performance shown at anandtech.

    Anyone know if Intel has plans to move to a 133fsb? That small change gave such a boost.

    By Mandorallen of Molynnr July 02, 2001, 05:12 PM

    quote:Originally posted by Doward:
    Actually, they DID overclock the Athlons. The only problem was, the one they wanted to (the MP) didn't take too well to it


    Then why didn't they benchmark the OC'd CPUs? Doesn't really make sense. It's kinda old news that the Palomino doesn't OC well in it's current form.

    By dullard July 02, 2001, 05:21 PM

    quote:Originally posted by Mandorallen of Molynnr:
    Then why didn't they benchmark the OC'd CPUs? Doesn't really make sense. It's kinda old news that the Palomino doesn't OC well in it's current form.

    It makes perfect sense. The non-overclocked P4 was compared with the non-overclocked Athlons. Didn't you notice that the overclocked P4 was at the bottom of every graph? It was not ranked along with the others.

    edit: They saved time and money by combining two articles. The normal P4 to the nomal Athlon comparison was one article. The overclocking abilities of the 1.8 GHz was a second article. These were convienently shown together.

    By Mandorallen of Molynnr July 02, 2001, 05:24 PM

    quote:Originally posted by dullard:
    It makes perfect sense. The non-overclocked P4 was compared with the non-overclocked Athlons. Didn't you notice that the overclocked P4 was at the bottom of every graph? It was not ranked along with the others.


    It would have shown how well the Athlon continues to scale in higher MHz, and if AMD can release a processor on their current Thunderbird architecture above 1.4GHz, and how it will perform. That's why it doesn't make sense. Why show how well the P4 can scale and not the Athlon?

    By Moridin July 02, 2001, 05:25 PM

    quote:Originally posted by Whíte:
    In that case shouldn't the P4 be doing much better than the Athlon in ALL OGL based programs, cause from other reviews it seems to be only Quake III that really gives the P4 an edge over the athlon. Im lead to believe Quake III makes good use of hardware prefetch and that this is what nVidia is basing their Speculated 30% performance gain on (which is what the nForce should bring according to them, since it uses Hardware prefetch) think i'm onto something here? or am I just rambling?

    I agree, though I think that in general the P4 does prefer open GL games.

    The problem is that the PIII also excelled at QIII and it had no HW prefetch, so there is more to it then that. When it comes right down to it, it is very difficult to say processor a is faster then processor b or vise versa. All you can really say is that processor a, is faster then processor b at a given task.

    It is very common for the web to oversimplify things. For example the Athlon X87 FPU is superior to the PIII and P4 X87 FPU, so whenever the PIII is faster then the Athlon in a FPU intensive application you will here the cry of SSE. For the P4 the cry will be "bandwidth" of "SSE2".

    The truth is that while these things can be very important they are not the only things that count. We are talking about hugely complicated systems. There are close to 100 times as many distinct parts in a P4 are there were in an Apollo moon rocket. Just because SSE and bandwidth are important doesn't mean that any time a PIII or P4 beats an Athlon these are the reasons.

    If you have the time you should drop by the SPEC website at www.spec.org and check out the scores of some of the individual benchmarks is the SPEC suit. What you will find out is that even though you have 14 different FP intensive applications the performance of a given processor can very wildly. All processors are simply faster at some jobs then they are at others, and this doesn't necessarily come down to bandwidth or SSE.

    By dullard July 02, 2001, 05:39 PM

    quote:Originally posted by Mandorallen of Molynnr:
    It would have shown how well the Athlon continues to scale in higher MHz, and if AMD can release a processor on their current Thunderbird architecture above 1.4GHz, and how it will perform. That's why it doesn't make sense. Why show how well the P4 can scale and not the Athlon?

    The reason, is that they already wrote an Athlon overclocking article. Why present the same article twice? If they wanted to show the same information twice, then they would be better of on a separate overclocked Athlon vs. overclocked P4 article.

    Since they didn't compare an overclocked P4 to an non-overclocked Athlon in their rankings, so I don't see your problem.

    By KJY9 July 03, 2001, 09:50 AM

    Current pricing on Pricewatch shows the P IV 1.8 @$765.00 and the Athlon 1.4@$169.00.
    I'm not sure how many knowledgeable people are going to pay $596.00 more for a chip that performs marginally better at best??

    By Moridin July 03, 2001, 10:52 AM

    quote:Originally posted by KJY9:
    Current pricing on Pricewatch shows the P IV 1.8 @$765.00 and the Athlon 1.4@$169.00.
    I'm not sure how many knowledgeable people are going to pay $596.00 more for a chip that performs marginally better at best??

    Pricewatch figures taken within a week or two of a chips release are seldom accurate. Wait a few weeks and you will see the part drop down to a price that is more in line with what Intel is charging for it. (the low $500 range)

    By Mandorallen of Molynnr July 03, 2001, 11:43 AM

    quote:Originally posted by dullard:
    The reason, is that they already wrote an Athlon overclocking article. Why present the same article twice? If they wanted to show the same information twice, then they would be better of on a separate overclocked Athlon vs. overclocked P4 article.

    Since they didn't compare an overclocked P4 to an non-overclocked Athlon in their rankings, so I don't see your problem.

    Because it seemed so much better when Tom's article was done like this :

    http://www.tomshardware.com/cpu/00q4/001206/index.html

    And it'd give an indication of how AMD's future processors would perform against Intel's future processors.

    It's not much of a point but still..

    By Xcom_Cheetah July 03, 2001, 11:47 AM


    By Motoman July 03, 2001, 01:57 PM

    quote:Originally posted by Moridin:
    Just though I would throw in a quick scorecard. Of course any scorecard of this type can be misleading since you often see similar benchmarks won by the same processor.

    By benchmark.

    P4 1.8 has 7 wins
    Athlon 1.4 5 wins
    Athlon 1.2 MP 1 win
    And the over clocked P4 finished on top in all but one benchmark.



    ...do I need to keep pointing out that AMD trades punches with Intel while always running at significant clock speed disadvantage? Intel should be ashamed of themselves for producing CPUs that have to be compared with AMD chips of 300-500Mhz less to come out on top. Take a P4 1.4Ghz and compare THAT to the Athlon 1.4Ghz - then see what happens!

    By JabberJaw July 03, 2001, 03:42 PM

    quote:Originally posted by dullard:
    When the 2.0 and 1.9 GHz models come out in a few months, then this "Expensive" 1.8 should fit right where the 1.6 now resides, about $260. I think that $260 sounds great for the performance shown at anandtech.

    The 1.8Ghz P4 is "Expensive". Even if the street price for the 1.8 quickly stabilizes to around $500, do you really think that its' 100Mhz stepping is worth the 54% ($175) premium over the 1.7Ghz (at $325)?

    As far as your pricing predictions are concerned, unless you have inside info, you're just guessing. There is no natural law of neatly slotted "pricing levels" for cpu's, Intel is going to adjust their pricing as they see fit. As previously stated in another post, I dislike their attempt to trend their prices back to higher levels, but perhaps they are looking to the mass market and a predicted turnaround in business buying later this year.

    Also, your "few months" is too far into the future for anyone to try to predict how good a buy any CPU will be. They would need to be compared with their contemporaries at that time, in terms of both price and performance.

    By JabberJaw July 03, 2001, 03:44 PM

    quote:Originally posted by Moridin:
    Just though I would throw in a quick scorecard. Of course any scorecard of this type can be misleading since you often see similar benchmarks won by the same processor.

    By benchmark.

    P4 1.8 has 7 wins
    Athlon 1.4 5 wins
    Athlon 1.2 MP 1 win
    And the over clocked P4 finished on top in all but one benchmark.

    Although I do understand why Anand included the Athlon 1.2 MP in the article, IMHO it really doesn't belong in this comparison at all. It's misleading to directly compare its performance to the others, particularly in 'scorecard' fashion. See Anands other article, and how well the Athlom MP compares in its intended SMP configuration. Better yet, wait a couple weeks for the MPX chipset mobos and the 1.4MP to come to market, and then compare its price/performance to the competition.

    By dullard July 03, 2001, 04:07 PM

    quote:Originally posted by JabberJaw:
    As far as your pricing predictions are concerned, unless you have inside info, you're just guessing. There is no natural law of neatly slotted "pricing levels" for cpu's, Intel is going to adjust their pricing as they see fit.

    Yes, Intel can adjust their pricing as they see fit. However, Intel has only done that a few times in its entire history. About 95+% of the new processors have fit into Intels slotted "pricing levels". In fact, the pricing levels can be found right on Intels website. I see no reason to predict that Intel will STOP using its default price levels (unless you have inside information otherwise).

    By SSXeon July 03, 2001, 04:28 PM

    quote:Originally posted by dullard:
    When they introduced the 1.7 GHZ, Intel simply dropped their highest price level and instead sold the 1.7 at the 2nd highest price level. Now that the 1.8 is here, it fits nicely into that vacated highest price level. Thus, I don't feel this necessarily means that future prices will rise. When the 2.0 and 1.9 GHz models come out in a few months, then this "Expensive" 1.8 should fit right where the 1.6 now resides, about $260. I think that $260 sounds great for the performance shown at anandtech.

    Anyone know if Intel has plans to move to a 133fsb? That small change gave such a boost.

    This is very true:

    "Because of the processor's integrated heat spreader there is no danger of damaging the fragile CPU core underneath, which has left some Thunderbird/Duron owners with an AMD keychain instead of a processor on their hands"

    Thats a good point bout the P4's cooling ... ver good if you ask me .... and the 533FSB we will see when Rambus makes the PC-1066 and PC-1200 .... prob at end of the year .... northwood "might" have the 533fsb, But im not sure. Well i think i am getting the 1.8GHz P4 i dunno cuz my mom's helping me pay and ill let here buy the cpu lol.

    SSXeon

    By JabberJaw July 03, 2001, 04:50 PM

    quote:Originally posted by dullard:
    Yes, Intel can adjust their pricing as they see fit. However, Intel has only done that a few times in its entire history. About 95+% of the new processors have fit into Intels slotted "pricing levels". In fact, the pricing levels can be found right on Intels website. I see no reason to predict that Intel will STOP using its default price levels (unless you have inside information otherwise).

    Well, I have http://www.intel.com/intel/finance/pricelist/
    bookmarked, and have visited it weekly for several months. If this is the page you're referring to, these prices change periodically as reflected in the "% Decrease" column, and some CPUs of different clock frequencies share the same prices. Perhaps I'm misunderstanding what you mean by "slotted"?

    BTW, I am not using "quotes" to emphasize my points in your face, but rather to specify which terms of your statements that I'm addressing. No offense was meant. Same goes for whether you had inside info or not. I suspect that there are some here who do, so I saw it as a real possibility. (I don't have any inside pricing info)

    By Xcom_Cheetah July 03, 2001, 05:05 PM

    i m really surprised when ppl keep referring that the "intel should be ashamed that they have to compare their processor with 300 - 500 MHZ slower processors.." i think everybody knows that intel designed P4 in such a fashion that it can scale very well may be upto 10 GHz and for this they have to shed some performance... and secondly as far as the roadmaps says this gap will not close anywhere in the near future... so practically intel's 1.8 GHz is supposed to compete 1.4 GHz AMD's offering... and intel 2.2 GHz will be competing 1.7GHz ... may be AMD's next core will be even more scalable than we may say intel 5 GHz processor performaning equivalent to 4.5 GHz P4... so instead of looking at the speed grades its a lot more logical that we compare the best offerings from both the companies.. or the processor for the same segment of market..

    lastly i m not an Intel or AMD zealot..

    By Moridin July 03, 2001, 06:57 PM

    quote:Originally posted by Motoman:

    ...do I need to keep pointing out that AMD trades punches with Intel while always running at significant clock speed disadvantage? Intel should be ashamed of themselves for producing CPUs that have to be compared with AMD chips of 300-500Mhz less to come out on top. Take a P4 1.4Ghz and compare THAT to the Athlon 1.4Ghz - then see what happens!

    Or is it AMD who should be ashamed of themselves producing chips that run 400 MHz slower then the P4?

    The two metrics you should be looking at when comparing the quality of a design are overall performance and process in which that performance was achieved. (moving to a better production process can raise the performance of a chip by 50% or more)

    Current benchmarks show the P4 with a slight edge in overall performance, but I don't think there is any doubt that the .18 Cu process AMD is using for TBirds is better then the .18 Al process Intel is using to produce the P4. This will change later this year when Intel moves to a .13 Cu process.

    The P4 stomps all over any processor AMD has ever made in .18 Al. This includes Athlon classics and Durons, neither of which has reached 1 GHz in quantity.

    By Motoman July 03, 2001, 07:04 PM

    quote:Originally posted by Moridin:
    Or is it AMD who should be ashamed of themselves producing chips that run 400 MHz slower then the P4?

    The two metrics you should be looking at when comparing the quality of a design are overall performance and process in which that performance was achieved. (moving to a better production process can raise the performance of a chip by 50% or more)

    Current benchmarks show the P4 with a slight edge in overall performance, but I don't think there is any doubt that the .18 Cu process AMD is using for TBirds is better then the .18 Al process Intel is using to produce the P4. This will change later this year when Intel moves to a .13 Cu process.

    The P4 stomps all over any processor AMD has ever made in .18 Al. This includes Athlon classics and Durons, neither of which has reached 1 GHz in quantity.



    The way I see it, Intel is pandering to people who don't know any better than to just buy the "fastest" CPU out there - in other words, your average Joe Schmuck looking for a computer just sees a 1.8Ghz P4 and buys it, without even considering that it's performance may actually be slower than a 1.4Ghz Athlon, which has a superior design.

    If Intel could produce a chip with the efficiency of the Athlon Palomino, and then also cranck out their clock speeds, then I'd be impressed. But it has looked for quite some time that their only drive in life has been pure clockspeed, and the actual processing capability of the processor has become secondary. It just seems to me that things are the other way around at AMD, that's all.

    By Moridin July 03, 2001, 07:09 PM

    quote:Originally posted by JabberJaw:
    Although I do understand why Anand included the Athlon 1.2 MP in the article, IMHO it really doesn't belong in this comparison at all. It's misleading to directly compare its performance to the others, particularly in 'scorecard' fashion. See Anands other article, and how well the Athlom MP compares in its intended SMP configuration. Better yet, wait a couple weeks for the MPX chipset mobos and the 1.4MP to come to market, and then compare its price/performance to the competition.

    This is the only chip available from AMD's next generation Athlon core. It's single processor performance is just as important as its performance in DP systems. Would you rather Tom tried to hide the fact that AMD's "next generation" Athlon core is currently not competitive with the P4?

    By Bolweval July 03, 2001, 09:22 PM

    It's not Intel's fault AMD can't make there chips any faster, IMO AMD shouldn't have released the 1.4 or even the 1.2 with the incredible heat that comes with them. Intel would never release a chip that runs that ridiculously HOT. This O/C forum is polluted with posts about "Why is my T-Bird so hot???"

    I'm not trying to say that the T-Bird is slower, its not, it is a VERY fast CPU, I just question there decision to release the faster chips without a better cooling solution, or address the problem in some way?

    quote:Originally posted by Motoman:

    ...do I need to keep pointing out that AMD trades punches with Intel while always running at significant clock speed disadvantage? Intel should be ashamed of themselves for producing CPUs that have to be compared with AMD chips of 300-500Mhz less to come out on top. Take a P4 1.4Ghz and compare THAT to the Athlon 1.4Ghz - then see what happens!

    By Moridin July 03, 2001, 10:49 PM

    quote:Originally posted by Motoman:

    .

    If Intel could produce a chip with the efficiency of the Athlon Palomino, and then also cranck out their clock speeds, then I'd be impressed. But it has looked for quite some time that their only drive in life has been pure clockspeed, and the actual processing capability of the processor has become secondary. It just seems to me that things are the other way around at AMD, that's all.

    Designs that increase instructions per cycle inherently reduce the final clock speed. Your design method has been tried in the past. It almost always fails. Anybody remember the K5?

    By Conrad Song July 03, 2001, 11:59 PM

    quote:Originally posted by Motoman:

    If Intel could produce a chip with the efficiency of the Athlon Palomino, and then also cranck out their clock speeds, then I'd be impressed. But it has looked for quite some time that their only drive in life has been pure clockspeed, and the actual processing capability of the processor has become secondary. It just seems to me that things are the other way around at AMD, that's all.

    Let's examine this in closer detail:

    Forgetting the Athlon for the moment, let's look at the two major x86 designs on Intel's .18 Al process: Pentium III and Pentium 4. The Pentium III has a 10-stage pipeline (equivalent to the Athlon) and is clock-for-clock equal to Athlon (with the exception of the FPU), yet is limited to ~1.0GHz. The Pentium 4 on the other hand, has a 20-stage pipeline, but looks to hit 2.0GHz (2x clock faster), yet performs about 50-60% better (sometimes more, sometimes less) at 2.0GHz than the Pentium III at 1.0GHz... all on the same process.

    If you ask me, any design that gives that significant a performance increase due to architecture alone, earns my admiration. And the performance will only increase as more software and drivers are optimized.

    Add a little copper... add a little shrink... tweak a little more...

    The Athlon core is showing its age. The Athlon is not ramping as fast, nor is the performance scaling as well as the Pentium 4. I honestly don't think that an Athlon core can compete in a .13u process field, even with the benefit of SOI.

    We'll have to wait for 'Hammer'.

    BTW: Nice rig!

    By kritical July 04, 2001, 12:31 AM

    i got one thing to say. I will not spend 300+ more dollars on a processor only to get marginal performance increases. I would rather get something like an expensive video card or more ram..

    By Moridin July 04, 2001, 12:42 AM

    quote:Originally posted by kritical:
    i got one thing to say. I will not spend 300+ more dollars on a processor only to get marginal performance increases. I would rather get something like an expensive video card or more ram..

    Everybody has his or her own criteria for selecting what chip to buy. I agree with you, I wouldn't select a P4 if I was buying a processor today either. On the other hand you can't assume that the best choice for you is the best choice for everybody or even anybody else.

    On the other hand you can have a discussion on the chips based purely on the technology without ever bringing in such subjective opinions as which is the best chip for me. I usually prefer the later type of discussion since I am only rarely in the market to buy a chip, but I am always interested in technology.

    By Motoman July 04, 2001, 12:56 AM

    quote:Originally posted by Moridin:
    Designs that increase instructions per cycle inherently reduce the final clock speed. Your design method has been tried in the past. It almost always fails. Anybody remember the K5?


    Yeah, I know you - you can't have your cake and eat it too. At best, I think you can simply position Intel and AMD as taking two different roads to the same place - Intel excels by clock speed, and AMD by processing prowess.

    By Motoman July 04, 2001, 12:59 AM

    quote:Originally posted by Conrad Song:
    Let's examine this in closer detail:


    If you ask me, any design that gives that significant a performance increase due to architecture alone, earns my admiration. And the performance will only increase as more software and drivers are optimized.


    BTW: Nice rig!


    By all means - I agree - any design improvement that results in better performance is good. But it's kinda like putting a supercharger on a Ford Escort - OK, the thing makes 500 horsepower now, but it still handles like a slug.

    By Intel's Boy July 04, 2001, 01:02 AM

    quote:Originally posted by dullard:
    Yes, Intel can adjust their pricing as they see fit. However, Intel has only done that a few times in its entire history. About 95+% of the new processors have fit into Intels slotted "pricing levels". In fact, the pricing levels can be found right on Intels website. I see no reason to predict that Intel will STOP using its default price levels (unless you have inside information otherwise).

    Do you have bechmarks or your DUAL XEON P1.7???

    By SSXeon July 04, 2001, 02:21 AM

    quote:Originally posted by Motoman:

    The way I see it, Intel is pandering to people who don't know any better than to just buy the "fastest" CPU out there - in other words, your average Joe Schmuck looking for a computer just sees a 1.8Ghz P4 and buys it, without even considering that it's performance may actually be slower than a 1.4Ghz Athlon, which has a superior design.

    If Intel could produce a chip with the efficiency of the Athlon Palomino, and then also cranck out their clock speeds, then I'd be impressed. But it has looked for quite some time that their only drive in life has been pure clockspeed, and the actual processing capability of the processor has become secondary. It just seems to me that things are the other way around at AMD, that's all.

    Motoman now its showing that the P4 is beating the AMD w/ a Sledgehammer ... i mean come on ... lets go the the "benchmarks" like everyone used to tell me when the 1.3Ghz T-Bird is killing the 1.5Ghz P4 ... wait...

    hehe .... motoman your cool but just get it P4 1.8GHz is the fastest cpu out! Look at the "benchmarks" hehe.

    SSXeon

    By Moridin July 04, 2001, 10:14 AM

    quote:Originally posted by SSXeon:
    Motoman now its showing that the P4 is beating the AMD w/ a Sledgehammer ... i mean come on ... lets go the the "benchmarks" like everyone used to tell me when the 1.3Ghz T-Bird is killing the 1.5Ghz P4 ... wait...


    hehe .... motoman your cool but just get it P4 1.8GHz is the fastest cpu out! Look at the "benchmarks" hehe.

    SSXeon

    True but the margin is very small. There are also some area's in which the Athlon is clearly faster. This is just another round in the ongoing battle, it is not a knockout or even a knockdown. The Athlon on the other hand still has a decided advantage on price.

    BTW the Athlon MP is palomino, not Hammer (the K8)

    By JuveDuke July 04, 2001, 10:38 AM

    Intel is just upping the clock speed because as long as their cpus have a higher clock speed they seem better to the average consumer. (I mean think about it, Intels are built to not overheat with minimal cooling, because most people who buy their stuff wont know how to put a CAK38 on). Most people dont realize that clock speed isn't everything. Some computer idiot will buy a 1.7ghz and brag about how much higher his clock speed is than someone else with a 1.4ghz tibird. The t-bird owner will just quietly laugh to himself. Until the average consumer becomes smarter and more computer literate I dont think Intel will care about the "processing prowess" and continue with improving clock speeds to make their cpus seem better. Just my $.02. Dont take offense Intel fans, i still think they make good processors, just not for the price.

    By Moridin July 04, 2001, 11:25 AM

    quote:Originally posted by JuveDuke:
    Intel is just upping the clock speed because as long as their cpus have a higher clock speed they seem better to the average consumer. (I mean think about it, Intels are built to not overheat with minimal cooling, because most people who buy their stuff wont know how to put a CAK38 on). Most people dont realize that clock speed isn't everything. Some computer idiot will buy a 1.7ghz and brag about how much higher his clock speed is than someone else with a 1.4ghz tibird. The t-bird owner will just quietly laugh to himself. Until the average consumer becomes smarter and more computer literate I dont think Intel will care about the "processing prowess" and continue with improving clock speeds to make their cpus seem better. Just my $.02. Dont take offense Intel fans, i still think they make good processors, just not for the price.

    Read Conrad Song's post. He explains the issue nicely. You must remember that while clock speed isn't everything, it also isn't nothing.

    The decision to push clock speed in the P4 at the expense of IPC was a design decision aimed at performance. It is silly to think that a marketing department controls design decisions.

    By KJY9 July 04, 2001, 11:32 AM

    quote:Originally posted by JabberJaw:
    Although I do understand why Anand included the Athlon 1.2 MP in the article, IMHO it really doesn't belong in this comparison at all. It's misleading to directly compare its performance to the others, particularly in 'scorecard' fashion. See Anands other article, and how well the Athlom MP compares in its intended SMP configuration.

    Not sure, but I think they used the Athlon MP in the comparision to give us an idea how the Palomino will do, as the MP is a Palomino that's certified for MP boards.

    By JuveDuke July 04, 2001, 11:46 AM

    haha, i come with such crazy ideas. Everything is market driven i tell you

    By Motoman July 04, 2001, 01:18 PM

    quote:Originally posted by SSXeon:
    Motoman now its showing that the P4 is beating the AMD w/ a Sledgehammer ... i mean come on ... lets go the the "benchmarks" like everyone used to tell me when the 1.3Ghz T-Bird is killing the 1.5Ghz P4 ... wait...

    hehe .... motoman your cool but just get it P4 1.8GHz is the fastest cpu out! Look at the "benchmarks" hehe.

    SSXeon


    Nice one, SSXeon...pick out the benchmarks where the 1.8Ghz P4 actually managed to outrun the Athlon with a 400Mhz handicap. What about all the ones where the Athlon won, despite it's handicap?

    By Motoman July 04, 2001, 01:19 PM

    quote:Originally posted by Moridin:

    The decision to push clock speed in the P4 at the expense of IPC was a design decision aimed at performance. It is silly to think that a marketing department controls design decisions.



    ...you've obviously never worked in a high-tech sales-driven company. Ugh.

    By JuveDuke July 04, 2001, 01:24 PM

    haha nice point moto, he seems to have conspiculously (check spelling) left out the benches where intel gets raped. And who the hell games at 640X480?

    By KeeperMarius July 04, 2001, 01:26 PM

    I'm actually thinking on getting rid of my current AMD setup and purchasing a P4 rig.

    <<>>

    At this speed, I could honestly care less so much about performance. My P3-900 rig plays Counter-Strike, my main game, just as fast as my 1.2ghz T-Bird does. Diablo II receives awesome performance benefits from high performance memory (RDRAM), although my T-Bird doesn't flinch in that game to begin with anyways.

    So, why would I do it? Simple, the HEAT !! With 3 case fans and a modified 5000rpm fitted onto a ThermoEngine heatsink, I still have to keep the case open just so this (f)ucker doesn't overheat.

    It gets annoying. Really, it does, all the in name of reduced cost for a mass exodus of welcomed convenience.

    Is it really worth it?

    By JabberJaw July 04, 2001, 01:46 PM

    quote:Originally posted by Moridin:
    This is the only chip available from AMD's next generation Athlon core. It's single processor performance is just as important as its performance in DP systems.
    Yes, I've already stated that I understand why Anand included an Athlon MP in the review. For the above reason, he's eager to report on desktop Palomino performance. However, from the article:

    "Motherboard support for the Athlon MP is currently quite poor, mainly because it's only intended to work in 760MP motherboards so motherboard manufacturers have no real reason to worry about support on the desktop side of things."
    "VIA is another month away from releasing a newer revision of their KT266 chipset that is supposed to offer much higher performance, so we'd suggest that you wait until then before purchasing any KT266 based motherboards."
    "AMD only wants you using the Athlon MP in 760MP motherboards"

    Anand stuffed an Athlon MP into a premature desktop platform with a half-baked Via chipset, in an invalid configuration that's unsupported by AMD. How can you expect to get results that are anything more than rhetorical. Maybe some degree of insight into the Palomino's performance parameters can be intuited from this comparison, but in my opinion until the AMD-supported desktop Palomino systems actually hit the streets, it's too soon for a meaningful evaluation. Otherwise, these previewed results mislead people into making just the sort of scorecard comparison you've previously posted. The Palomino does not yet belong on that scorecard.

    quote:
    Would you rather Tom tried to hide the fact that AMD's "next generation" Athlon core is currently not competitive with the P4?

    I'm certain that's a rhetorical question intended to express irony, however my attitude is exactly opposite - I'm trying to maintain my objectivity despite the mentality of 'AMD Vs Intel' fandomonium that's so prevalent on web boards. I am impressed with the performance of an inexpensive dual Athlon MP system as compared to the pricier Xeon 4, and disappointed at the prospect that Intel seems to see fit to trend their CPU prices upward even when any performance advantages of their expensive platforms seem, at best, spotty and ethereal.

    By Moridin July 04, 2001, 02:31 PM

    quote:Originally posted by Motoman:

    ...you've obviously never worked in a high-tech sales-driven company. Ugh.

    You've clearly never looked at the design process a deep sub-micron devices with nearly 50 million transistor and a five + year design cycle goes through. There simply isn't any room in this process for marketing to say I want this or I want that. The basic fundamentals of this processor were probably set in stone years before the marketing people even heard about it existence.

    By Moridin July 04, 2001, 02:47 PM

    I see no particular reason to expect a large increase in Athlon MP performance anytime soon. It's biggest problem in its current form is it's inability to reach clock speeds higher then 1.2 GHz, otherwise performance is right about where it is expected to be.

    Whether or not it's performance will increase in future boards aside, this is a distinctly different product form the Tbird and it is on the market now. To exclude it just because it is immature and not performing well would be wrong. After all did anybody suggest the P4 shouldn't be included in benchmark reviews in the time after it was released? It was suffering from similar problems at the time, it was immature and not performing well.

    By Motoman July 04, 2001, 02:55 PM

    quote:Originally posted by Moridin:
    You've clearly never looked at the design process a deep sub-micron devices with nearly 50 million transistor and a five + year design cycle goes through. There simply isn't any room in this process for marketing to say I want this or I want that. The basic fundamentals of this processor were probably set in stone years before the marketing people even heard about it existence.


    I'm not saying that product marketing is having design input per se, but they can definitely declare that what the "market wants" is high clock speeds irregardless of processing efficiency, and then development has their marching orders. On the other hand, they could declare that processing efficiency is paramount, and then development goes after that...

    By Moridin July 04, 2001, 03:30 PM

    quote:Originally posted by Motoman:

    I'm not saying that product marketing is having design input per se, but they can definitely declare that what the "market wants" is high clock speeds irregardless of processing efficiency, and then development has their marching orders. On the other hand, they could declare that processing efficiency is paramount, and then development goes after that...

    What I am saying is that the process that this type of device goes through during its development make this all but impossible. The whole architecture frame work is laid out and every the performance of conceivable application is modeled over and over again to see what effect on performance any changes to the architecture will have.

    A company of Intel's size and resources knows with at least some degree of accuracy how well a new processor will perform before it even sets the location for a single transistor. Do you think they do all this work just to ignore it because it isn't what the marketing department wants then choose to go with a lower performing design instead?

    What you are suggesting implies the marketing department oversees this entire operation and blatantly disregards performance data in favor of something they hope will increase clock speed. By all means that are target areas and markets that went into the design procedure, but out of necessity these must be based on performance estimates, not clock speed.

    It doesn't address you concerns directly, but you should read this article anyway and tell me if it sounds to you like the P4 design team had its hands tied by the marketing department.
    http://www.eetimes.com/story/OEG20001101S0077

    By Bolweval July 04, 2001, 04:50 PM

    Good points! Same reason I won't get and AMD chip. There are without a doubt VERY fast chips, and VERY hot, and LOUD, and fragile, and the amount of money you spend on cooling almost offsets the money you save buying the chip. Not to mention that they draw twice the power of an Intel chip, Christ the money you save on your electricity bill over a year of using it will probly end up making it cost more than an Intel chip in the long run.


    quote:Originally posted by KeeperMarius:
    I'm actually thinking on getting rid of my current AMD setup and purchasing a P4 rig.

    <<>>

    At this speed, I could honestly care less so much about performance. My P3-900 rig plays Counter-Strike, my main game, just as fast as my 1.2ghz T-Bird does. Diablo II receives awesome performance benefits from high performance memory (RDRAM), although my T-Bird doesn't flinch in that game to begin with anyways.

    So, why would I do it? Simple, the [b] HEAT !! With 3 case fans and a modified 5000rpm fitted onto a ThermoEngine heatsink, I still have to keep the case open just so this (f)ucker doesn't overheat.

    It gets annoying. Really, it does, all the in name of reduced cost for a mass exodus of welcomed convenience.

    Is it really worth it?

    [/B]

    By gaffo July 04, 2001, 08:22 PM

    quote:Originally posted by Moridin:
    Designs that increase instructions per cycle inherently reduce the final clock speed. Your design method has been tried in the past. It almost always fails. Anybody remember the K5?

    What?? IPC dez. is "in the past" - and lousy IPC is what we ahve to look forward to from Intel?? - Something in the middle (as in all other things in life) is the most reasonable. The g-4 was too far into the IPC without regard to clock speed, the p-4 is too far the other way. The k-7 is in the middle. Enough said.

    By gaffo July 04, 2001, 08:31 PM

    quote:Originally posted by Conrad Song:
    Let's examine this in closer detail:

    Forgetting the Athlon for the moment, let's look at the two major x86 designs on Intel's .18 Al process: Pentium III and Pentium 4. The Pentium III has a 10-stage pipeline (equivalent to the Athlon) and is clock-for-clock equal to Athlon (with the exception of the FPU), yet is limited to ~1.0GHz. The Pentium 4 on the other hand, has a 20-stage pipeline, but looks to hit 2.0GHz (2x clock faster), yet performs about 50-60% better (sometimes more, sometimes less) at 2.0GHz than the Pentium III at 1.0GHz... all on the same process.

    If you ask me, any design that gives that significant a performance increase due to architecture alone, earns my admiration. And the performance will only increase as more software and drivers are optimized.

    Add a little copper... add a little shrink... tweak a little more...

    The Athlon core is showing its age. The Athlon is not ramping as fast, nor is the performance scaling as well as the Pentium 4. I honestly don't think that an Athlon core can compete in a .13u process field, even with the benefit of SOI.

    We'll have to wait for 'Hammer'.

    BTW: Nice rig!

    Respectfully - I dissagree 100-percent and see the inverse. The k-7 can compete in clock at the same die size. I little less clock, but still faster and so competetive. As long as AMD can come out with .13 microm within 6-months of Intels's release, the k-7 will continue to kick p-4's butt, and software will be optimized for p-3 (and maybe the k-7, and NOT the p-4!!) - depending upon marketshare. Just note that if software was optimized for the k-7, and not the p-3 as is current, the performance lead of the k-7 over the p-4 will increase by 20-percent WRT fpu. There are alot of k-7 used now, and maybe more folks aught to think less about p-4 optimized programs over the horizon and think a little about the k-7 optimied apps that may come out due to the athlons current marketshare - just my 2-cents.

    By gaffo July 04, 2001, 08:55 PM

    quote:Originally posted by Moridin:
    You've clearly never looked at the design process a deep sub-micron devices with nearly 50 million transistor and a five + year design cycle goes through. There simply isn't any room in this process for marketing to say I want this or I want that. The basic fundamentals of this processor were probably set in stone years before the marketing people even heard about it existence.

    Really - I don't know myself. Just when was the p-4 core dreamed up??. before the Merced fiasco, or after?. I remember the mid-90's and then Intel was working on a p-4 core, but cancalled it around 96?97? - did they see the Merced as the next p-4??, or not? The p6 core was (is) dragged out of years, I think originally because intel saw the Merced as the next core. But with that dissaster of timeliness, they had to hack a fourth x86 core. Today's p-4 is not the origninal p-4 that was cancelled in the mid-90's. that core had an x87 fpu, the current is a joke - I personnaly see it as a cheap hack to promote a temporay SSE-2 fpu until the intended Merced core arrives on the desktop. Intels original x87 core was castrated (out of inability to compete?, or just marketing - how knows) and SSE-2 centric rushed p-4 chip (to fill the p-3/Meced/Itanic void) was pushed on the consumer. End of rant.

    P.S. I know that intel had a p-4 generation core that was worked on and then cancelled way bweck in the mid-90's, and that the curent p-4 core is NOT that original core, and that this core has a weaker than intended x87 core. BUT the intended p-4 x87 (i.e. the current p-4) was a little better than the one in the p-4 chips shipped ----- BUT this is still not the original p7 core deZ..... Alot has happened behind Intel doors and at least three p-4 generation cores have been worked on. I see most Intel moves as damage control myself - due to Merced fiasco and AMD's competance.

    By Moridin July 05, 2001, 12:24 AM

    quote:Originally posted by gaffo:
    What?? IPC dez. is "in the past" - and lousy IPC is what we ahve to look forward to from Intel?? - Something in the middle (as in all other things in life) is the most reasonable. The g-4 was too far into the IPC without regard to clock speed, the p-4 is too far the other way. The k-7 is in the middle. Enough said.


    The G4 is an entirely different ISA and doesn't really apply. If you to make the comparison though the Athlon is a much wider chip then the G4. In fact the G4 is neither a wide (high IPC design) nor a fast (long pipeline) design. It is however a very elegant design that makes the best possible use of the resources it has.

    Back to the central issue. What makes a wide (high IPC) design in one ISA will not constitute a high IPC design for another ISA. The X86 ISA in particular has limited amounts of parallelism that can be extracted. In other words it favors high speed processors over high IPC processors.

    By Moridin July 05, 2001, 12:30 AM

    quote:Originally posted by gaffo:
    As long as AMD can come out with .13 microm within 6-months of Intels's release, the k-7 will continue to kick p-4's

    The last I heard was that AMD won't be moving to .13 until late 2002. You should also realize that AMD is already using Cu and Intel is not. (Cu gave the Athlon a 400 MHz performance boost.) So AMD will not benefit as much as Intel from moving to .13 since Intel will start using Cu ate the same time.

    The Athlon is not kicking the P4's but or even beating it on most applications right now, and we are not even seeing full P4 optimizations yet.

    By Moridin July 05, 2001, 12:38 AM

    quote:Originally posted by gaffo:
    Really - I don't know myself. Just when was the p-4 core dreamed up??. before the Merced fiasco, or after?. I remember the mid-90's and then Intel was working on a p-4 core, but cancalled it around 96?97? - did they see the Merced as the next p-4??, or not? The p6 core was (is) dragged out of years, I think originally because intel saw the Merced as the next core. But with that dissaster of timeliness, they had to hack a fourth x86 core. Today's p-4 is not the origninal p-4 that was cancelled in the mid-90's. that core had an x87 fpu, the current is a joke - I personnaly see it as a cheap hack to promote a temporay SSE-2 fpu until the intended Merced core arrives on the desktop. Intels original x87 core was castrated (out of inability to compete?, or just marketing - how knows) and SSE-2 centric rushed p-4 chip (to fill the p-3/Meced/Itanic void) was pushed on the consumer. End of rant.

    P.S. I know that intel had a p-4 generation core that was worked on and then cancelled way bweck in the mid-90's, and that the curent p-4 core is NOT that original core, and that this core has a weaker than intended x87 core. BUT the intended p-4 x87 (i.e. the current p-4) was a little better than the one in the p-4 chips shipped ----- BUT this is still not the original p7 core deZ..... Alot has happened behind Intel doors and at least three p-4 generation cores have been worked on. I see most Intel moves as damage control myself - due to Merced fiasco and AMD's competance.

    P4 development started in early 1996 when the PPro team finished that project and moved on to the P4 project. Some of the architectural work had already been done before this though. The P4 was never at any time canceled or replaced with a different version.

    The only core design Intel has killed in recent memory was the one they were working on pre IA-64. The design team from the Pentium classic was working on a core for a short time (about a year) when Intel and HP decided to combine their design efforts and produce Itanium/Merced (same chip different name) The Pentium team was taken off the project they were working on and moved over to IA-64.

    IIRC this was about 1993. (Itanium was late as far back as 1998)


    By jagojago12 July 05, 2001, 03:45 AM

    quote:Originally posted by Moridin:
    The decision to push clock speed in the P4 at the expense of IPC was a design decision aimed at performance.

    Doesn't it seem a little contradictory to think inefficient IPC, yet high clock speed will yield in more performance? Let's think hypothetically, since you intel fans love the scability of the Pentium4, if the Via C3 processor were to suddenly have developed a huge 40-stage pipeline of upto 7.8GHz, with initial laucnh chips at 2.3GHz, 2.5GHz and 2.8GHz would you want to buy it? Probably not, most of you know a C3 processor at 800MHz is about as fast as a Celeron at 650MHz. Same situation many of us are trying to get across. I do agree, that clock speed with the 20-stage pipeline was probably the best choice intel had (probably because intel had no CPU that HAD high IPC designed).

    At the current rate Intel is ramping up clock speed we'll see them hit the barrier for the 20-stage pipeline faster than the 10-stage pipeline the Pentium 1-3 had. At least the P3 was worthy of its MHz rating.

    Seems a little ridiculous that the Pentium4 needs a whopping 533MHz FSB to actually become a worthwhile CPU. I think the quote "Less is more" should start some controversy with my post.

    All in all I think more MHz without the same level of work done is a cheap and shallow way of marketing. In my eyes any Pentium4 processor is at least 200MHz slower than it's rated performance. I believe AMD/Cyrix CPU's used to have PR's (Performance Ratings) of their equivalent performance. Perhaps Intel should market the Intel Pentium IV 1.5GHz (PR 1.3GHz).

    rant off. sorry for the long post.


    Contact Us | www.SharkyForums.com

    Copyright © 1999, 2000 internet.com Corporation. All Rights Reserved.


    Ultimate Bulletin Board 5.46

    comment on SharkyForums.com
    previous page
    home




    HardwareCentral
    Compare products, prices, and stores at Hardware Central!


    Copyright © 2002 INT Media Group, Incorporated. All Rights Reserved. About INT Media Group | Press Releases | Privacy Policy | Career Opportunities