P4's future is looking GOOD to me.

[kami]

I'm more interested in that 0.13 micron version of the P3. P4 doesn't look too promising to me.

I find it funny when people think AMD is suddenly just going to take over the processor world...when Degenerate just made a point on how Intel can go around spending way more than AMD makes in a year.

AMD is good...competition is good...but be realistic please.

 

[Sunner]

Ok, Im just nitpicking here, but the bandwidth reduction wont be 300%, since you cant reduce something by more than 100% unless you wanna end up with negative numbers

The reduction will be ~67%.

But good post anyway

 

[MGMorden]

Polar: Hate to burst your bubble, but P4 is still x86 archetecture. P4, P3, Athlon, Cyrix III, whatever, just go about a different means of implementing x86. Hence, there's no reason for Athlon to dry up soon. The core is actually running faster clock for clock and can be pushed as far as 2ghz (if P3 wouldn't have topped out at ~1ghz intel wouldn't have even need p4).

Now Macs aren't x86. Notice how they run totally different software. Different archetecture. SPARCS not x86. Diferrent software, different archetecture. DEC Alphas. Different archetecture. Understand?

 

[sandorski]

If the economy remains sluggish, the P4 may have a hard time succeeding. On the flipside, a slow economy may help AMD enormously. The reason for this is that people will/may become more particular about their purchases. When/if they compare the P4 and Athlon, they'll discover that the P4 doesn't make much sense. AMD has established themselves technologically, this year they may break out and gain much more significant marketshare.

On a related note, read the articlehere that suggests that some mobo makers are in tough shape right now. Of particular interest is the comment from one mobo maker about P4 mobos selling very poorly, Socket370's selling poorly, but there is a bright spot with SocketA mobos. Not really conclusive of anything, but if that is a trend, we will see AMD become much more enthusiastically supported by mobo makers and that could alter P4s future success.

About Intel's expenditures: Certainly, Intel will spend more this year on upgrades and R&D, but much of that expenditure will be going towards the non-processor side of their operations. Intel is quite diversified now and they are diversifying more, that's good for their bottom line, but may divert needed attention from their core processor business.

 

[NOX]

Oh?no not this again! You know, you Intel and AMD people need to stop speculating so much about what?s going to happen in the future. Maybe it makes for good discussions, but I just hate it when I see so much misinformation in a thread like this one.

<< First let me address RAMBUS memory. Serial memory while having lower pin count has much much higher latency (bad bad bad). >>

Please dude, you have been disproved with that statement more then once. You keep saying it?s bad, bad, bad, yet you fail to prove how. More so I don?t know how you could prove it when the fact remains SDR, DDR, and RDRAM?s latencies are in the same neighborhood (50-70).

<< As an aside, since MS, Intel, Micron, Infineon (Seimens), and many other large corporations are aligned against RAMBUS, they are assured a loss in their legal battles over DDR patents. >>

One, where did you pull this one out from? You must be joking right?! Assured a loss, LOL don?t make a fool out of yourself buddy! If that was so their patents on SDR/DDR would have been debunk already, and none of those memory manufacturers would have signed any royalty agreements.

<< Pin count is an issue, but not that much of one >>

Well let me tell you right now buddy, Bert McComas, from Inquest Inc. disagrees with you (read his review over at simmtester.com ?DDR vs. Dual Channel RDRAM?)! I?ve talked to him before and he has admitted to me the issue of pin count does matter, unlike how you are trying to down play it. He also suggested it will be more of an issue for DDR/II. Seeing that DDR-II will have a pin count of 232!

<< Back to reality. The consumer may well not know or care that the P4 1.8Ghz he is buying is not even fit to compete with a 1Ghz PIII at less than half the cost, so as far as purchasing goes, it may do alright if the media fanfare is sufficient (which I am sure Intel will see to). >>

The reality is the consumer can buy a Dell 8100 w/P4 1.3GHz w/128mb RIMM for $1880. Now they can also get an identical system from Gateway using an AMD 1.2GHz w/128mb PC100 for $1999 non deluxe. The moral of that little story is that the consumer will most likely buy the P4 because it has Intel and Dell printed on it, and because it?s a tad bit cheaper then Gateway?s highest AMD system. Oh but wait, I though all AMD based systems were suppose to be cheaper then Intel?s P4! Sorry to burst your bubble, but that?s not entirely true!

I?m not going to suggest Intel?s P4 is the greatest, not like some of you who tend to attack another companies downside but never mention its upside. Like how RDRAM is outperforming DDR by more then double with the i850! But how the Athlon is making the P4?s Integer look powerless! But as any informed tech would know that does not really matter.

<< Intel would have been better off evolving the PIII core for the short run while they re-designed the P4 to better cope with todays market. I don't think Intel had any idea that P4 would be facing such stiff competition by the time it was released or that RDRAM would simply be rejected in the market. >>

Dude, that statement couldn?t have been more wrong. Intel is still marketing its PIII line at full speed! The P4 was brought to the market to address the deal they had with Rumbus! Fully supporting RIMMS to it?s potential, and they did a great job at accomplishing that seeing how RIMMs perform on the i850! RDRAM is not even been rejected by the market, what world do you live in buddy!? RDRAM is running its way into a Market wide open for competition, and has done a nice job according to Rambus annual financial reports, and many analysis on Wall Street. Look at the latest poll at simmtester.com, RDRAM is not being rejected!

Dulanic,

<< DDRII will be serial >>

DDR-II will not be serial, it will be parallel.

 

[Midnight Rambler]

OneEng, you conveniently &quot;forgot&quot;:

a.) Intel has already came up with a 20% reduction in P4 cost (packaging/assembly cost savings).
b.) Intel will have 0.13 P4's much sooner than you seem to think




<< AMD's business model makes them hugely profitable at $100.00 ASP. Intel needs $200.00 to pull the same margin. >>

You need to take a serious look at both AMD's and Intel's numbers again. With all due respect, you couldn't be more wrong, esp. about AMD. Their margins are horrendous.

 

[viper007]

sandorski:

That proves to me that Intel's main sales are comming larger OEM (Compaq, IBM) companies which use there own mainboard's rather then other brands. Intresting to know although!

Cheers...

 

[MadRat]

How many P4's are being sold in the server market? As far as I know, few if any at all. The P4 w/RAMBUS is not a design I'd think would lend itself to a good server performance.

 

[NOX]

C-Mon MadRat, that?s like me asking how many Athlons are being used in the server market? None as far as I know, even after 2 years of its entry into the world, it still has not been accepted into this area. You know why? Because it was not developed to do so! So why are you asking How many P4's are being sold in the server market? I would hope none, why would Intel put it?s P4 up against it?s own Xeon processor, and Alpha offerings for that matter? These two processors are very dominant in this area, that?s what they were made for, not the P4, not the Athlon. The P4 is so new to the market, and it was not developed for the server side of things.

That?s a below the belt questions there bud.

 

[Soccerman]

Please dude, you have been disproved with that statement more then once. You keep saying it?s bad, bad, bad, yet you fail to prove how. More so I don?t know how you could prove it when the fact remains SDR, DDR, and RDRAM?s latencies are in the same neighborhood (50-70).

while I'm not comparing SDRAM to RDRAM, SDRAM IS better, latency wise.

and yes, latency matters.

what are you going to do, when the 2ghz processors come out? each tick of the CPU clock is then about 500 picoseconds (.5 nanoseconds) in length, compared to 50-70 nanoseconds.

that means that, with higher clock speed, the CPU takes a greater and greater hit when waiting for slow memory.

hmm, I guess lower latency isn't such a bad thing now is it? it's like comparing hard drives today with SDRAM (well not quite as drastic, that's comparing milliseconds to I think nanoseconds), but you get the idea, if we can eliminate the RAM bottleneck altogether, we'd see a major improvement.

Well let me tell you right now buddy, Bert McComas, from Inquest Inc. disagrees with you (read his review over at simmtester.com ?DDR vs. Dual Channel RDRAM?)! I?ve talked to him before and he has admitted to me the issue of pin count does matter, unlike how you are trying to down play it. He also suggested it will be more of an issue for DDR/II. Seeing that DDR-II will have a pin count of 232!

ok, where do you get that spec of 232 pins for DDR-2?

btw, DDR SDRAM currently runs on a 64 bit bus, doesn't that mean that there are 64 wires going between the RAM and the northbridge?

and yes I agree, lower pin count = better, but higher clock speeds aren't a solution to the limits of a low pin count, if it means reduction of yeilds, and increase in power and heat.

a.) Intel has already came up with a 20% reduction in P4 cost (packaging/assembly cost savings).
b.) Intel will have 0.13 P4's much sooner than you seem to think

wow that's got to be hard, considering the P4 @ 1.4 ghz sells for about $1700 Canadian here.

 

[NOX]

Soccerman,

Please, read this article. I have nothing to hide about how I feel towards both technologies. I have been an advocate for RDRAM, as well as SDR/DDR in the past and present. I feel and believe both technologies have a place in today?s market. If you feel I?ve only advocated for RDRAM, well, that?s because everyone seems to attack RDRAM only, expose its bad sides but never willing to mention its good sides.

You?ll find this article very factual, and has no biased towards DDR or RDRAM. It points out a number of bad things about RDRAM, and includes great explanations as to why it?s bad, instead of just saying it?s bad, bad, bad, and with no explanation. He also points out a number of good things about RDRAM, and also points out a few bad things about DDR, and future Ram using SDR technology.

Ars

<< ok, where do you get that spec of 232 pins for DDR-2? >>

You can get that from JEDECs website.

<< wow that's got to be hard, considering the P4 @ 1.4 ghz sells for about $1700 Canadian here. >>

Well I can get one here in the US for $400, so what does that tell me about Canada? Stuff cost more there then in the US. The article I pointed to adds a great explanation as to why systems using RDRAM are more expensive, but that?s not necessarily the case (as the article also suggest) looking at Dells prices.

Samsung

So, still in the same ballpark! If you think these issues will only be issues for RDRAM, it won?t be. DDRAM will face these same challenges!

[EDIT]

Fixed link to JEDEC site.

 

[Sunner]

<< but that?s not necessarily the case (as the article also suggest) looking at Dells prices. >>


Dell's prices are redicilously low on P4 systems, Intel is probabaly supporting those systems financially to push the P4 into the market.
Not that it matters to the customer though

 

[theplanb]

Say, if we have faster enough disk storage device, would we still need the main memory and caches on the CPU? That would reduce the PC cost significantly. not to mention increasing the reliability. Umm.. too far fetched?

 

[Gunbuster]

&quot;faster enough disk storage device&quot; call me when your hdd is at least 400x faster then it is now

 

[Sunner]

Too far fetched.
There's no possible way mechanical storage(HD's, tape, CD's etc) will ever reach the latencies of electronical devices such as SRAM's, SDRAM's, RDRAM's etc.
Just compare the fastest HD's out there, a Cheetah X15 has a speced access time of 3.9 ms, which turns into something along the lines of 8-9 ms in the end.
good PC-133 SDRAM's have access times of 7 or 7.5 ns, and in the end, after the signal has been processed all the way, it will be up to around 50-60 ns or something along those lines.

Make it 1 ms VS 100 ns which is an extreme bias towards the HD, and the SDRAM would still be 10 000 time faster.

The of course there's the issue of RAW transfer rates, that same X15 can pull something like 40-45 MB/Sec while dual channel RDRAM can do well above 1 GB/Sec(I dont remember exactly, there are tests on Aceshardware for the picky).

 

[KarlHungus]

The reality is the consumer can buy a Dell 8100 w/P4 1.3GHz w/128mb RIMM for $1880. Now they can also get an identical system from Gateway using an AMD 1.2GHz w/128mb PC100 for $1999 non deluxe. The moral of that little story is that the consumer will most likely buy the P4 because it has Intel and Dell printed on it, and because it?s a tad bit cheaper then Gateway?s highest AMD system. Oh but wait, I though all AMD based systems were suppose to be cheaper then Intel?s P4! Sorry to burst your bubble, but that?s not entirely true!

Identically configured system? Perhaps you should look again. The Gateway you referenced has a Geforce2 Ultra, the Dell has a TNT2 M64. Not to mention the Dell had a 17&quot; montor to the Gateway's 19&quot;. Configured as identically as I could get them the Dell came in at $2299 vs. $2128 for the Gateway.

The other question you should ask is why are you comparing a 1.2GHz Athlon with a 1.3GHz P4? You should really compare a 1.2GHz Athlon to a 1.5GHz P4, or a 900MHz/1GHz Athlon to a 1.3 GHz P4.

Now, I've got nothing against Intel, but as it stands right now I consider the P4 to be a waste of money.

 

[NOX]

<< Configured as identically as I could get them the Dell came in at $2299 vs. $2128 for the Gateway. >>

Thanks for confirming the fact you can get a P4 system for roughly the same price as an Athlon system.

Though go back and look again at Dells site, now they offer a Limited Edition P4 system, starting at $1539. When configured identical to the Gateway system $1959. This include a 19? monitor, 32MB ATI Radeon, 40GB HD, DVD-ROM SBLive etc.

<< The other question you should ask is why are you comparing a 1.2GHz Athlon with a 1.3GHz P4? You should really compare a 1.2GHz Athlon to a 1.5GHz P4, or a 900MHz/1GHz Athlon to a 1.3 GHz P4.

Now, I've got nothing against Intel, but as it stands right now I consider the P4 to be a waste of money. >>

You?re a badly informed person, or you simply don?t know what you?re talking about. You compare the 1.2 Athlon to the 1.3 P4 because it?s the prudent thing to do. If you look at Toms P4 vs. Athlon final recount the 1.4, and 1.5 P4 outperformed Tom?s o/c Athlon at 1.4, 133MHz FSB DDR system in Q3A. Same for some areas of BAPCo.

In business-to-business, Athlon is still way behind Pentium 4. Even the overclocked Athlon at 1466 MHz cannot reach Pentium 4 scores. ? Tom?s

Now why should anyone compare the 1.2 Athlon to a 1.5 P4 when the 1.4 Athlon with DDR could not outperform a P4 at the same speed? I?m not suggest the P4 is there for better, I?m just saying the P4 is more of a contender then most of you people give it credit for!

I believe Tom said it best here ?Pentium 4 has got clear flaws, but it's not a piece of crap either.? Clearly it does, I don?t believe anyone is denying that, but it does have its advantages, which most of you here don't want to hear.

 

[TravisBickle]

I could tell that NOX was spreading misinformation on those systems anyway, but good work.

P4 only has memory performance going for it, in a few specialist applications.
rambus looks like it has a big problem with 400mhz chips to me. the higher the mhz, the poorer the yields and the more expensive. most graphics chips aren't even 200mhz. so what rambus requires is several 400mhz chips on one rimm alone! it's not going to get cheap soon.

 

[Train]

what are you talking about, I just got a 256MB stick of Rambus (PC800) for $250, thats the same price i payed for Muskin PC133 4 months earlier!

 

[NOX]

TravisBickle,

<< P4 only has memory performance going for it, in a few specialist applications.
rambus looks like it has a big problem with 400mhz chips to me. the higher the mhz, the poorer the yields and the more expensive. most graphics chips aren't even 200mhz. so what rambus requires is several 400mhz chips on one rimm alone! it's not going to get cheap soon. >>

What are you talking about, RDRAM is already running at 400MHz, and its already cheaper. Where have you been the past few months?

<< I could tell that NOX was spreading misinformation on those systems anyway, but good work. >>

Oh?yes, I have been doing a lot of that! That is why I link to sites with factual information right? Go check the prices for yourself before you accuse me of spreading misinformation!

 

[TravisBickle]

oh dear we have another flamewarrior who takes himself real seriously. don't worry about it NOX.

 

[KarlHungus]

<< Though go back and look again at Dells site, now they offer a Limited Edition P4 system, starting at $1539. When configured identical to the Gateway system $1959. This include a 19? monitor, 32MB ATI Radeon, 40GB HD, DVD-ROM SBLive etc. >>

Perhaps we have different ideas of what identical means. When I say identical it means that both computers have a Geforce2 Ultra in them. Yes, the P4 can be outfitted with a Geforce2 Ultra on Dell's site. If you want to compare that particular P4 to an &quot;identical&quot; Gateway I would suggest you find a Gateway with a 32MB Radeon.

<< You?re a badly informed person, or you simply don?t know what you?re talking about. You compare the 1.2 Athlon to the 1.3 P4 because it?s the prudent thing to do. If you look at Toms P4 vs. Athlon final recount the 1.4, and 1.5 P4 outperformed Tom?s o/c Athlon at 1.4, 133MHz FSB DDR system in Q3A. Same for some areas of BAPCo. >>

Now I'm badly informed? Perhaps you would like to explain your logic here. I assume you think you laid the smack down on me by quoting some stats from Tom's P4 review. Perhaps you should refresh yourself by reading Anand's review again. The P4 1.5GHz gets beat by the Athlon 1.2GHz in every game but Q3. Anyway, this point is moot considering these tests are fillrate limited at the resolutions everyone actually plays at. I won't even bring up the results in the business suite tests since you obviously don't want to acknowledge them. All in all the P4 performs well in Q3, some streaming applications, and certain theoretical tests. In most of the others it gets stomped by the Athlon or the PIII.

As for the reason to compare the P4 1.5 to the Athlon 1.2. They have vaguely the same performance level (i.e. they both beat the PIII 1GHz for the most parts at least), and are the highest clock speed/price available. The rationale behind comparing the P4 1.3 and Athlon 900/1G is just as simple, they are both the lowest rated speed before dropping to the next level down in processors (PIII and Duron respectively).

Yes, the P4 has some good points, but I feel they can be mostly attributed to the bus speed and <shudder> Rambus. If Intel had made this good of a chipset back in the days of the i820 I doubt there would have been very many people who would not have accepted Rambus.

 

[Cknyc]

Im kinda looking forward to AMD's 64 bit solution. That should be a kick a$$ processor, coupled with either ddr or rdram.

 

[steelthorn]

I have to agree. I believe the p4 is a killer chip, and I don't think AMD will be able to do anything against it. Sorry Amd but your dead in the water.

 

[osgood111]

Read this page How Intel Blew It. :Q