Prescott 4GHz cancelled

I sent this to a bunch of my co-workers, many of them who continuously re-affirm their opinion that AMD is garbage. While there is no question Intel never fails on stability, I take a different approach to this news:
Some people are Monday Morning Quarterbacks. Me, I'm a Friday Freelance CPU Analyst. Today's topic: The Internal Memo from Intel attached above.

I found this information very telling, especially since the company has now confirmed the text in public. The bottom line is that Intel is cancelling the 4GHz P4 to focus on other "performance enhancing" technologies. The reasoning behind the actual ending of the P4's continuous GHz increase is obviously technical; yields with the .09micron process have been notoriously bad and the power requirements/heat output for the Prescott core have been horrific in comparison to AMD's current chips. They could have put out a low-yield P4-4GHz core for a little while, but the engineers finally convinced the marketeers that it wasn't worth the money.

What's fascinating to me is that AMD realized this three years ago when they started producing things like the Athlon XP 1600+, which performed like a 1.6GHz processor but only ran at 1.4GHz. They had to do this to keep up numbers with Intel, who was pushing a P4-based Celeron 1.8GHz that could be outperformed by a P3-based 1.4GHz. And Intel has spent most of the past three years claiming a GHz-gap over AMD as technical superiority; meanwhile AMD has a P4-3.8GHz equivalent, the Athlon64 3800+, which runs at a nice, steady 2.4GHz actual clock speed. And trounces the yet-unreleased P4-3.8GHz in benchmarks. And will likely cost less.

The first sign Intel was wising up was the P4 "Extreme Edition." This included a couple of megs of extra L2 cache to improve performance, but the prices have been so high that not many consumers have been biting. The next sign was Intel's new processor numbering scheme that came out in June; instead of a "P4 3.6," it's called an Intel Pentium 4 360, taking the actual clock speed out of the marketing of the processor. Then we had the revelation that new chipsets were being cancelled to focus on evolution of the Pentium-M platform (which is the evolution of the shorter-pipelined, lower-GHz Pentium 3 core, if you'll recall). Intel also finally admitted that Itanium (IA-64) was NOT a reasonable 64-bit solution, and licensed AMD's 64-bit extensions to put in their own processors (since AMD64 is now the accepted 64-bit platform for Windows, a major coup for AMD) The final nail in the coffin of the P4 was the multi-core demo that AMD did with multi-core Opterons in an HP Proliant DL585. This forced Intel to pull off a hastily-arranged multi-core "technology preview" where no actual muti-core processors were shown, only a proof of concept demo.

Now the announcement that higher caches and multiple cores at lower speeds will be Intel's future in growing performance. How can anyone still question AMD's importance in innovation? Intel's clearly playing catch-up. With AMD undercutting Intel on price while matching or exceeding performance, I am going to re-affirm my membership in the AMD !!!!!! club. Hopefully the competition will allow AMD to gain some marketshare and force Intel to improve their product even further... that way, we all win!
Click to expand...
 
ZenOps said:
I like the new Dothans where they are right now, 400Mhz FSB @ 21 watt max. Really I'm not interested in the Dothan "upgrade" to 533mhz FSB if it means it will add another 6 watts.

It would be great if it works with the new 915M chipset, one huge request I would add would be the ability to have the ability to have CAS 1.5 timings at 266 or 333mhz when using very high quality PC3200+Cas2 or a FSB that matches 400mhz.
Click to expand...
I agree on the mem timings, but my thoughts on the additional 6 watts is the extra power drain will only be an issue under full load. The shorter the period of time under full load, seems the better control of power drain...
90% of the time, the Centrino's I've sold and used stay at ~600MHz, barely even producing heat, much less drawing power worth mentioning. A short burst to 2GHz drawing 27watts would not bother me one bit if the job got done quicker.

Now if you're talking 6 watts extra at all times... yea, I agree, screw it...
 
RogerX said:
yields with the .09micron process have been notoriously bad
Click to expand...
Please provide some backup for this statement. I have yet to find anything that confirms this claim, and seeing as how the 'Dothan' performs well in terms of frequency scaling and thermal control, there's one strike against your claim that the 90 nm process itself is the issue. Now, if you had meant to say that the Pentium 4's 90 nm iteration is facing low yields, then there's no problem.
The first sign Intel was wising up was the P4 "Extreme Edition." This included a couple of megs of extra L2 cache to improve performance
Click to expand...
I don't see this as "wising up," more like "keeping up with the Joneses," my reference of course being to the Athlon 64 FX, the competing super-processor. The EE did nothing to address the fundamental issue of frequency scaling under the NetBurst architecture; it only increased IPC like every Pentium 4 revision has attempted to do.
How can anyone still question AMD's importance in innovation?
Click to expand...
Innovation? I think not. It's only perceived as innovation. The idea of a more parallel processor than the Pentium 4 has been demonstrated by the POWER series from IBM, many other big-iron processors, and of course, Intel's own Itanium, Pentium M, and Pentium III. AMD was not the first to a 64-bit processor; it was beaten by Intel, IBM, DEC, and Alpha, if I recall correctly. The naming scheme is no innovation; the Cyrix processors "featured" these things. What about multicore processors? IBM's POWER5 processors are dual-core right now. No, AMD's pull is perceived as innovation because Intel can no longer pull off such high frequency scaling under today's manufacturing and market conditions. I'm no friend of Intel, or AMD for that matter, but let's not get carried away.
 
Not innovation, but a wise decision (it's late, ignore spelling errors).

Has there been a P4 to the market that is compatible with x86-64? I am admittidly AMD right now, so take it easy on me :D

If there isn't, does it meen there won't be now?

Vette
 
Yes, it's available in some markets but not the U.S. as far as I know.
 
This x86-64-compatible P4 will likely stealth-launch in the US (as opposed to a big-hoopla launch) for several reasons:

1. Other than Linux and UNIX, there are no operating systems that would run it at retail.
2. Intel is waiting on Microsoft to get XP Professional 64-bit out of beta.
3. An AMD-compatible Intel processor? That alone would give the Intel marketers hives!
4. The announcement of the cancellation of the 4 GHz P4 gives Intel some cover to intro the EMT64 P4s (P4 600-series, possibly?) at the northern end of the P4 lineup in North America.
 
RogerX said:
I sent this to a bunch of my co-workers, many of them who continuously re-affirm their opinion that AMD is garbage. While there is no question Intel never fails on stability, I take a different approach to this news:
Click to expand...

cheers to that!
 
Anyones else kinda urked that it was the end of 2002!!!! when we hit 3ghz? I mean its soon going to be 24 months since 3 ghz and we havent even hit 4ghz/4000+. Can you imagine how slow we would be if we did this dam 33% improvement every 2 years thing back in 98. Talking 450mhz, 2000-600mhz, 2002=800mhz, 2004=1ghz.

This is really really pathetic to have almost no performance gain by either party for 2 years. 33%, come on. Even 6ghz now would be considered imbarasing. Ya ya you can pull your a64 outa your ass and stuff, but its still not much faster then the 2002 stuff. I kept my 2.4c for like a year and then sold it for 80% of cost. Thats how bad the market has got. If it werent for video cards we would still be playing UT and counter strike.

wad a min....

gad dammit.
 
nightanole said:
Anyones else kinda urked that it was the end of 2002!!!! when we hit 3ghz? I mean its soon going to be 24 months since 3 ghz and we havent even hit 4ghz/4000+. Can you imagine how slow we would be if we did this dam 33% improvement every 2 years thing back in 98. Talking 450mhz, 2000-600mhz, 2002=800mhz, 2004=1ghz.

This is really really pathetic to have almost no performance gain by either party for 2 years. 33%, come on. Even 6ghz now would be considered imbarasing. Ya ya you can pull your a64 outa your ass and stuff, but its still not much faster then the 2002 stuff. I kept my 2.4c for like a year and then sold it for 80% of cost. Thats how bad the market has got. If it werent for video cards we would still be playing UT and counter strike.

wad a min....

gad dammit.
Click to expand...
what about fx53 and others...?
 
nightanole said:
Anyones else kinda urked that it was the end of 2002!!!! when we hit 3ghz? I mean its soon going to be 24 months since 3 ghz and we havent even hit 4ghz/4000+. Can you imagine how slow we would be if we did this dam 33% improvement every 2 years thing back in 98. Talking 450mhz, 2000-600mhz, 2002=800mhz, 2004=1ghz.

This is really really pathetic to have almost no performance gain by either party for 2 years. 33%, come on. Even 6ghz now would be considered imbarasing. Ya ya you can pull your a64 outa your ass and stuff, but its still not much faster then the 2002 stuff. I kept my 2.4c for like a year and then sold it for 80% of cost. Thats how bad the market has got. If it werent for video cards we would still be playing UT and counter strike.

wad a min....

gad dammit.
Click to expand...

I wouldnt mind ethier way, slow processors mean that developers would need to sell gameplay instead of tech. =)
 
I agree there, but I shudder to think of my vid editing with a 1GHz CPU...

Better software, but nothing helps vid editing like pure processing power.
 
nightanole said:
Anyones else kinda urked that it was the end of 2002!!!! when we hit 3ghz? I mean its soon going to be 24 months since 3 ghz and we havent even hit 4ghz/4000+. Can you imagine how slow we would be if we did this dam 33% improvement every 2 years thing back in 98. Talking 450mhz, 2000-600mhz, 2002=800mhz, 2004=1ghz.

This is really really pathetic to have almost no performance gain by either party for 2 years. 33%, come on. Even 6ghz now would be considered imbarasing. Ya ya you can pull your a64 outa your ass and stuff, but its still not much faster then the 2002 stuff. I kept my 2.4c for like a year and then sold it for 80% of cost. Thats how bad the market has got. If it werent for video cards we would still be playing UT and counter strike.

wad a min....

gad dammit.
Click to expand...

Yep, things have definitly slowed down. Look on the bright side! Now you can spend money on things like PCI-E and DDR2 that have absolutely no performance benefit!
 
Jonsey said:
Yep, things have definitly slowed down. Look on the bright side! Now you can spend money on things like PCI-E and DDR2 that have absolutely no performance benefit!
Click to expand...

atm they both dont have any real benefit... but i will agree with you on ddr2.. man is that shit worthless.
 
Jason711 said:
atm they both dont have any real benefit... but i will agree with you on ddr2.. man is that shit worthless.
Click to expand...
it's currently no speed beneift over existing technologies, but they will in the future.
 
There are rumors though that even the newer stuff won't use DDR2 if it doesn't have to; meaning if they have to make a new socket instead of using LGA775, it won't use DDR2.

But then again those are just rumors.
 
DropTech said:
There are rumors though that even the newer stuff won't use DDR2 if it doesn't have to; meaning if they have to make a new socket instead of using LGA775, it won't use DDR2.

But then again those are just rumors.
Click to expand...
well if we've reached the GHZ limit maybe we can up the FSB to match the clock speeds?!? lol
Get some ddr2 or ddr3 and have some dual or quad cores, 8-16mb cache, and shorter pipelines. then we'd start to see some beautiful things. :D

edit: and oh yea, we need some fucking solid state drives ASAP.
 
I would like to see the p4 extreme processors to come out as P4 Reloaded.

But I do like to see something other than sata drives, maybe revamp the fiber drives and come out with fiberoptic Hd's. If they can stream data over fiberoptic why can't they do it with hd's? Because they already do it with networks, think you could have a fiberoptic controller on the mb and since it is using light to stream data, who knows what could be possible.
 
SAS is the next thing to come out for hard drives. It's a combo of SCSI and SATA.
 
It's called Fiber Channel, and it's already common in enterprise networks and computers.

http://www.webopedia.com/TERM/F/Fibre_Channel.html

And light emission with integrated circuits like the motherboard controller you described is a real design challenge that hasn't been solved yet.

It'll be quite some time before we see hard drives that store data as an optical medium, if we see one at all. The real problem with speed in today's hard drives are electo-mechanical shortcomings.
 
Thank you for finding that info, I knew it was fiber something. Having optical media was a good thought but if someone did come out with a optical hd it would be kind of interesting.
 
i saw a 4gb solid state drive on CDW's site one time i think it was around $3,000 to $5,000. It even had an IDE interface. Now if we could find a controller to make sure it wasn't bottlenecked and get the prices down a few grand, solid state would be great :D
 
DDR2's latency doesn't appear to affect the P4 much, but it is more bandwidth oriented than latency. Pentium M or Athlon of any version probably won't benefit from DDR2 unless the clock speed is alot higher.

I really wonder whether DDR2 is the new RAMBUS?
 
If Intel got around to a Pentium M that has dual channel regular DDR with Cas 1.5/2 support, I'd bet it would easily surpass current P4's in terms of speed in anything memory intensive (like games, but not things like encoding)

Its coming soon with the 915M chipset, one I'm greatly looking forward to for desktops as well as laptops.
 
its not a win for amd



Intel makes Billions


AMD doesnt



Intel has contracts

AMD does not,



AMds top chips only 5% of market compared to intels

It will take more than 6 years while intels contracts run out for AMD to even start touching or hurting Intel In anyway
 
liquidtrance123 said:
i saw a 4gb solid state drive on CDW's site one time i think it was around $3,000 to $5,000. It even had an IDE interface. Now if we could find a controller to make sure it wasn't bottlenecked and get the prices down a few grand, solid state would be great :D
Click to expand...

no doubt, i wonder how much a 36GB version would cost. :)
 
ZenOps said:
If Intel got around to a Pentium M that has dual channel regular DDR with Cas 1.5/2 support, I'd bet it would easily surpass current P4's in terms of speed in anything memory intensive (like games, but not things like encoding)

Its coming soon with the 915M chipset, one I'm greatly looking forward to for desktops as well as laptops.
Click to expand...

cas 1.5 would be nice to see for sure!
 
You know, after reading the Inquirer article and also through a lot of these posts, I have to say that Intel kind of put themselves into this difficult situation to begin with. While AMD continued to focus on the balance of performance and features, Intel decided to drop the P3 architecture and instead invest in higher-latency architecture to simply satisfy the marketing machine.

I find it such an irony that the very same company that dumped such great architecture just to win the "MHz race" is now realizing that raw MHz speed isn’t the way to go. I’m forced to wonder if this isn’t what happens when you start to believe your own marketing hype. Still, if you look at the trends from both AMD and Intel, they are both trying to find the most efficient bridge between now and dual-core CPU’s. I’m not too impressed with the "let’s increase the cache" road, especially since it tends to add much more price than it’s worth.

So, I guess the real question is, "what next?" :rolleyes:

- James
 
I'm really curious to see how the dual-core P4s handle since Intel is having to void the entire principle of their current architecture and mend it into something else. The P4 architecture was designed around raising the Mhz to get higher performance. It'll be interesting to see them try to rengineer the core to change with the growing trends with processors. Personally I hope they just throw out the prescott and bring in dual-core Pentium Ms, those are gonna give AMD a hurtin.
 
a little while ago i recall reading about intel boasting of a 10ghz p4. :) (projected, not implemented)
 
You must log in or register to reply here.
Back
Top