I wouldn't say that Intel is doomed... they can come back. My latest computer uses AMD's Athlon XP. Its my first build and its rock solid. I used to own a Gateway computer that had a P3 @ 450Mhz that is still living large next to the Athlon XP computer. In my opinion, both companies have great chips and will continue to do so. Right now Intel is trying something new. If it pulls through (which it will since its Intel), then AMD will have to push their 64-bit processors more towards the general public (but that may hit a snag since the only OS that I know of that supports the 64-bit chips is Linux).

Bottom line: It will be a battle that will be waged for an eternity with both sides overtaking the other frequently.

You know, we could always say screw both of them and go with less costly and faster RISC processors.
You wouldn't necessarily have to stray far from the Windows GUI either, as there are Linux kernels running on RISC.


I have yet to see benchmarks of the Prescott with hyperthreading turned on or with adequate cooling.
Since the P4, Intel has integrated a thermal control that cuts the CPU clock when the processor starts to overheat.

That could possibly be an issue.


You guys gotta remember, however, that AMD is still struggling to get over the 3ghz mark.

With a bit of cooling and overclocking, a P4 @ 2.4 can hit close to 4ghz, and a 3 can go above 4ghz.

Clock for clock, Intel is definitely winning. But in the long run they're losing because AMD has been using a 128kb L1 and Intel has stuck with an 8kb L1.


The problem I see is that Intel has opened up far too many labs across the world, if they closed down two of them, spent a few extra months producing processors, and focused on improving on what they have instead of adding to it.. they'd throw AMD out of business.

Many people don't get Intel processors from the US, but there is a lab that produces processors in the US. This is the same lab where Intel tests everything. They would save more money if they kept their corporate offices in the US, but stationed their lab in Canada or Mexico... hell... if they wanted to they could save even more money and open it in Cuba. Then they could easily send mockups to Puerto Rico and their other labs.


Intel's worst problem is lack of funding for better R&D.


I'm sure most people can agree that the best solution would be cut the core down and build it back up from an entirely new starting point, because if you look at the P3 core, not much as changed.

AMD hasn't made it over the 3GHz mark on its own. Put on a prommie and you can hit 4.6GHz. That's like a P4 at 5.4. Since AMD uses that shorter pipeline with more data-directing cache, they don't need as much speed. If they were at, say, 3.2GHz then Intel would be s**tting its pants right now in terror. Intel has nothing that stands up to the A64 series NOW, just wait until the FX53 is here! Intel will probably say "Oh well you know we haven't had much time, blah blah blah, new itanium, etc...".
Microsoft should be against the law. There's three main triangles I've figured out here:
nVidia + Intel + Microsoft = axis of evil
ATI + AMD + SCO/BSD/Linux = the young jedis
Yeah I use an nVidia card, but it turns into a keychain in a few weeks when my R9600XT comes. Hey, who would have ever guessed my FX5200 could get a steady 60FPS full detail in UT2K4??? Hey, I know! I'll put an Iceberq4 on it and sell it (for $100) to this Vietnamese kid from the middle school across the street! He'll never know! He uses a DELL!!!
Are all Finns computer geniuses?

I felt the article was garbage imho, no where does it say what the testing methods were anywhere in the article. I figured it would have been something to the effect of Prime 95 if not Prime 95 itself. I don't understand how this program could possibly offer any real world examples of life with a prescott. As Prescott has HT support I figure it will behave much like my Northwood "C", as in it will never see 100% load in any applications other than Prime 95, possibly Sandra and maybe Folding@home.

I don't believe that the Prescott is as bad as poor review sites are suggesting. Infact, another review site: http://www.legitreviews.com/Reviews/prescott_cooling_2.shtml shows the effect of a Thermalright's SP-94 with a 80mm 80cfm fan keeping the Prescott at a cool 52c underload/44c idle. Granted this solution will be loud however 92mm fans are more popular and less audible. While debating at nVnews.net forums, when compared to a A64 3400+ w/retail HSF and My Northwood 2.8C@3.5Ghz w/retail HSF, the temps appear to be within 5c for either the A64 or Northwood. I went so far as to run both UT2K4 and Farcry demo's simultaneously to see if I could get 100% CPU usage out of what I think is as "real" as oppose to Prime 95's results. To some of the other members surprise, it would not get close to 100% CPU usuage, let alone hit 70% usage.

Pic: http://www.battlefieldgrunts.net/graphics/Screenshots/Isthatrightstill.JPG

So when that result presented itself, I began to contemplate how Prime 95 represents what daily or common folk endure. So I have come up with a list of suggestions that could possibly help you in becoming a better reviewer of products when it comes to the amount of heat they generate. Especially under the circumstances that no one that I know of actually plays more than one game at a time, let alone can put a 100% load on a Intel P4 with HT. Some of these will appear down right odd, though they can easily relate to anyone who buys a pc either through Dell, Gateway or the typical user.

1) Games: Pretty obvious, can't go wrong there right? One at a time ofcourse.

2) Apps: Office 2003 suite - Ya, it is rediculous... though it is actually what people use correct? At least that is typically included in the pc packages when someone buys a Dell or Gateway comp. Infact, the entire list of options you see listed as potential purchasing decisions on Dell or Gateways websites.

3) Video Encoding: Very popular now

4) Anti-Virus software: Almost everyone has some sort of Anti-Virus software right? It does put a load on the CPU as well, could be a good option.

5) Ad-ware and Spybot Spam buster software: Very useful and quite common as spam is everyones problem.

6) CAD Software: Again, it has potential right?

7) Adobe Video Editing software suites: Another common proggy

8) Muya 3D software: A great deal of web designers, game dev's and game modders use this

9) 3D Studio Max: same reasons as above

10) MP3 ripping software: Way too common

That is ten I can list off the top of my head right now. They are pretty common and some are very boring actually but still real world applications.

Peace,

=)

It sucks! No way around it.
My friend's roommate got one and it sucks! It's not too great at gaming, but it encodes with LAME fine. That's about it. Seems like more of a server chip.

I'm I missing something here? What about Win XP 64? I thought it was designed for the Itanium??? Besides there are no applications programs for the 64 yet. If it had been designed to make 32 bits run faster or for that matter run at all like the AMD 64 then all of this might have some meaning.

Somebody mentioned RISC processors... anyone know of just an inexpensive dual processor board that Win XP will support that allows a 2.4 Pentium 4 compete with the 3.2 Northwood? Hey when you can afford the fastest CPU and their dragging your feet making them any faster then why not try two or three instead of only one?

Which one of you has the specs or a reference?

XP 64 is not for Itanium. Itanium died out years ago. Linux has support for 64-bit, and XP 64 will be AMD64 optimized. UT2K4 and HL2 have confirmed future 64-bit counterparts, and UT2K3 already IS 64-bit. Media programs are going 64, Vue D'Esprit and similar renderers are 64-bit optimized now, and even media players. Caching programs are 64-bit. Master boot-records now have 64-bit optimizations. Memory headers and buffers have 64-bit optimizations (primarily Corsair XMS). Server apps are almost all 64-bit now.
"...making 32-bits faster..."
The A64 does 32 bit better than most if not all P4s and AXPs, and it's not even specialized at 32-bit! Imagine the 64-bit performance!

So then if we follow the trend from 8 to 16 to 32 to 64... and consider the speed improvements per each bus then why even bother with making CPU's faster and just go for an expanded bus and for that matter a bunch more CPU's? I means all this haggle over CPU speed comes down to the maximum speed of one transistor and we know how to make one transitor go faster by making it out of superconducting semiconductor material (which by the way could be slightly dopped silicon and run at speeds that produce no noice). What is really important in the short term is not CPU speed but system speed and more CPU's or an expanded bus seem to be the solution there. Sounds like to me that the computer industry could make a giant leap but has decided to plan ahead and no matter what is going on in an AMD or an Intel lab the plan is to stick you and me. If I add up the cost of all the systems I have purchased since my first Trash 80 and all of the software I have purchased over the years I could probably afford to build my own super computer cluster out of a very respectable number of obsolete PC's. I'll stop here. Want to keep my plans under wrap as well...

Carter, you have no idea what you're talking about.

First of all, for anyone that doesn't understand the difference between 32bit processing and 64bit processing, let me clarify.

32bit processors have built in instruction sets for 32 bit applications and operating systems. As we all know, hardware based functions are faster than software based functions.

But by no means does this indicate you can't run a 64bit application on a 32bit proccessor, it just means the application will be entirely run in software emulation. It might be laden with errors, due to no hardware speculation, but it will execute into memory at least.


On a 64bit platform you have seperate instruction sets or one combined pool of instruction sets with a limitation in between the 32 and 64 bit instructions.

You can't translate a 64bit instruction to a 32bit application, and vice versa.

Despite the processor speed, all 32 bit processors run at the same performance...... 32bit.

It's the other die components such as the pipeline, latency, caches, etc that determine the speed of the processor.

So Carter, where you said look at the performance of the AMD in 32bit proccesses..... it's not gunna matter in 64bit.


Windows XP 64bit is NOT dead, it just isn't widely used for two obvious reasons:

1) What you can do on Windows, you can do twice as better on Linux, provided you have the right software.

2) The only true 64bit CISC processors are the Itanium 2's.


The Athlon 64 should not be part of this discussion. It's only competition is itself, because Intel has yet to release a 64bit processor to the public. The Prescott is still in the 32bit range, and will stay there until mid-2006.

IBM collaborated with AMD on the production and design of the 64, AMD can't take credit because they've only adapted IBM's achievements to their line of processors.


Here is the bottom line:

+++The Prescott is only hot for stupid reviewers that are using the stock heatsink. Intel and AMD both recommend purchasing higher rated cooling, their stock heatsinks are only a cheap way to cut costs for retail PC sellers. Each company uses their heatsinks for real world tests, but the processors they use in their offices are all using higher powered cooling.

+++AMD's 64 will either set them back 2 years or set them ahead a year. Back because their processor isn't fully 64bit, or ahead because they're stepping into areas Intel has yet to go.

+++The general public won't notice a difference in 64bit processing unless they know where to look:

- Gaming will NOT change drastically. Most games rely on the graphics card to process and display material, no graphical processing is done on the cpu.

- Office work will not be faster or more reliable, it will stay the same as it has for the past few years.

- Audio processing won't receive any benefits.

- Video editing will become faster, because of enhanced instruction sets.

- Rendering and imagery will recieve the largest speed and efficiency increase. They are most dependent on instruction sets.

- Servers will become more efficient, due to a more solid platform 64bit processing provides.




Everyone stay on topic or make a new thread.

Thanks Sean for clearing that up.

The point I'm making is that the Prescott, like any other component, generates waste heat. Sinking the waste heat is not the problem although this is 50% of the topic at hand.

The problem is why the Prescott, or any other component for that matter, generates waste heat. 45% of the topic is that waste heat is generated by overclocking.

That leave 5% for speculation as to what the remedy is.

I'm ready to start another thread on potential remedies for the problem of resistance imposed by the conductors and semiconductors on the die per rate of switching or operating frequency.

However, I feel that it is a foregone conclusion that superconducting semiconductors are the solution whether they are high temperature or low temperature.

So that brings us back to using a higher capacity air cooled sink where we may have found the opportunity to introduce water cooled heat sinks.

On the other hand water cooling may be waiting for the processor that has exhausted all of the volume inside the case (especially when you are talking about Small Form Factor).

Heat pipes may have an intermediate role but lets face it 90% of the vehicles on the road use water (mixed with coolant) and not air to reduce the highest temperatures before using air to dissipate it to ambient.

So where do we go from here?

superconducting semiconductors? isn't that a contridiction? if AMD or Intel had a material that was superconduting at the temperatures that CPU's operate at they could make alot more money making superconducting stuff then they ever would making chips (currently the only way to make anything superconducting is to get a good conductor and cool it to hundreds of degrees below 0)

Also what is the "waste" heat you are talking about? If you have electric current running through anything with any electrical resistance you generate heat by disipating power the power disipated is measured in watts... and can be calculated by multiplying the current (measured in amperes) by the voltage. most modern processors disipate 60 - 80 watts so your processor makes about as much heat as a lightbulb except its surface area is much smaller than a lightbulb so the heat is much more intense. the only way to reduce the heat is to lower the resistance of the material or decrease the voltage or current flowing through it.

Like the man said bone up first and then start a new thread.

thanks, intel brainwashees. i know what 64-bit is, dick. and one transistor doesn't make the difference. it's how well they all work together. and a 32-bit cpu doesn't have the framework for 64-bit apps, hence the 'errors' and crashes. and multi-cpus are only good multithreaded, as in servers and hackboxes. dont call sanders stupid. he's got more experience in the field than you. the prescott's a hot son-of-a-bitch. our IT guy even hates them for how hot they get, and he ONLY uses P4s! learn the facts. don't go all out for a CPU until you've used it yourself. it's not that great, don't have an orgasm over it.

There's no reason to start a new thread, this was what the topic was supposed to be before someone (I won't name names, but they know who they are) took it in the wrong direction and started making it about other companies and processors.


The solution to the cooling issue is to move away from silicon, which acts like a heat container, and move to a metalloid.


Intel released specs for their Prescott months ago with indication of a K-metal (someone please come up with the exact name, I don't want people, including me, to think I'm talking out of my ass here..... ).

Now, either I was on an acid trip or I read this Intel-published article based on the new silicon substitute they'd be using.

I still can't figure out which of the two produced the pool of drool I'm encompassed in.


I will make an effort to post the link to the article in my next post.... but for now let me run by the key points of this.


Using a metal would increase electrical conduction, allowing for a higher throughput on current transistors.

Transistor distance will decrease as heat increases, due to heat expansion, allowing for smaller but less densely packed transistors to be used.

Thermal properties of any metal or metalloid are better than a silicon, allowing for a rapid heat exchange.

Because of heat expansion and compression, hotter processors will run faster(within limit), whereas colder processors will produce the same results with greater efficiency.



Here's the bottom line with the Prescott:

It runs hot because it's running on current technology. Intel has just found a way to cut down their P4 and add new features.

Once Intel realizes that they have customers to please, they'll shape up. They'll wake up their R&D team and produce an entirely new core, like they've done for every Pentium.

The heat spreader featured on all Pentium 4's aren't gold like they should be, but rather a cheap aluminum/copper.

Not only that, but a small pocket of air gets trapped between the die and the heatspreader.
With a smaller die, more air gets trapped, storing more heat around the processor.
If you remove that heatspreader, you reduce the heat production in half.


Don't believe me? Go buy a cheap Celeron and remove the spreader with a razor blade. It will drop the temps 10-15°.

The mobile design of the P4 doesn't feature this heatspreader, and it runs at a lower voltage, if you purchase one seperately and install it on a nice watercooling rig, you'll be able to overclock at least 500mhz more than a desktop counterpart would. (a little off topic, I realize, but I just wanted to bring it up)

Sean B is smart. You all could learn something from him. Everything he said in his post is true, even about the K-something metal. However, VIA announced they are done with metals and silicons. The new wave is biological material that can conduct electricity hundreds of times better than silicon, even at the 130nm format. This will later lead to voltages below 1V. And yes, the Prescott is on current technology, but only in the 32-bit realm. One of you mentioned no translations or emulations from 32-bit to 64-bit, which proves my case that the Prescott will die out soon, even with "CP" or "Yamhill" extensions. It is still a 32-bit CPU.
AMD uses a solid aluminum/copper alloy heatspreader design with the cache in a seperate area than the main core (on the A64). I bet if you lapped an A64s with some 1200DPI sand-paper down to the core you would see 3 different squares - one being the core, one being the L1, and the one the size of a Barton core must be the L2. AMD has been working on refining their trace methods. The Prescott runs hot because:
1. It is constantly handling data tracing for its larger L2 cache.
2. Its L2 cache is taking more voltage away from the central core. This leads to more heat dissipating AROUND the core instead of IN the core.
Just wait, the FX53 will be AMD's shining hour.

Okay. Time for new thread. Here it is: Superconducting Semiconducting

Yer know that 64 bit software will probably have twice the errors and crashes found in 32-bit because it will be far more complex. That's right.

That's wrong. There will be two different ways of reaching the same software interface, so if one method fails it could just use another.

Hey now Carter you to the whole concept of clustering right out of my mouth.

In case of failure!

Duplication and redundancy can not only be used for the purpose of bullet proof reliability but to help speed things up in the absence of failure.

The only things that are needed to accomodate a 64bit platform are the proper hardware and operating system.

Nobody noticed the transition from 16bit to 32bit because it was so quick.
The same transition can happen in this day and age in half the time.

Why? Numerous reasons.

For one, the user base on computers has changed drastically. It used to be that computing was for hobbyists, scientists, industry professionals, and the government. We now have preteens, teens, and the elderly booting up and logging on.

It's the whole concept of supply and demand. The largest demand for speed, reliability, and just plain new technology has happened in the past 5 years.

Hence the reason for the Prescott. We all know it's just a stepping stone processor for those who don't have a computer, or need more speed and can't wait until a new platform comes out.


Some could concur that the Athlon 64 is also a stepping stone processor, since it isn't using the BTX design, doesn't feature PCI-X, or any of the other functions next generation systems will utilize.


I'm perfectly happy with my 2.4 running at 3.5..... and yes that's with air.


{edit}
Oh, one last thing, if you made a clustered farm of the Via C3's, you'd have a cost effective and efficient system without any cooling at all.

They suck alone, but so do ants. Once you get them together they will devour anything you put them up against.
{edit}