Intel document confirms Prescott dissipates 103 W

eden · Aug 22, 2003

Lol, a FLAME war. It's a FLAME war.

Now, did you get the hint? It's in the war kind, our target....muhahahaa.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

Groveling_Wyrm · Aug 22, 2003

Bah, Flamewars of the deep were filled with insults and absolute trash.

I wouldn't call this a flamewar at all.

Sorry...I had to reply to this...

I didn't say that it was fully back to normal....but in the process of "returning" to "normal" LOL!

....and look...the mere mention of the term "flame war" brings em outa the woodwork...ROFL!!!

 Ok, so you have to put your "2 cents" in, but its value is only "A penny's worth". Who gets that extra penny?

Dark_Archonis · Aug 22, 2003

Written by Eden
You did not even read right if that's how you see it.
I said I slowly changed my mind given that now there is more evidence that AMD is responding and improving.

Well, in that stock thread, you show a sudden surge of hope for AMD. Now, you say that things have changed over the past month. I say they haven't. AMD has had IBM support over a month ago; the Opteron still does not perform that great (even at AMD Zone), and AMD still has not shown us their ability to raise core clock speeds. Yes, Microsoft will soon add x86-64 support to Windows, but AFAIK, that will only be in Server 2003 SP1. Plus, at the same time, they will add IA-64 software emulation support for x86. Not to mention Deerfield will soon make it's appearance. I'm just confused as to why you suddenly believe in AMD again.

Written by Eden
BTW Dark, how DO you feel about nVidia lately, if I may ask?

Well, I do not feel that great. TO be honest, I expected more out of the NV3x architecture. The IQ for Nvidia's cards still does not equal ATI unfortunately. Also, thier prices IMHO, should be a little more competitive. With regards to the whole "cheating" thing, I have no comment and do not really care. Why? Because it has yet to be discovered in a game. It was only discovered in 3dmark. Furthermore, the cheating cannot be confirmed now with the new drivers.

I am hoping that the NV40 wil improve Nvidia's situation.

Sure, Nvidia's cards are still good products, and have good drivers, and I still trust Nvidia more for stability and compatibility than ATI. I just feel that Nvidia has gone off track a bit.

- - -
"... In the semiconductor industry, it's good to be paranoid ..." - [Andy Grove]

Edited by Dark_Archonis on 08/22/03 01:56 PM.

Dark_Archonis · Aug 22, 2003

Written by dhlucke
Look. Let me make this perfectly clear.

Heat. It has to go somewhere.

Is there a difference between power output and heat displacement and what is it for the prescott?

Heat Management? What does that mean? Just because a HSF is good at getting the heat away from the core does not mean that the heat just dissapeared after that.

Well, obviously it goes somewhere. But, there's a difference in how fast the heat moves away from the core, and also a difference in the amount of area the heat has to work with (die size). Strictly comparing the die sizes of Northwood and Prescott, Prescott has a smaller die size, AND it outputs more heat per square mm. When Intel moved from Williamette to Northwood, they moved from the 180nm process to the 130nm process, moved to copper interconnects, rearranged the core, and had some small undocumented tweaks. Moving to copper was of bigger significance then actually moving to the 130nm process. Copper is more efficient and a better conductor than aluminum, and it has more heat capacity too. This is one of the big reasons why Northwood outputted less heat than Williamette. Sure, the 130nm process helped, and the more aggressive clock gating didn't hurt either. Now, when moving to the 90nm process, Intel has added strained silicon, which does not really help alleviate heat output, and low k dielectric, which does. Strained silicon allows for faster electron movement, and thus allows for a faster clock. More importantly, Prescott features a whack of core improvements. One of those is a rearranged core. The core was rearranged primary for better performance, including FP performance, and HT performance. The fact that Prescott features a better HT implementation reduces the effectiveness of clock gating. Now, AFAIK, Intel has used the same IHS for Prescott as Northwood. They are currently working on a core revision to reduce Prescott's dissipation. The fact that the dissipation was higher than Intel's best estimates indicates that it probably is leakage which is the problem. Over at Anandtech, an Intel employee himself stated that leakage is a significant problem on the 90nm process. Many members here call BS on my leakage comments, but it makes perfect sense. Why then did Intel move to a new low-k supplier? Why then did that Intel employee at Anandtech state that? Why then does the majority over at Ace's and Overclockers.com agree that this IS leakage?

Now, I do not know how Intel will alleviate this heat putout problem with a core revision. It might be a new IHS (which Tejas will use), or a new core rearrangement, who knows. At the very least, there will probably be a new HSF.

Intel is starting to hit a lot of thermal problems. With Tejas, Intel cannot really do a lot. Though, they are introducing a new HSF, IHS, and case design to alleviate the heat problem. When Nehalem comes out on the 65nm process, Intel will use FD-SOI, to say the least.

- - -
"... In the semiconductor industry, it's good to be paranoid ..." - [Andy Grove]

Mephistopheles · Aug 22, 2003

Power output doesn't necessarily equal heat output.

Ouch. That's not a physically acceptable statement there, sorry. All power that is used by a processor will be dissipated as heat, and that is fact. However, the heat has to be dissipated by the heatsink. And the two things that you have to keep apart are heat and temperature. Now heat can be dissipated without a temperature increase - it just needs a good heat conducting device for that to happen. I think that is what some people referred to as "heat management" or something.

Essentially, heat output itself is not the problem at all. So prescott outputs 100W of heat, so what? Big deal! The problem is that it is hard to deal with that much thermal output on a small surface which has to be kept at under, say, 60C. If Intel can develop a standard heatsink that keeps Scotty running at 50C, then the fact that it is dissipating 100W of thermal energy is completely irrelevant, except for room temperature increases.

And for those worried about that 35W or so increase in thermal output from current processors to Scotty, I'd just have to say this: get real. This is roughly the same as putting one of those old, weakest-model 40W-tungsten filament lamps in your room. Your stereo probably outputs more heat than that. Your TV certainly outputs much more heat than that. It's a 30-35W increase,for crying out loud! The problem is cooling the processor to an acceptable temperature, that's all. If that has already been achieved, then unless your room cooling is poor (like you live in a thermally insulating room, which is already a dumb thing) your room temperature will increase so little you won't even feel it.

For those complaining about current 60W CPUs: Are you aware that the CPU is only one of the components heating up your room? Your computer might be running in excess of 450W or so, and you're blaiming the processor... And if it's that hot in your room, geez, just open your window!

:evil:

MephistophelesEdited by Mephistopheles on 08/22/03 04:08 PM.

dhlucke · Aug 22, 2003

I can't believe I can't get an answer.

If my Tbird is at 54.5W and it heats up my room, what is the Prescott at 103W going to do?

______________________________________________
<A HREF="http://www.teenirc.net/chat/tomshardware.html" target="_new">http://www.teenirc.net/chat/tomshardware.html</A>

dhlucke · Aug 22, 2003

You noted room temperature increases. I'm trying to point out that irregardless of the core temp, which we all know will be fine due to the HSF, the heat has to go somewhere.

If the Prescott is going to fill up my room with the heat output from 103W, it's going to be unbearable.

______________________________________________
<A HREF="http://www.teenirc.net/chat/tomshardware.html" target="_new">http://www.teenirc.net/chat/tomshardware.html</A>

eden · Aug 22, 2003

I believe you meant to quote and reply to ME on that statement.

My bad, you are right. I figured in my head it wasn't logical to transform power energy to heat and lose some. And as you stated, it has to turn entirely into heat. What I was more confused to is that even if it does dissipate well, are we lowering the temperature of the hot air, or are we just moving it away, and if so, isn't the case inside AND the exterior room where it is in, at a greater risk of having way too hot temperatures?

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

eden · Aug 22, 2003

Well, in that stock thread, you show a sudden surge of hope for AMD. Now, you say that things have changed over the past month. I say they haven't.

You're wrong. Big time. IBM has had chip foundry layoffs lately. AMD did NOT announce a 2.2GHZ Athlon 64 back then. There were major profit problems reported, and little rises in consumer confidence. In other words, AMD had NOTHING to provide to us to reassure us they just might get the crown in a month and a half.

AMD has had IBM support over a month ago;

Ya and? Where do you see me criticizing IBM does not support? The question is WHAT is provided. I bet all IBM is doing now is helping improve the SOI process, which they themselves could not even harness it properly, being the original creators.

the Opteron still does not perform that great (even at AMD Zone),

I don't believe entirely AMD Zone, but I am more than positive a 2.4GHZ Athlon 64 can beat a 3.2GHZ handily, with no troubles.

and AMD still has not shown us their ability to raise core clock speeds.

Really?
Weeks ago it was nothing but clock speed trouble news. Now we hear of a possible 2.4GHZ (one linked to this previously), and AMD has a 2.2GHZ ready. Sure it's nothing new, but at this time there have been enough articles pointing that they finally are getting closer to being able to raise clock speeds. Since the process used is identical to the K7, and MANY are able to reach 2.5GHZ on their Tbreds, (perhaps even Bartons) I say it is a given the A64 will reach that, and it's only a matter of finding the right cooling. If SOI works, it will be easy to cool.

that will only be in Server 2003 SP1

How do you know it won't be an entirely seperate OS? Fact is there WILL be. Opteron is getting that SP1 patch so there will be a current WINDOWS to use x86-64.

I'm just confused as to why you suddenly believe in AMD again

Look dude, deal with it. You obviously are still as biased as ever, and seeing an open-minded person who tries to look OBJECTIVELY at the PRESENT news and times, is only annoying you. You can't respect my views because they conflict with your bias.
I HAVE begun feeling more hope for AMD, due to RECENT NEWS. Either you didn't follow them, or you are just so close-minded you dismiss AMD news.

Once again, DEAL WITH IT, I look at things NOW and react to what's NOW. Got a problem with that? Leave the forum, ignore me, be happy.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.Edited by Eden on 08/22/03 03:38 PM.

eden · Aug 22, 2003

That is pretty much my concern as well. If the heat can't be controlled and its temperature is reduced, then yeah, where will it go and what will it do to the room it is in?

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

slvr_phoenix · Aug 22, 2003

dhlucke and Eden...

Yes, the power is mostly dissipated as heat. (Though some of that power is also dissipated in other forms such as electro-magnetic radiation.) Yes there is more heat with more watts. Just because heatsinks have gotten better at moving the heat away from the CPU core doesn't mean that CPUs themselves put out less heat. It just means that they run cooler because the heat is being transferred to the air in your room better. So yes, your room will get hotter with a processor that uses more power.

However the CPU itself is still a very small amount in the grand scheme of what heats up your room. Lights for example still put out far more heat than most CPUs. So do power supplies for that matter. But more than all of these things is you. The human being expells an awful lot of heat. You might think that it's your computer heating up your room, but often it is actually you. If you were to sit there with the computer off it'd still get hot just from your charming presence. 😉

That aside though, if a Tbird at 54.5W is too much already, then don't even bother looking at a Palomino, because every last one of them is over that. And while you're at it don't try upgrading to any Athlon past a TBred 1900+ because that's at 52.2W. An AXP 3000+ is 74.3W! And a Pentium4 is pretty much right out too. The only P4 that you could use would be a 2GHz NorthwoodA or Celeron (who's W range a little bit because they are offered in a couple different voltages). Anything higher would be too much.

I dare say that the magickal ThoroughbredB AXP1700+ would be the only real way to reduce the heat from the CPU and still gain any higher performance. Not that it would do much to make the room any less hot though.

It's too bad that the Pentium 3 is long dead because they were the real champs of cool operation with good performance.

If you look The Devil® straight in the eye and only see yourself then you must be standing in front of a mirror.

eden · Aug 22, 2003

It's too bad that the Pentium 3 is long dead because they were the real champs of cool operation with good performance.

I'll have you know that VIA has a CPU under 10Watts, HUMF.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

HardWareBoss · Aug 23, 2003

P3 is not dead as long as Piii man is around.

eden · Aug 23, 2003

And P2 is not dead as long as Crashman is around!

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

jmecor · Aug 23, 2003

More heat is added on PRESSHOT:

<A HREF="http://www.geek.com/news/geeknews/2003Aug/bch20030822021375.htm" target="_new">http://www.geek.com/news/geeknews/2003Aug/bch20030822021375.htm</A>


"Remain at stock speed" - The Overclocker.

phsstpok · Aug 23, 2003

Essentially, heat output itself is not the problem at all. So prescott outputs 100W of heat, so what? Big deal! The problem is that it is hard to deal with that much thermal output on a small surface which has to be kept at under, say, 60C. If Intel can develop a standard heatsink that keeps Scotty running at 50C, then the fact that it is dissipating 100W of thermal energy is completely irrelevant

Well said.

And for those worried about that 35W or so increase in thermal output from current processors to Scotty, I'd just have to say this: get real. This is roughly the same as putting one of those old, weakest-model 40W-tungsten filament lamps in your room. Your stereo probably outputs more heat than that. Your TV certainly outputs much more heat than that. It's a 30-35W increase,for crying out loud! The problem is cooling the processor to an acceptable temperature, that's all. If that has already been achieved, then unless your room cooling is poor (like you live in a thermally insulating room, which is already a dumb thing) your room temperature will increase so little you won't even feel it.

This argument doesn't quite work.

Think about putting your hand on that 40 watt light bulb. Now think about sticking that light bulb inside your PC case. It's a significant amount of extra heat.

An extra 30-35 watts of heat will challenge air cooling solutions.

56K, slow and steady does not win the race on internet!

tombance · Aug 23, 2003

I cant see how the argument about having a larger chip surface area makes any difference. A processor still gives out a set amount of heat and even if it is transferred to the HSF better that heat still has to go into the air. Your room would still get to the exact same temperature even if you were using a teeny HSF that was 70C at idle.

Also, the argument that the P4 has better heat dissipating qualities is wrong because the HSF still makes contact to the core, and that simply adds another layer of heat transfer to the equation. The only way a bigger chip surface area would help would be if the surface area came from from increased silicon die size, not simply from a slab of aluminum stuck on top.

<A HREF="http://service.futuremark.com/compare?2k1=6752830" target="_new">Yay, I Finally broke the 12k barrier!!</A>

dhlucke · Aug 23, 2003

The only reason I brought up the room temperature is because I have a 10'x12' (roughly) room and it gets hot. When I had my PC in a bigger room this was a mute point. But those people living in apartments or dorms or prison cells really need to consider the heat as a factor. I don't find sweating it out very fun.

______________________________________________
<A HREF="http://www.teenirc.net/chat/tomshardware.html" target="_new">http://www.teenirc.net/chat/tomshardware.html</A>

eden · Aug 23, 2003

True indeed. That is a bad thing anyways.

I must wonder though, how does watercooling work in heat transfer?
Would your room STILL be as hot with a standard air cooler than with a watercooler?

And does AC take care of the heat in the room and lower its temp?

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

simwiz2 · Aug 24, 2003

I have a 2.53 and Ti4600 in a small room, and when the central air conditioning is actually running it takes care of the heat fine, but when it goes off the room heats up much faster than the rest of the house and the room is uncomfortably hot before the air conditioning turns on again.

eden · Aug 24, 2003

Yeah, and it IS bothering, contrary to what some might say.
There is nothing else that could cause more heat like this in a room than the PC.

It is annoying in summer, since I don't have an AC in my room, but a fan near the door to attempt and let in the ACed air from the living room. They should try to research ways to keep CPUs from reaching such levels of heat. Even normal consumers will become bothered eventually.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

phsstpok · Aug 24, 2003

I cant see how the argument about having a larger chip surface area makes any difference

No one said it did. Read Mephistopheles' comment again.

The problem is that it is hard to deal with that much thermal output on a small surface which has to be kept at under, say, 60C. If Intel can develop a standard heatsink that keeps Scotty running at 50C, then the fact that it is dissipating 100W of thermal energy is completely irrelevant

Isn't he essentially saying the same thing you did?

If the heat can be sufficiently dissipated then who cares what the die size, TIM, or heatsink is.

I don't even care what typical die temp will be, 40 degrees or 70 degrees as long as the processor runs stably.

What I did say is that 30-35 watts of additional power is significant. Dealing with the extra heat, looking at the PC as a whole system, is a challenge.

Right now those of us with overclocked Athlons, Tbirds thru Bartons are already having to compromise between noise and cooling, at least those of us using air cooling are. We are dealing with processors that nominally are dissipating 50, 60, 70 watts and perhaps 80-85 watts overclocked. This is difficult enough to keep cool and quiet (whatever your definition of quiet is). However, now we talk about Presscott which needs to dissipate 103 watts at normal speed (pressumably) and higher if overclocked. Wouldn't such a processor present even more of a challenge to air cooling designs?

That's all I'm saying.

56K, slow and steady does not win the race on internet!

HardWareBoss · Aug 24, 2003

Looks like it's hard to kill those old Intel chips.

eden · Aug 24, 2003

I don't even care what typical die temp will be, 40 degrees or 70 degrees as long as the processor runs stably.

No offense but it is foolish to say that without acknowledging the shortcomings.
Having extremely high temps gets you MUCH closer to the theoretical die destruction threshold which is NOT good, and most importantly, you significantly reduce the life of your CPU.

--
<A HREF="http://www.lochel.com/THGC/album.html" target="_new">A sexual experience like never before seen</A>
Site has now even more sexy members, for your pleasure.

scamtrOn · Aug 24, 2003

i thought I was the target...


sex is bad, bad is sin, sin is forgiven so sex is in

Intel document confirms Prescott dissipates 103 W

Champion

Distinguished

Distinguished

Distinguished

Distinguished

Polypheme

Polypheme

Champion

Champion

Champion

Splendid

Champion

Splendid

Champion

Distinguished

Splendid

Distinguished

Polypheme

Champion

Distinguished

Champion

Splendid

Splendid

Champion

Illustrious

Share this page