Explain this to me, hot athlons

[m2]

Here is a direct quote from a very recent Anandtech article.

"This paints a picture of two problems with the Athlon, it?s running too hot, and it?s drawing too much power"

If that is the case, what is the difference between an athlon and an overclocked PIII, which runs at the same temperature ( apart from an obviously different architecture ) ? Are AMD having us on ?

This is a genuine question, not a an attempt to wind anyone up.



Thankyou


Eric

 

[SilverBack]

They don't run at the same temperature. The TBirds are running 10 to 15C above the P3's.

 

[AdamK47]

Thats true. My P3 650 at 1021MHz ran much cooler than my Athlon. The Hedgehog heatsink I'm using would barely get warm with the P3, but with the Athlon it is very warm.

 

[Mikewarrior2]

A p3 at 1ghz is roughly 30W

A t-bird at 1ghz is roughly 56W

So with a, say FOP38 as an example, the cpu would run 11C over ambient case for the p3, and ~11C over ambient with the t-bird.

They run warmer, but it usually is not a problem.


Mike

 

[shiznut123]

mikewarror2, so my arctic cooler cant cool down a tbird?

 

[Mikewarrior2]

It could, just don't expect it to perform miracles. It should run the core temp around teh same temp as the taisol cek734092(approximately .45C/W). How the mb measures teh temp is an entirely different issue.


Mike

 

[m2]

Thanks for the replies everyone, still confused though about why athlons draw so much power and generate so much heat


Eric

 

[pm]

Power, performance and design time are factors that seem, in the world of semiconductor design, to be something that you can get two of, but not three.

So you can have a short design time and high performance, but high power. Or a long design and high performance and low power. You get the picture.

As a CPU designer, I can design really fast circuitry really quickly, but it usually burns a huge amount of power (pseudo-NMOS, for example), or slow circuitry that uses relatively little power very quickly (static, pass-gate, etc) and the push around latches (which hurts performance). But if you want high-performance, low power circuitry, give me a lot more time to design it.

FWIW, I dispute the general statement that was quoted from the original Anandtech article. It says that it's too hot and uses too much power. IMO, it's only "too hot" if you can't use it in a specific application. So an Athlon currently is too hot to use as a mobile part in a laptop, but it's clearly not too hot to use in a desktop system (since plenty of people are). So saying that this is a problem for the Athlon is, IMO, too general a statement.


Patrick Mahoney
IPF Microprocessor Design
Intel Corp.

 

[m2]

Thanks for the contribution, Patrick,I think I understand the implication of your message. What I was trying to get at was that heat seems to be the biggest limiting factor in determining the speed at which a processor can run, hence the constant discussions on this forum about coolers etc. for overclocking.

So it appears to me as a layman, that in some way. the athlon is an inferior product to the PIII in that it has to be radically cooled to achieve its native clock speed and be stable. A slower badged PIII can also get to the same speed when overclocked and cooled.

The reason I am so interested is that I use music software which pushes the cpu to its limits even when idle ( plug-ins ), I am seriously worried about a heat problem with the Athlons even though I am very attracted by the price.


I see that AMD are taking some development time to improve the athlon core with the Palomino, so maybe everything will be ok soon



Eric

 

[jsbush]

I was using my retial intel heatsink fan that came with my celeron 566 on my duron 600 cus I couldn't wait for my other cooler to get in. It ran stable and relitively cool.

 

[pm]

To me, radical cooling doesn't really start until the processor dissipates more than 100W. Below 100W fans and heatsinks should be able to cool them effectively. I don't think that the Athlon is limited by power dissipation/heat, it's limited by it's design - and the design can be tweaked to go faster up to a point.

To me, "limited by heat" means that you could clock the processor at 2GHz, but then the power would be impossible to dissipate using standard passive cooling (where "passive" means radiative cooling using fans and heatsinks, as opposed to active where there are heat "pumps" which remove heat from one part to redistribute it to another). Once you throw refrigeration into the design, then cost starts to rise rapidly.

Anyway, I guess I can agree that the Athlon is "inferior" to the Pentium III in terms of power dissipation, but I don't think this means that it's an inferior product. It just means that it uses more power.

To me power is another variable in the design, something that you need to take into account, but something pretty far down the list of priorities (performance typically being near the top, and schedule usually following closely behind it, with cost usually interleaved into the top three as well). As long as you don't grossly exceed your power "envelope" with a design then you, as an engineer, should be ok.

I guess to me it doesn't really matter what the power that a CPU uses (within limits) because that's a concern of the design team - not the end-user. As long as AMD says, "this part will run at 900MHz without any problems", and, say, Gateway takes this account, then what does it matter to you, the end-user, whether it uses 1W or 50W? It's been designed to work. About the only concern in my mind is the cost of electricity - and that's not that big a deal usually compared to the monitor.

 

[AdamK47]

<< (where &quot;passive&quot; means radiative cooling using fans and heatsinks, as opposed to active where there are heat &quot;pumps&quot; which remove heat from one part to redistribute it to another). >>



Is that right? I never knew that...
I always thought active cooling was fans and heatsinks, while passive cooling was just the transfer of the heat into a heatsink.

 

[Mykex]

Ok the heat issue of which I do not see. If they are running hotter than the compeerable Intel product do you think it just may be the 200mhz FSB speed on the CPU? Since Intel's P3 is at 133 this is a rather large difference.

BTW non overclocked Tbird750
ambiant case temp 77f
cpu temp 93f

That Aint hot.

----------------
LOL@P4 Ill buy an Abacus

 

[Mikewarrior2]

Mykex,

AMD chips run significantly warmer when using the same heatsink as a comparably clocked pentium 3.

for example, with an FOp38(.37C/W) on a t-bird 1ghz(1.8V)(58W) versus a p3 1ghz(1.8V) (32W), the T-bird WILLrun roughly 8-9C higher than the p3.

Whether or not the socket-a(or slot A) mb registers this difference is an entirely different issue.


Mike

 

[Mikewarrior2]

I Also agree with PM that AMD-chips do not require radical cooling. They do run hot, but a good heatsink will do the job.

My only true wish is for AMD to implement an internal diode so we can eliminate all the BS &quot;reviews&quot; out there.


Mike

 

[pm]

At Intel - which doesn't necessarily mean that it's industry-wide terminology - &quot;passive cooling&quot; refers to radiative cooling (ie. heat comes off of the CPU and is radiated by the heatsink). A fan is an active device, I agree, but &quot;active cooling&quot; is reserved for devices that take heat from one location and move it to another - for example, peltiers, heat pumps, and refrigeration systems. I figured that this might be a source of confusion - heck, it bothered me when I first heard it - so that's why I defined it in my post.


Mikewarrior, you seem to know a lot about cooling technology, what's your opinion of &quot;active&quot; vs. &quot;passive&quot; cooling?

As far as the internal diode... I have no idea. The cons to one is that it &quot;wastes&quot; two pins which increases cost (a little bit anyway), and it's really not useful to 99% of the general public. It's only really is of critical use in a server CPU. Actually, I'd place a bet that they have one on there and just don't let anyone access it. They probably only pin it out on the TAP pin of their test parts. It's not like they are hard to design anyway. If they could put together a CPU like the Athlon, they surely could have designed it in if they wanted to - and I'd bet that they have.

 

[Mikewarrior2]

Hey PM,

To me, passive cooling is heatsinks/fans, the norm as far as cooling. Active to me is peltiers, watercooling, etc.

As far as an internal diode, I have a feeling that you are correct.


Mike

 

[m2]

Sincere thanks once again to everyone for their help.

I am quite happy to accept that Athlons are a viable solution with moderate cooling, but I still don't see that they are any better than overclocked Pentium III's, apart from the superior FP performance.



Eric