3.8GHz Prescott throttles back all the time with very little headroom.

[Zebo]

3.8 or 2.8 Ghz?

http://www.sudhian.com/showdoc...m?aid=630&pid=2382

 

[THUGSROOK]

LOL ~ isnt that a disgrace?
even for non OCers ~ its a disgrace.

what a freakin Edsel

 

[BonzaiDuck]

Again, that's why the two industry giants are looking at dual-core processors in the next go-around.

Even so, this surprises me a bit. The LGA775 socket design DID reduce some of the thermal problems.

Back to the drawing board!

 

[Zebo]

I'd like to see these reviews reviewing inside a case envrioment (read much hotter) than on thier wide open test beds.

 

[THUGSROOK]

intel has "other" very powerful chips they could release for desktop platforms.
i wonder why intel is ignoring them?

 

[Zebo]

Originally posted by: THUGSROOK
LOL ~ isnt that a disgrace?
even for non OCers ~ its a disgrace.

what a freakin Edsel

to be fair 90nm things aint so rosy as AMD fans like to pretend after about 2.8...

http://www.overclockers.com/articles1151/

RE: IBM's 90nm process which AMD is using

is indeed a very cool running chip up to about 2GHz. It chews up only 40-50 watts at that speed (and needs only 1-1.1V to do so).

Make it run at 2.5GHz, though, and you find that this processor now needs 100 watts at 1.3V. Quite a difference, isn't it?

Why does this happen? Read a bit more, and look at the first chart, and you'll find that once you push the chip past a certain point, it starts leaking power, a lot of power. Maybe not quite as much as Prescott, but still enough to qualify it as a relative.

As IBM puts it:

"This total power is very sensitive to voltage. Each power component has a steep response to voltage, for example, ac switching power is proportional to V2, whereas the subthreshold power is proportional to V3 and gate leakage power has an exponential (our emphasis) relationship to voltage."

 

[Zebo]

Originally posted by: THUGSROOK
intel has "other" very powerful chips they could release for desktop platforms.
i wonder why intel is ignoring them?

I think they were or still are run by the marketing department instead of engineering and R&D. Mhz sells simple as that. When netburst first started I heard at least one well respected chip designer for them quit in discust. I think the Dothan has higher IPC than even A64 but they sure don't seem in a hurry with it do they?

 

[klah]

Originally posted by: Zebo

Originally posted by: THUGSROOK
LOL ~ isnt that a disgrace?
even for non OCers ~ its a disgrace.

what a freakin Edsel

to be fair 90nm things aint so rosy as AMD fans like to pretend after about 2.8...

The last roadmap here didn't show anything faster than 2.6GHz planned until Q3 '05 anyway.

 

[clarkey01]

"When netburst first started I heard at least one well respected chip designer for them quit in discust."


Zebo, thats Bob Combwell (spelling), he gave a talk to some students which is online somewhere where he goes in depth why he thinks Intel is heading in the wrong direction, he had an argument with Andy Grove over netburst and he then quit IIRC.

 

[DrMrLordX]

Originally posted by: klah
The last roadmap here didn't show anything faster than 2.6GHz planned until Q3 '05 anyway.

Which is a shame, really . . .

 

[clarkey01]

Q3 05 ! . No I think with improved yields, Strained silicon they will get 2.8 Ghz out of the 90 nm process by Q2 , it might run a little hot but aslong as it's not prescott hot it should be ok.

FAB 30 has been known to pull of some minor miracles, lets hope it can do a few more.

 

[BonzaiDuck]

This wasn't the only thing published recently about the thermal shortcomings of the top-end Intel chip.

THG noted for the 3.6E or EE that this same throttling problem occurred in tests with the stock INtel cooler, and that buyers should simply skip even the notiion of using it -- that they should opt immediately for more exotic cooling solutions, if only copper heatsinks, heatpipes and better fans.

Someone else remarked about case-design test-beds. I'm not sure that the test-beds used in these reviews were BTX form-factor; I am willing to bet they were ATX i925-chipset motherboards. But I don't recall what was done for the THG review, and at this late hour, someone else with greater enthusiasm might enlighten me as to which motherboard was used, which heatsink-fan, or which case -- if any.

Anyway -- no corporate advocate I -- but the marketing department is generally responding to consumer preference. We wouldn't be looking at the advent of SLI and dual graphics adapters if it weren't for a consumer preference for gaming; there would not have been MMX instruction-set enhancements but for consumer multimedia preferences.

And also -- anyway -- this comes as no surprise. Last spring announced INtel's recognitiion of a "thermal limit", followed by a discussion about "dual-core" processors to circumvent the problem. Moore's Law notwithstanding -- all technologies bump up against obstacles. What will the Great Petroleum Culture of the Modern Age do when demand for oil outstrips supply? We WANT a paradigm-shift in high-end computer processors; we NEED a paradigm shift in our transportation technology.

 

[AnandThenMan]

Originally posted by: Zebo

to be fair 90nm things aint so rosy as AMD fans like to pretend after about 2.8...

http://www.overclockers.com/articles1151/

Overclockers.com is full of crap. They often go out of their way to diss AMD, this is no exception.

I am not pro AMD anymore than I am pro Intel, but I am seriously dissapointed with Prescott.

 

[LTC8K6]

Other reviewers such as XBIT had their 3.8's running at 4.2 on air with no issues, and made no mention of any clock throttling which would have shown up in the benches.

Firingsquad also reached 4.2 stable and no mention of clock throttling.

 

[Jhruska]

That's not necessarily true.

First, Prescott has a wider "acceptable" thermal range than any other Intel CPU I'm aware of. I've seen two retail processors tested in the same motherboard at the same time turn in full-load temperatures 5'C apart, even after adjusting and remounting the CPU's repeatedly.

Second, I *can* stop the 3.8 from throttling. All I have to do is keep my room cold enough. Using an aftermarket non OEM cooler would probably also do the trick.

Third, It is possible to override some of Intel's throttling. You can't override the automatic shutdown feature, but manufacturers can prevent throttling from occuring at certain points via BIOS hooks. You aren't supposed to--but you can.

The presence of confounding variables and multiple different test conditions is why FiringSquad and XBIT can do their tests and report valid data--and I can as well. Because I test using two instances Folding @ Home, I can compute the amount of time reported to process one "frame" of data. This length of time is a near-constant in any given work unit; it may vary from 5:21 per frame to, say, 5:24 per frame, but only major events will shift it outside a 3-5 second gap. When I see my per-frame calculation time jumping by an amount that coresponds to the amount of throttling reportedly occuring, that's a clear indication that throttling *is* occuring.

 

[LTC8K6]

What's not true?

 

[gururu]

Originally posted by: LTC8K6
Other reviewers such as XBIT had their 3.8's running at 4.2 on air with no issues, and made no mention of any clock throttling which would have shown up in the benches.

Firingsquad also reached 4.2 stable and no mention of clock throttling.

EXACTLY. but mentioning that in this thread would be akin to crapping in it, since its a 'Intel sucks' thread.

 

[Duvie]

Actuallyu I took some of the reports as saying that these prescotts have little or no headroom with stock coolers....Get an exotic cooler and yo8u a have some room...

The probelm is and most know I have never liked the prescott route, is that when you sell a stock chip you shouldn't be on the fringe like that....Think of the ppl who buy this and have no clue about adequate case cooling. they will be throttling left and right and the machine should feel lethargic....

I think if it :

1) can't be adequately cooled by stock cooler
2) assume ppl will buy 100 dollar case setups with 4-5 fans
3) show lttle improvement over a northwood cores to date and yet already about tapped out for speed....


It shouldn't have been made, period....

The only reason I got an A64 is I have no confidence in the direction of Intel right now and in the immediate future. I think I may have a couple of upgrades on this or a nforce 4 mobo before Intel may have something worth getting again...

 

[Jhruska]

Gururu,

I cant' speak for anyone else, but its not an "Intel Sucks" thread from my viewpoint. Saying that a chip or chip speed grade has an issue where it runs too close to the brink of throttling is different than going to OMFGINTELSUXX0RS AMD FOR TEH WIN route.

 

[gururu]

Originally posted by: Jhruska
Gururu,

I cant' speak for anyone else, but its not an "Intel Sucks" thread from my viewpoint. Saying that a chip or chip speed grade has an issue where it runs too close to the brink of throttling is different than going to OMFGINTELSUXX0RS AMD FOR TEH WIN route.

come on now, you have an OP with an 'A64 overclocking guide' in his sig, making a thread about an Intel product with the comment "HOT HOT HOT".

you're trying to tell me this isn't bashing Intel. Plus, there have been plenty of threads on E0 stepping cpus but because most of them were actually informative, they wouldn't get this particular message across?

 

[Duvie]

Originally posted by: gururu

Originally posted by: Jhruska
Gururu,

I cant' speak for anyone else, but its not an "Intel Sucks" thread from my viewpoint. Saying that a chip or chip speed grade has an issue where it runs too close to the brink of throttling is different than going to OMFGINTELSUXX0RS AMD FOR TEH WIN route.

come on now, you have an OP with an 'A64 overclocking guide' in his sig, making a thread about an Intel product with the comment "HOT HOT HOT".

you're trying to tell me this isn't bashing Intel. Plus, there have been plenty of threads on E0 stepping cpus but because most of them were actually informative, they wouldn't get this particular message across?

You show me a link where when they tested the cpu they used stock cooller to obtain much of anything past stcok 3.8ghz??? the fact is we know most of these sites in their test labs place them outo a table surface...get this thing in a closed case environment with vid cards, and optical dirves producing heat and tell me this wont be a problem for any non enthusiast who may be suckered into this cause itis 3.8GHZ!!!!! That person likely wont know about the heat issue and assumes everything is A OK!!!

 

[gururu]

Originally posted by: Duvie
You show me a link where when they tested the cpu they used stock cooller to obtain much of anything past stcok 3.8ghz??? the fact is we know most of these sites in their test labs place them outo a table surface...get this thing in a closed case environment with vid cards, and optical dirves producing heat and tell me this wont be a problem for any non enthusiast who may be suckered into this cause itis 3.8GHZ!!!!! That person likely wont know about the heat issue and assumes everything is A OK!!!

I'm not saying that the retail package is an overclockers dream. But IMO, people with stock sinks shouldn't be overclocking anyway. What concerns me, is that this Sudhian review is a total of 5 pages and 2 experiments and now you and others claim that the cpu is an 'edsel'?!!!! Compare this to the far more complete x-bit review of the 570J. They ran it through a battery of tests and included overclocking results and made solid conclusions based on their observations. Sudhian, again, could only speculate.

 

[Duvie]

Originally posted by: gururu

Originally posted by: Duvie
You show me a link where when they tested the cpu they used stock cooller to obtain much of anything past stcok 3.8ghz??? the fact is we know most of these sites in their test labs place them outo a table surface...get this thing in a closed case environment with vid cards, and optical dirves producing heat and tell me this wont be a problem for any non enthusiast who may be suckered into this cause itis 3.8GHZ!!!!! That person likely wont know about the heat issue and assumes everything is A OK!!!

I'm not saying that the retail package is an overclockers dream. But IMO, people with stock sinks shouldn't be overclocking anyway. What concerns me, is that this Sudhian review is a total of 5 pages and 2 experiments and now you and others claim that the cpu is an 'edsel'?!!!! Compare this to the far more complete x-bit review of the 570J. They ran it through a battery of tests and included overclocking results and made solid conclusions based on their observations. Sudhian, again, could only speculate.

Did yXbit use the stock cooler??? I think that is the point...Think of it not as overclockers but a person buying a prescott for the first tim ecause they have some money and "want the fastest Intel chip"...They likely wont OC, but to have it throttle at load at stock speed is ridiculous....

The prescott is for the serious user I agree but I don't think it says that on the package for minimum requirements!!!

 

[gururu]

Originally posted by: Duvie
Did yXbit use the stock cooler??? I think that is the point...Think of it not as overclockers but a person buying a prescott for the first tim ecause they have some money and "want the fastest Intel chip"...They likely wont OC, but to have it throttle at load at stock speed is ridiculous....

The prescott is for the serious user I agree but I don't think it says that on the package for minimum requirements!!!

I don't know. I know that for overclocking, they explicitly said they attached a Zalman. But they ran their tests at default, so I don't know if for those whether they used retail or not. The truth is, I can't stand behind the P4 E0s because I wouldn't buy one for the price they are asking. But going back to why I originally posted, this IMO has all the makings of an Intel bashing thread. Which is good and all, but it's relying on a WEAK review that didn't really address any real world problems that the throttling would influence. Are we going to get frame skipping during games, are movies going to all of a sudden drop 10 frames, etc. Yet, they SUGGEST that this is the reality. Good reviews report observations, they don't create them. So I am doubtful that the throttling is a big deal.

 

[Duvie]

I didn' t take this as bashing...maybe goating but that is hard since amd on so many levels has had a much more successful 90nm transformation...any news by whomever posting it will not shine as bright for INtel....

The fact is...that was not a stock cooler in that review and most test labs dont put in case where it is subjetct to higher heat from other components..also most test sites probably are done in nice AC'd rooms. Not likely th eaverage for most ppl....

The problem is little or no headroom...heck the 3.4c bnorthwood at the time we just switched to prescotss had more headroom yt intel seems fit not to pres on with that line....