Intel Prescott To Dissipate 100W+ Of Heat
SimGuy
Ottawa, Canada
Intel Prescott To Dissipate 100W+ Of Heat
Break out the LN2 cooling...
A Japanese web-site revealed some information about heat-dissipation of Intel Prescott processors due to come this November at a $637 USD price-point. The chip will be extremely hot and hungry and in general this is a reason why it will not be compatible with the vast majority of i865 and i875P-based mainboards that were meant to be compliant with the Prescott.
Despite of the fact that the code-named Prescott will be made using 90nm fabrication process using Intel Strained Silicon technology, the babe will be very hot and will eat loads of electricity. According to this report, the Prescott 3.60GHz processor’s Thermal Design Power will be about 103W, 15% more from initial expectations of 89W. A version of Prescott processor for DTR notebooks will typically dissipate about 94W heat.
In order to meet such requirements Intel also had to increase the current for such CPUs, and that is an explanation why the most of current mainboards will not be able to work with Intel Prescott processors. For instance, the original Prescott 3.60GHz was intended to work with 78A IcccMax, whereas the real version will only be able to function with current increased to 91A.
The TDP of the Tejas processor manufactured using 90nm technology for Socket T platforms will be well above 100W; keeping in mind that Tejas is expected to achieve 4.0GHz milestone, we can conclude that massive cooling-solutions will be required for cooling-down such monsters.
Source: X-Bit Labs.com
Break out the LN2 cooling...
A Japanese web-site revealed some information about heat-dissipation of Intel Prescott processors due to come this November at a $637 USD price-point. The chip will be extremely hot and hungry and in general this is a reason why it will not be compatible with the vast majority of i865 and i875P-based mainboards that were meant to be compliant with the Prescott.
Despite of the fact that the code-named Prescott will be made using 90nm fabrication process using Intel Strained Silicon technology, the babe will be very hot and will eat loads of electricity. According to this report, the Prescott 3.60GHz processor’s Thermal Design Power will be about 103W, 15% more from initial expectations of 89W. A version of Prescott processor for DTR notebooks will typically dissipate about 94W heat.
In order to meet such requirements Intel also had to increase the current for such CPUs, and that is an explanation why the most of current mainboards will not be able to work with Intel Prescott processors. For instance, the original Prescott 3.60GHz was intended to work with 78A IcccMax, whereas the real version will only be able to function with current increased to 91A.
The TDP of the Tejas processor manufactured using 90nm technology for Socket T platforms will be well above 100W; keeping in mind that Tejas is expected to achieve 4.0GHz milestone, we can conclude that massive cooling-solutions will be required for cooling-down such monsters.
Source: X-Bit Labs.com
0
Comments
after throttling down to 1ghz.....
Intel hasn't gone to SOI yet have they? I think that should drop temps.
Compare the Opteron with a Prescott instead.
Besides, let's wait until Athlon64 has been available for a few months or even a year and see where things are. It will take a while for software to be optimized for the Athlon64 just like it took a while for software to be optimized for the p4.
I highly doubt the early benchmarks of the Athlon64 will be indicative of the performance once optimizations have been made. I also am very dubious that the Prescott chip will be benching as high as it's capable of in the early stages of its release. Give them both some time and then we can talk, with a bit firmer ground to stand on, about heat dissipation vs GHz vs work done.
I agree with one thing though and that is with 100 watts of heat, the normal Svensson who buys a Dell or a HP with a system temp of 40C (the new Pavillions is that hot with a 3.0 P4), what kind of temps would 100W give them?
Seems more like 50mhz after watching that video.....
NS
Both Intel and AMD CPUs are "kicking directly." It's always the software that has to catch up to the new features of the processor, but that's an issue for both brands.
Despite numerous architectural changes Intel has done on the same series of processors, Intel's major performance boosts come from the core logic they produce, not the processors. Considering Intel's track record, I can't seriously believe that any of you are putting stock in the "Bigger, better, faster!" mantra from the Intel camp. It's been heralded before, and not much exercised.
Speaking in a matter of reality, Intel's very own Pentium 3 could school the P4 if it had the clock-speeds to do so. The leap from the Wiliamette to the Northwood garnered a nice performance increase, but at the cost of higher prices, a socket change, a heatsink upgrade and a performance jump smaller than predicted.
Now let's apply a similar template to the Prescott:
1. Socket change? Check
2. Heatsink upgrade? Has to be.
3. Price jump? Check.
The only piece of the proverbial puzzle is the "Smaller than predicted" performance boost from the 90nm process. Do not discount Intel's ability to concoct grandiose theory, and royally **** the execution thereof.
Clocking in at 2.2-2.6 like the Athlon64 is expected to do, the Prescott better make room to place its tail between its legs.
Pentium 4 3.0C (800 FSB) -> $392.00 USD
Where's the value now?
Apparently if you rearrange some letters, you can get the word "Value" out of my statements!
Let's clarify this:
1. I was using the Opteron to illustrate the very small rift between the new hammer architecture and the P4s.
2. The Athlon64 is supposed to be cheaper.
3. The Athlon64 is supposed to be faster than the Opteron.
And so despite this little bump in the road, everything I just said is still true.
I'll take two, please.
And where are the prescotts?
I simply gave a startingly accurate observation of the past and extrapolated it to future endeavors relevant to this topic.
Take it for what you will.
The 3200 Barton is about the same in performance as the P4 3.0c right? Not sure about the prices over there but the P4 3.0C is 50 bucks cheaper here. I cant see any value in that even if that wasnt what we where talking about. How would the Athlon 64 be any cheaper then? I just cant see it.
Take it for what you will.
Xeon 3.06 512K 533mhz $550 & up
Xeon 3.06ghz 512K L2 1MB L3 $700 & up
In the server markets CPU price is a minor consideration...
======================================
Pentium 4 3.0C ($392) & Abit IC7-G ($175) = $567
Athlon 3200+ 2.2ghz ($449 ) & Abit NF7-S v2.0 ($114) = $563
Look like a wash to me....(overclocking considerations excluded)
And that clock speed difference is Glaring.....
That should be real good on insulation lifetime.
If the CPU is buning 100W then the power section will be burning about the same.
This could make my oc'ed TBird look easy to cool.