Is it true that P4 775 Prescott CPU's run so hot they require an intake fan near the CPU to stay cool?!! (BTX-standard)

[Viditor]

OK, my turn to stick my neck out and chime in...

1. Prescotts are NOT too hot IF you are getting a 3.2 GHz and slower. At those clockspeeds (especially the 2.8), they are just fine for heat, though it's true that the A64 is substantially cooler.

2. NX bit is only important if we get a new virus/worm like the Blaster circulating. In that case it could be invaluable... Intel will be releasing their version of NX bit P4s in January.

3. Win XP will be upgradeable to Win XP-64 in the next few months at no cost to many (though a small charge to some). It is PREDICTED that this will allow for a 20% increase in performance, though we have yet to see the proof of this as many of the drivers aren't ready yet.

4. I can't imagine that you'll notice the difference in speed between the two. In benches, the P4 is slightly faster than the A64 for video, and in everything else the A64 is slightly faster than the P4, but again, neither will be so much that you would really notice...

 

[Viditor]

Actually when Prescot first came out, Dell sent their stock back to Intel and cancelled all orders in favor of northwoods

Actually, both Dell and HP cancelled many Prescott models for 2 months. However, they did NOT send their stock back, and they stated it was due to not being able to receive enough chips...

 

[Audiophile1980]

2. NX bit is only important if we get a new virus/worm like the Blaster circulating. In that case it could be invaluable... Intel will be releasing their version of NX bit P4s in January.

There are several buffer overflow exploits which are unpatched except by the NX bit. Surf some seedy porn areas for a few days and you'll see what I mean.

It is PREDICTED that this will allow for a 20% increase in performance, though we have yet to see the proof of this as many of the drivers aren't ready yet.

There have been a lot of benchmarks of the latest BETA which show a 7% performance boost on 32bit apps and promising results on 64bit apps. If that's the kind of performance attained with the BETA and BETA drivers, I think 10% is a fairly modest estimation. Reglardless of whether it's 7%, 10%, or 2000%, it's an improvement which neither S478 northwood or prescot will realize which was rather the point.

However, they did NOT send their stock back, and they stated it was due to not being able to receive enough chips...

I can't dispute the stock return as I can't find the news coverage ATM but I do remember reading that. Regardless, Dell is a close Intel partner and had to explain the conspicuous absense of Prescots in their line. Their explanation was obviously a cover as supply was decent enough for us white box builders and #1 buddy Dell always gets the chips first. In fact, that's part of their deal with regards to hot top speed chip supplies...uh, pun not intended.

 

[LTC8K6]

In March Dell wasn't selling Prescotts? So what if that's true?

That article is wrong all over the place including what it says about AMD, imo.

PCI-E, DDR2, and SATA are all here now, just a few months later.

 

[LTC8K6]

BTW, both our CPU Database and other reports from the field indicate that overclocked PressHots are only yielding 3.5-3.7GHz using high-end air, and blowing a lot of hot air in the process. This is barely better than what recent Northwoods are doing.

Yep, Prescotts were so much hotter than Northwoods, that they were overclocking better than Northwoods on air.

That part of the article supports what I have said all along, Audiophile. Prescotts are not that much hotter than Northwoods. Not enough to worry about. Nothing that a good HSF can't handle.

 

[Wingznut]

Originally posted by: Audiophile1980

However, they did NOT send their stock back, and they stated it was due to not being able to receive enough chips...

I can't dispute the stock return as I can't find the news coverage ATM but I do remember reading that. Regardless, Dell is a close Intel partner and had to explain the conspicuous absense of Prescots in their line. Their explanation was obviously a cover as supply was decent enough for us white box builders and #1 buddy Dell always gets the chips first. In fact, that's part of their deal with regards to hot top speed chip supplies...uh, pun not intended.

You having a few OEM chips is VERY much different than Dell having enough in stock. You have no idea how important supply is to Dell.

 

[JohnAn2112]

Originally posted by: LTC8K6
Yep, they're so hot that everyone is having trouble with cooling them. No one can even get them to run without clock-throttling unless they use water cooling. The stock heat sink and fan are totally useless.

So many people are having trouble that Intel can't even sell the chips anymore. In fact, people are sending them back because a standard heatsink won't keep a Prescott cool enough to run for more than 10 minutes.

Funny how my 2.8 Prescott is running stable at 3.2 with the stock heatsink and fan. :roll:

 

[LTC8K6]

Funny how my 2.8 Prescott is running stable at 3.2 with the stock heatsink and fan.

Ummmmm.....Yeah I know, that's why I made up all the ridiculous stuff in that post.

How could you possibly read this thread and think that post was serious?

 

[JohnAn2112]

Originally posted by: LTC8K6

Funny how my 2.8 Prescott is running stable at 3.2 with the stock heatsink and fan.

Ummmmm.....Yeah I know, that's why I made up all the ridiculous stuff in that post.

How could you possibly read this thread and think that post was serious?

Doh! Speed reading at work = bad. LOL

 

[Viditor]

There are several buffer overflow exploits which are unpatched except by the NX bit

I agree. But most of them aren't devastating like Blaster was...

Dell is a close Intel partner and had to explain the conspicuous absense of Prescots in their line. Their explanation was obviously a cover as supply was decent enough for us white box builders and #1 buddy Dell always gets the chips first

They didn't cease ALL Prescotts, just those over 3.0 GHz...
White Box and Dell both received the higher binning Prescotts, but Dell (and HP) has a MUCH larger minimum requirement. That means that if they don't receive X number of chips, they can't release the product. The WB OEMs have no such limit...

BTW, I do agree with you that AMD has the superior product across all lines right now, and I see no change to that possible for at least the next year...
In fact, I believe that so strongly that as of a few weeks ago, over 60% of my stock portfolio is invested in AMD. That said, it is VERY important that I "keep it real"...Intel is not a company that you should ever underestimate!

 

[Audiophile1980]

That's right. I have great hopes for the dual core pentium-Ms with massive caches.
Those are currently looking slightly more efficient than some hammers per clock and they're already at 2Ghz. All they need to do is ad the NX bit and x86-64 (with real memory addressing unlike Nocona, prescot) and they have a very competitive chip....with AMD's chips today. Hopefully AMD will stand completely still for them

 

[Audiophile1980]

BTW, I was reading something that one of the designers of the Intanium 2 wrote. With regards to an on-die memory controller, he had said that he would like to use one at some point in time but for non-technical reasons it wasn't likely to happen.

Of course, Intel's brain fart BTX is horrible because it forces the motherboard to have the processor in a certain place and the ram in a certain place etc. This means that you can't have an on-die memory controller because the memory is too far away from the CPU. Stupidity. Form factors are there to enclose technology, not strangle it.

Now, I'm thinking about dual core prescots and how they're so heavily cache/fsb/ram speed dependant because of their heavy, heavy, branch mis-predict penalty and how dual core means these things are going to be choking for bandwidth.

Much in the same way that 4way Opterons will always widen their performance lead signifigantly over the starving 4way Xeon systems. This would seem to apply to dual core as well.

Now, I realize that the Pentium M is far less bandwidth dependant much like the K7/K8 but the dual core opterons do have the advantage of the hypertransport link between cores. The hammer was designed for multi-core from day 1.

I would speculate that Intel must be preparing some kind of answer to this because I don't see how they can pack dual cores without it.