Prescott in the bat cave computer

[Ross]

For all intents and purposes, this is absolutely the end of the world on this 2.8E. It would need to be 30°F in this room to make a solid 4GHz on water....oh man, so close, but 286FSB wasn't happening no matter what I tried....and I tried everything An extra .250Vc probably would've done it, but that just can't be done....I am telling ya, this proc or this proc in this mobo (since AI7s don't seem to have the same problem?) took a self-defense class...hehe. 1.625+Vc = automatic no boot.

For the monumental effort it took to get this from ~3850, I probably could've gotten 3.8 out of the 2.6C...tomorrow's goal

I was lucky it even loaded the progs without crashing, so obviously no Sandra CPU benches, but Super Pi 1M was not much quicker than at 3850 though....34 seconds.

I am frozen solid, so now I have to go thaw out

 

[lbbo2002]

Almost 4gig your so close Ross but a great effort.

Is everyone hitting a similar wall of only 100-150 mghz over where they have to raise the volts to max?

Hoping Big Nuttz cracked the 4gig stable that would be one at least

So which motherboard don't boot over 1.6 volts on the CPU?

 

[Ross]

Yeah, it's killing me! It's supposed to be in the mid-20s tonight, so I am going to risk pneumonia and get that damn extra 14MHz (maybe). It seriously, seriously was an effort for that, right down to unplugging all fans on mobo headers to free up drain on mobo traces and hanging a fan in front of the CPU/PSU.

Vc (and all voltages) seemed a little more stable (bouncing from 1.50x-1.52x instead of 1.48x-1.50x at idle in BIOS) and I don't know if it was because I unplugged the fans from the mobo or because the PSU or mosfets were running a little cooler because of the fan over them.

For sure this 2.8E in my P4C800-E mobo will absolutely not even boot if I set the Vc over 1.600. I've tried it even down at stock FSB. At the very least, I will need to reboot twice and if it gets back into BIOS like that, the Vc is magically back running at 1.344ish still with a 1.625Vc BIOS setting.

This 1.600Vc lockout is just killing me. It wouldn't be so bad if the ASUS mobo overvolted like usual and it was 1.625ish actual at 1.600 setting, but the Prescott just sucks everything dry. Maybe one of the drawbacks to the independent rails on the Antec PSUs is that when one is being pulled on that hard, it has no where to get reserves from?

Big Nuttz seems like his stuff is together....1.42Vc and already at 3850+. He maybe one of the lucky ones that can reach 4250...

 

[flapperhead]

the power and heat issue has got to be resolved b4 we see anything near 4.5-5 ghz, not to mention the cooking of your mb.im pretty dissapointed with this cpu. oh well, i think im gonna build another enclosure around my lian li and pump alot of outside cold air to the beast b4 i mess with my 2.8e.the outside air is pretty cold this time of year i think it'll help alot not only with the cpu but also with all the mb components. last year when i did this it got 10-15c cooler on the proc. and my nb got as low as 10-15 c. maybe that will help with stability.

 

[TC]

Reporting in with my 2.8 and AI7. I'm on water (koolance unit) and mine tops out at 3800 with HT off, or 3700 with HT on. It posts at 3900, but will not load windows. This is with a vcore of 1.55 actual 1.6 something bios setting. Temps under load are about 53C which doesn't seem bad considering the FX51 I had ran about the same with water. This is according to abit's uguru thing, so take it for what it's worth. What bothers me is the PWM temp hovering between 87-95C.

Since everyone else seems to have run the usual sandra and 3dmark benches I'm adding Seti. It does a single unit in 1:36 with HT off, or 2 work units every 2 hrs 5 minutes with HT on. Compare that with my FX51 that did a single unit in 1:29 at 2600MHz, but couldn't do as many per day without HT.

 

[micamica1217]

What bothers me is the PWM temp hovering between 87-95C.

I see dead mobos....man, that's some scary temp readings.
I hope that's a wrong reading, or I would start praying that your mobo doesn't die on you.

mica

 

[Tedinde]

Put a fan blowing on the board it will drop 30Celsius!!!!

 

[althes]

This is getting more fun by the minute.

 

[batboy]

I've been thinking about mosfet cooling and have decided to install mosfet heatsinks on one of my IC7 mobos and also on my AI7 mobo. Was looking at my collection of small heatsinks and measuring stuff to see what I could cut down. Then I wondered what was already available to buy. Check this out. Little heatsinks from Microcool.

http://www.microcool.it/index1.asp?ris=1024&lin=in

http://www.sidewindercomputers.com/michcokit.html

http://www.frozencpu.com/cgi-bin/frozencpu/vid-25.html?id=mT9ypmUg

http://www.frozencpu.com/cgi-bin/frozencpu/vid-24.html?id=mT9ypmUg

http://www.frozencpu.com/cgi-bin/frozencpu/vid-23.html?id=mT9ypmUg

 

[krag]

If you do get the Micro-cool make sure you strip off the factory TIM and use something better. I bought a package of them and installed them on my IC7, but I didn't notice any benifit. I pressed them on each chip (pll, memory voltage chip, mosfets) as hard as I could and thought they would make good contact. It wasnt' until 3 months later, when I took them off and looked at the underside for the perfect square impression...that I was surprised to see very imperfect impressions. Such as 1/2 to 1/4 square impression in the factory TIM.

All though, since the Prescott is raising the mosfet temps up to 90c, the Micro-cools might give some benefit. I don't think my IC7 really needed them.

 

[batboy]

I generally use Arctic Silver epoxy when I do stuff like this. Now thinking that maybe I'll cut down some of my extra RAM sinks for the IC7 mosfets and buy some Microcool sinks for my AI7 mosfets. Should be interesting to see if it helps much with the Prescott.

 

[JCLW]

Thanks to all those pioneers who bought early Prescotts.

Would someone mind running clibench on a 2.8C and 2.8E to see the difference?

http://glibench.sourceforge.net/pub/clibench-0.7.15.zip

Test menu -> Select all CPU tests,
Test menu -> Run selected tests.

On the test configuration you can pick the number of threads to run (results for both 1 and 2 would be great). You don't have to fill anything else in unless you're going to submit the result.

It's an old benchmark but I still find the results interesting.

From clibench homepage:

CliBench is a suite of synthetic benchmarks

It includes the following benchmarks. They are partially self developed or rewritten industry standard benchmarks.

1. Dhrystones: An industry standard benchmark which is an indicator for integer performance
2. Whetstones An industry standard benchmark which is an indicator for floating point performance
3. Eight queens problem: A famous algorithm that depends on the latency time of the CPU. (The longer the pipeline, the bigger the latency time)
4. Matrix operations: By calculating matrices you'll get a sign for the CPUs level 1 performance
5. Number Crunching: Raw integer performance
6. Floating point: Raw floating point performance
7. Memory throughput: This shows the transfer speed from the CPU into the memory

 

[Chaser]

good info man, thanks

 

[blade901]

Just picked up my 2.8E today (Replacing my 2.6C@ 3.3 paired with a Thermalright SK6+ with a Panaflow fan). Running it with an IC7 Version 1.0 with bios 21. This chip is HOT. Running idle at 56 degrees C and high 60's at load. Living near the Equator doesn't really help either.

System is in a full aluminium casing with 2 front intake fans and 1 rear outlet fan. Not looking to overclock it as I'm planning to run a silent PC with a Zalman CNPS7000A-AlCu but with this kind of temp, the 2.8E is gonna go back. I'll pick up the 3.0C instead. Sad to lose the extra 512kb cache but I'll just have to do without it.

Just my 2 cents.

 

[batboy]

Hey, I got my new AI7 motherboard paired up with the 2.8E Prescott (Thermalright SLK-947U air cooled) and it's running now. Loading on the programs now. I'll keep everyone updated on how it does with this rig.

 

[batboy]

Temps for air cooled Prescott 2.8 overclocked to 3.4 gig on an Abit AI7.

Idle: CPU=39, system=30, PWM=29

Load: CPU=51, system=33, PWM=48

Vcore is at default of 1.3875v listed in the BIOS and measured at 1.4v in Abit Eq.

CPU load temp is about the same as on my watercooled IC7. Northbridge cooling is stock, so it's higher than my modded IC7. Dang if the PWM got hot though. The warm air blowing off the CPU heatsink must heat up both the N/B and power mosfets. I guess some cooling mods will need to be done if I want to leave the Prescott on the AI7.

 

[NiTCOM]

Ok. here is load temps for the Prescott 2.8 overclocked to 3.4 gig.

CPU=52, system=19, PWM=38.5

Vcore is still at default of 1.3875v listed in the BIOS and measured at 1.36v in HW Doc.

It wasn't stable at 3.6 gig with default vcore... but I just remembered I still had the PC4000 RAM in using the 1:1 ratio (I couldn't run those setting with the Northwood 2.8C either). So I'll go back to 3.4 gig and run all the benchmarks before swapping in the Mushkin PC3500 level 2 RAM at the 5:4 ratio.

Where did you guys purchased the prescotts ????

 

[batboy]

Got mine from newegg.com. If you are looking for one, perhaps you should PM me.

 

[Vio1]

so how are the space heaters? Havent heard from Prescott owners for a while.... hype died down?

 

[Ross]

I got mine from Newegg too.

Vio1--
Once they're OC'd, they're not as bad as everyone makes them out to be in terms of performance. They do run hot, but that doesn't really seem to effect them in any negative way. I don't think anyone has had one long enough to see if the mosfets on the mobos are going to have a problem.

All this of course, based on my OC'd 89W 2.8E...might want to hear from the very few that have 103W 3.2Es I have personally pulled mine since it's not worth the heat/power consumption over my NW, but I definitely plan to try a D0 step when they are released...