I7 920 temperature tested

[nyker96]

HP tested the 920 on H2O//TRUE//Stock coolers under OC conditions.
http://enthusiast.hardocp.com/...wxLCxoZW50aHVzaWFzdA==

from what I can see, the max temp 920 can go to is 90C (after which BSOD). At 3.8 TRUE is doing 83C at 100% load in open case, I think you have to add 4-5C to that in closed case. This means temperature could be the limiting factor in a i7 overclock. At least for air coolers. From the looks of it, 4Ghz will be a max for most air OC of this line of cpus. Still pretty impressive. of course those with water blocks can push the chip more.

 

[flxnimprtmscl]

Jesus tits, that thing is a space heater. It makes Pent D's look good... I don't know if I could be comfortable running a chip at 80-90C.

 

[Drsignguy]

Also to add what was noted, is that from full load to idol, temps dropped in 5 seconds with the H2o and Air and this was a from an open bench. Now if the choice is air-cooled, a closed case will have to be, as with all cases, to have great ventilation to remove that circulating warm air. Water removes it fairly quickly. This is something that all should take note on as if you want great (lower) temps, remove the hot air quick.

Took note on the 80c - 90c temps. If this is any indication of how warm these chips get, it would take some getting used to as I don't see those kind of temps from my yorkie, wolfdale and Kentsfield.

 

[TidusZ]

More reason for me to wait for I7 version 2.0, simply not too impressed so far.

 

[Zap]

Originally posted by: Drsignguy
Took note on the 80c - 90c temps. If this is any indication of how warm these chips get, it would take some getting used to as I don't see those kind of temps from my yorkie, wolfdale and Kentsfield.

My first Prescotts ran that hot regularly.

In my testing with a Core i7 965, a Kill-A-Watt reported about 150W difference between idle and load (Prime95 loading 8x) running around 3.5GHz.

 

[Drsignguy]

Originally posted by: Zap

Originally posted by: Drsignguy
Took note on the 80c - 90c temps. If this is any indication of how warm these chips get, it would take some getting used to as I don't see those kind of temps from my yorkie, wolfdale and Kentsfield.

My first Prescotts ran that hot regularly.

In my testing with a Core i7 965, a Kill-A-Watt reported about 150W difference between idle and load (Prime95 loading 8x) running around 3.5GHz.

Oh and how you are correct. I too have a prescott 3.6 and that thing itself can heat my Garage....

So, how much of a difference is the wattage compared to say.. the yorkie? I haven't ever tested one but just seems like a 150W jump is pretty significant.

 

[Zap]

Originally posted by: Drsignguy
just seems like a 150W jump is pretty significant.

Well, remember that this is "from the wall" at just over 80% efficiency of the PSU, so it is drawing around 120W more, in theory.*

*Some evidence suggests Kill-A-Watts and other measuring devices sometimes are fooled by the PSU's APFC.

 

[dmens]

Originally posted by: nyker96
HP tested the 920 on H2O//TRUE//Stock coolers under OC conditions.
http://enthusiast.hardocp.com/...wxLCxoZW50aHVzaWFzdA==

from what I can see, the max temp 920 can go to is 90C (after which BSOD). At 3.8 TRUE is doing 83C at 100% load in open case, I think you have to add 4-5C to that in closed case. This means temperature could be the limiting factor in a i7 overclock. At least for air coolers. From the looks of it, 4Ghz will be a max for most air OC of this line of cpus. Still pretty impressive. of course those with water blocks can push the chip more.

temperature do not cause crashes. speed limiters do. the temp only moves the limiter out a bit. the speed difference between 0C and 90C on a nehalem is like 250MHz, not all that much at all. also, speed binning is done at 90C which means the in-die temp is even higher.

to be honest the whole thing about people being "not comfortable" about temps is kinda pointless. damage to the cpu is with high volts and cold. hot is just a noisy fan.

 

[dmens]

Originally posted by: TidusZ
More reason for me to wait for I7 version 2.0, simply not too impressed so far.

what, like a westmere? have fun waiting. if you're "not too impressed" by temperature, high temps do not mean high power draw, it means high power density on some localized spot on the die. go read anand's power rundown for some actual numbers:

http://www.anandtech.com/cpuch...howdoc.aspx?i=3453&p=3

oh btw, nehalem has far more accurate temperature measurements than C2D.

 

[error8]

Why does it go so high, on a TRUE and with only 1,35V?? It just doesn't make sense... A TRUE is capable in dissipating efficiently, at least 120-140W of heat. Even on water it reached 80C under load. Something must be wrong there.

 

[Zap]

Originally posted by: error8
Why does it go so high, on a TRUE and with only 1,35V?? It just doesn't make sense... A TRUE is capable in dissipating efficiently, at least 120-140W of heat. Even on water it reached 80C under load. Something must be wrong there.

What's wrong might (or might not) be expectations of the performance of a cooler made for one platform being used on another platform. LGA 1366 is capable of getting hotter than LGA 775. The very first LGA 775 CPUs were 65W and eventually crept up past 100W TDP. The first LGA 1366 chips are already past 100W. Also, Intel made the heatsink mounting area larger than LGA 775, allowing for larger coolers. The "stock" cooler that comes with a Core i7 965 has a fan around 100mm in size (see pics at HardOCP of the two Core i7 retail coolers).

 

[error8]

So these new chips are like the old Pentium Ds, in terms of power consumption, but this time they're not lacking performance.

 

[Nemesis 1]

Originally posted by: Zap

Originally posted by: error8
Why does it go so high, on a TRUE and with only 1,35V?? It just doesn't make sense... A TRUE is capable in dissipating efficiently, at least 120-140W of heat. Even on water it reached 80C under load. Something must be wrong there.

What's wrong might (or might not) be expectations of the performance of a cooler made for one platform being used on another platform. LGA 1366 is capable of getting hotter than LGA 775. The very first LGA 775 CPUs were 65W and eventually crept up past 100W TDP. The first LGA 1366 chips are already past 100W. Also, Intel made the heatsink mounting area larger than LGA 775, allowing for larger coolers. The "stock" cooler that comes with a Core i7 965 has a fan around 100mm in size (see pics at HardOCP of the two Core i7 retail coolers).

Nice to see ya got the thread going the right direction. Because your right the die is alot differant. Also Dmens is correct about the sensor being better.

I have done a bit of testing . With cpu running 8 threads at full load. Your right she can draw some power and build heat You nned to look at the Amount of work being done with those loads and you will quickly find . That IC7 is very efficient. But you guys are very funny lol. Running 4 threads at fullload isn't bad at all . In fact its better than anything out there. Temps are easily control up to 3.6 by stock cooler. You might think thats bad . But in fact its rather good. I KNOW for fact that comparing IC7 to P4P is about as bad as saying ATI can never beet NV in performance. I can't help that AMD failed with K10. and K10.5. . But to say that IC7 shares anything with P4P other than a few design features is laughable. Or is this the problem . THE RETURN OF NETBURST. Alot went into IC7 from P4C . Add in the lessons intel learned from P4P and you have IC7. Ya I know that IC7 comes from c2d type core . So the C2D core shares these things with P4C also.

No matter how ya spin it . Intel has hit a home run with 4 batters in a roll and still no outs.

 

[error8]

What scares me is that during the last years, we've noticed a reduction or at least a stagnation in terms of power consumption, on cpus. But now, with these i7 cores, Intel took power consumption a step higher and that is not a good thing. I really hope that from now on, we're not going to see cpus becoming more power hungry on each new generation, like the gpus have.

 

[Nemesis 1]

Where on earth did you get that from . Its not true. Do you have an IC7 in your hands?

The AMD force is growing strong after 2 years behind the 8 ball.

They have a nice thread over at XS. Comparing the k10.5 to penryn . Its really funny . Because Penryn basicly kicks but. Until you come to server loads . The thread quickly is haulted when server numbers show up for IC7 its a brutal beating by IC7 on AMds K10.5.

Why the thread was suddenly stopped. It was stopped out of kindness just to eas. The pain of those who thought they had a winner.

 

[dmens]

Originally posted by: error8
But now, with these i7 cores, Intel took power consumption a step higher and that is not a good thing.

the only thing thats not good is your lack of reading comprehension skills.

 

[error8]

Originally posted by: dmens

Originally posted by: error8
But now, with these i7 cores, Intel took power consumption a step higher and that is not a good thing.

the only thing thats not good is your lack of reading comprehension skills.

With my lack of reading comprehension skills, I'm able to read that i7 platform sucks more juice then core 2 quad platform. So what is your problem really? Am I missing something?

http://www.tomshardware.com/re...7-Nehalem,2057-10.html

 

[dmens]

Originally posted by: error8
With my lack of reading comprehension skills, I'm able to read that i7 platform sucks more juice then core 2 quad platform. So what is your problem really? Am I missing something?

http://www.tomshardware.com/re...7-Nehalem,2057-10.html

you should read the AT article on power instead of tom's shitastic pile of bs. here is the link again: http://www.anandtech.com/cpuch...howdoc.aspx?i=3453&p=3

tom's numbers are garbage. he doesn't even say how many threads he is running. if he's running a 1 thread prime95, there's no way the nehalem system power can exceed c2q because 3 cores would be sleeping. maybe its 4 or 8. cant run 8 on a c2q. is it 4? 4 single thread nehalem cores would be well below a q9770 from everything ive seen.

tom's has no credibility left after making all that noise on overclocking the 920/940 nehalems when intel released a detailed memo telling them exactly how it was a non-issue.

 

[error8]

Didn't know about that. I only read what Tomshardware had written there and I thought that was the general idea. My bad then.