i7 920 high temps under load

H

hawk82

[H]ard|Gawd
Joined
Oct 2, 2001
Messages
1,473
I just built out a new system with a Core i7 920 cpu with stock Intel heatsink and am putting the system through stress testing. I just installed Vista Business 64bit for test purposes. I am not overclocking and don't really plan on that in the future. I have noticed that the cpu temps are at 189˚F under full load using Prime95, and roughly 100-110˚F under minimal load in Vista Business (no apps running).

Is that a normal temperature under load for this cpu? I am a bit worried by those temps, so I took the heatsink off, cleaned the cpu and heatsink, re-applied thermal grease, and put it back on. I am still getting the same temps after doing this. The computer has not locked up at all however. Also, I can surf the web, and do other minor things while the torture test Prime95 is running.

Also, I'm not sure if the Intel heatsink is on tight. I really dislike the plastic screws on these stock Intel heatsinks. I am planning on replacing the stock with a different one that uses bolts and nuts to hold the heatsink on solid. Not sure what to get however.

Full specs:
Intel Core i7 920
Stock Intel heatsink/fan
Asus P6T Deluxe
Corsair DDR3 1600 6GB kit
Asus GTX 260
Asus Xonar DX
two Seagate Barracuda ES.2 500GB harddrives, running in RAID0 for the moment
ENERMAX MODU82+ EMD625AWT 625W powersupply
LIAN LI PC-60BPLUSII case

I am coming from an old p4 2.4ghz so I'm a little rusty with the optimal temps for the new processors.
 
Next time, degrees Celcius.
189 = 87c
100-110 = 37-40c

@ Boostage, it's not that high. You will see similar temps for C2 45nms.
 
Pretty much everything that's been posted so far about the load temps on these i7's is that they run HOT . . . much hotter than we've been used to with our lga775 CPU's. So your temps aren't abnormal.
 
Really, there's two or more methods to apply thermal compound. Myself, I prefer #2.

1 - The dot method. Apply a half-pea sized dot of compound onto the center of the CPU, then put on your heatsink. It will flatten out over a short time period to cover the CPU.

2 - Spreading method. Put a small half-pea sized dot of thermal compound onto the center of the CPU. Then spread it so it makes a thin layer on the CPU. You can use a flat razor to do this. I use a plastic zip-loc back wrapped around my finger to spread it around though. It works a little better.

Also, I too hate those plastic clips they use on the Intel boards. I like the tension bracket with lever type, or screw down mounts.

Anyway, good luck and let us know if you can remedy the solution.

For reference, At maximum power 100% usage with the stock heatsink, Intel specs the i7 at 67.9C degrees. They use the equation TEMP(C) = 43.2 * 0.19 * watts. (using 130w) At least that's how I think it's read if I'm reading it right.

http://download.intel.com/design/processor/designex/320837.pdf
 
67.9C is Tcase @ 130W. The die temp will be higher at that load/IHS temperature.

I haven't looked through the other spec sheets to see where the processor will start to throttle (i'd guess around 100C). From the [H] review, the high temperatures using the stock heatsink are normal.
 
Okay thanks guys for the reassurance and the link to the tech datasheet. I couldn't find it earlier.
 
"Tcase" is a temperature measurement from the diode reading temps centered directly below the cores. I think there's another set of sensors called "Tjunction" that read direct core temperatures. From what we saw on the LGA775 Core2's, there was around a 5C to 10C delta between Tcase and Tjunction. If that holds true for the i7, then max core temps could maybe even be up to 80C.
 
lowteckh said:
Next time, degrees Celcius.
189 = 87c
100-110 = 37-40c

@ Boostage, it's not that high. You will see similar temps for C2 45nms.
Click to expand...


I have a Q6600 @ 3.5ghz that never gets past 51c degrees on full load that's with 1.47v so I figured a 45nm i7 would run much cooler. even though im running a xigmatek s1283. I didnt realize they ran that hot
 
Core i7 runs much hotter than Core 2 Duo. The OEM heatsink is barely adequate for Core i7. The good news is that during early testing, it seems that they run reliably when overclocked a good 15C to 20C hotter than Core 2 Duo can..

From what we saw on the LGA775 Core2's, there was around a 5C to 10C delta between Tcase and Tjunction.
Click to expand...

rge over at XS recently did some real world testing. He cut a hole in the top of the IHS and mounted a calibrated sensor like Intel recommends. The difference between core temperature and Tcase temperature when running Prime was approximately 27C. As long as your processor isn't throttling you don't have to worry about it being too hot.

RealTemp 2.86 has been updated and is working correctly with Core i7. It logs whether you've reached the thermal throttling point and displays it on the main page.
 
Oh god, not this again :rolleyes:
Phenom gives off more heat then this does but this runs hotter because of the denser transistors under the IHS. Just to reiterate: Heat =/ temp i.e 140w of heat (Phenom) at 60ºC > 130w of heat (i7 920) at 80ºC
 
Uberbob102000 said:
Oh god, not this again :rolleyes:
Phenom gives off more heat then this does but this runs hotter because of the denser transistors under the IHS. Just to reiterate: Heat =/ temp i.e 140w of heat (Phenom) at 60ºC > 130w of heat (i7 920) at 80ºC
Click to expand...

I'm sorry, did I mention heat? :rolleyes:

I'm just saying that it is hotter than Phenom, so did I say anything wrong? Is it hotter than Phenom or not?
 
As far as I'm concerned, "hot" refers to level/degree of temperature. I have no idea why he injected heat in terms of wattage dissipation into the argument. We were only talking about temperatures.
 
unclewebb said:
rge over at XS recently did some real world testing. He cut a hole in the top of the IHS and mounted a calibrated sensor like Intel recommends. The difference between core temperature and Tcase temperature when running Prime was approximately 27C.
Click to expand...
Thanks for posting that
 
Temperatures aren't nearly as important as people think they are.

I ran a Prime95 Small FFT test today on my E8400 but decided to unplug the fan on my Tuniq cooler. I let it run for 3 hours where it constantly bounced off the thermal throttle point just below TJMax.

hote8400fw5.png


Only once did it get hot enough that it actually hit TJMax at 100C which you can see in the SpeedFan graph. The rest of the time it would intermittently throttle when ever it hit a Distance to TJMax of 2. This cycled the multi down to 6.0 (2400 MHz) which kept the temps from getting too out of control. That's exactly how Intel designs these things.

So far it looks like Core i7 is even more capable of running reliably at some very high temperatures. If you're stable and not thermal throttling then there's no need to worry about core temperatures. Intel makes great processors that can take care of themselves.

Here's a link to a YouTube video showing rge's testing. The difference during this test was about 23C when running Linpack and comparing core temp to TCase but the difference continues to increase, depending on your cooling and how close you get to TJMax.

Thermal throttling is based on data from the core temperature sensors and these are calibrated to keep the TCase temperature within spec even under extreme conditions.
 
That poor poor e8400. YOU'RE A MONSTER. Also, comparing an i7 to a q6600 isnt really valid when comparing the temps it runs at. A q6600 has 1.5x as much die space compared to an i7 with less transistors. When you stuff more transistors into an even smaller area, you are going to have more heat output. Cant wait til swiftech gets their mounting brackets out for the GTZ so i can get mine set up.
 
I just built a very similar pc to what hawk82 has and I can report pretty much the exact same temperatures. I'm thinking about getting a better heatsink/fan but this is only the first day I've been running this machine so I might look at some other options first.
 
My idle is around 36c and max under ibt 68c. So you guys are sayind that i can go much higher with my oc and wont dmg my cpu

Check kyle review on i7. Those are one hot little buggers
 
I was using the ASUS utility PC Probe II which came with my P6T and it is a terrible program. I just installed SpeedFan and what I have noticed is that sitting at idle my "CPU" temp is at 47 degrees Celsius but "Core 0" to "Core 5" are all 25 to 27 degrees Celsius.

I tried to figure out what the difference between CPU and core temperatures is and I still don't know if I understand completely. From what I read, the CPU temp is from a diode on the motherboard (under CPU) and the core temps are internal from the processor itself. If that is correct, is it safe to assume the motherboard diode that shows the CPU temp is not accurate or calibrated properly?

I still have to see what happens under load.

EDIT:
Under 100% load on all cores it looks like the CPU temp always reads about 20 degrees higher than the core temps. My CPU temp shot up to the low 80s and the core temperatures hit the low 60's.
 
Rage what cooler are you using?
I don't think that's accurate, try using Coretemp or realtemp because you CPU temp looks closer to what your CPU core temp should be (Core 0-3 because 4-7 are logical and don't actually exist on the die).
 
i found with my msi board that if you don't lock down the vcore when you overclock, the board will do it on its own. i've also noticed that speedfan and coretemp display wildly different numbers. i'm dumping .03v into the cpu to achieve max load stability. cpuz and speedfan measure as much as 1.18v at load. when the op posted core voltages in excess of 1.3v to get 3.2ghz i was a little stunned.
 
Uberbob102000

I'm using the stock cooler for now. I installed CoreTemp and you are right, it reads core temps of 40 to 43 degrees C while idle. Under 100% load the cores get to mid 70's pretty quick, I'm guessing they'll top out at about 80 unless I crank up all my case fans.
 
Xpletive

Youe voltage is way to high for 3.2. You should be able to do it easily at stock voltage, or under 1.2v for 3.2. ;)

My i7 920 can do 3.6 with 1.2v rock stable...
 
LEVESQUE said:
Xpletive

Youe voltage is way to high for 3.2. You should be able to do it easily at stock voltage, or under 1.2v for 3.2. ;)

My i7 920 can do 3.6 with 1.2v rock stable...
Click to expand...

Thanks for the note, I copied the settings from this article in preparation for 4ghz. I will try lowering it for now
 
His voltage is probably set to auto. Thats roughly the voltage my i7 is using at 3.2, but thats because i have been to lazy to adjust it. Just using the built in "dummy overclock" that comes on the evga x58 bios.
 
You must log in or register to reply here.
Back
Top