10 Degree Difference between Core 0 and Core 3

[dabesq]

I am concerned about the temps I'm getting on a new i7 960. From rest to peak through one run (5 minutes) of Intel's Burn Test, my cores went 39->65, 35->62, 38->61, and 33->55.

There is a huge difference between the core temps. I am using a CM HAF 932, CM V8, and Arctic Silver.

It is possible I put on too much paste. I put it on both the heat sink and the CPU. My goal was a thin-as-possible layer on each, but the layers were not translucent.

Any suggestions? These temps are without any OC'ing.

 

[OvrClkr]

Never go by individual core temp, as one or more cores will always be warmer than the others. This is normal operation, just go by CPU temp and you will be fine.

 

[CompuTronix]

dabesq,

Burn Test and OCCT use LinPack, which is a cyclic 115% workload that yields cyclic temperatures which peak extremely high. While this is useful for stability testing, it is inappropriate for thermal testing. Prime 95 Small FFT's is the Standard for thermal tesing, which is a steady-state 100% workload that yields steady-state temperatures. Real Temp also uses Prime95 Small FFT's for sensor testing.

Shown below from left to right is Idle, then 10 minutes of Prime 95 Small FFT's, then back to Idle. Small FFT's provides a steady-state thermal signature due to a steady-state 100% workload.

This is why Small FFT's is the Standard for thermal benchmarking. Since 97% saturation is typically reached within 7 to 8 minutes, a 10 minute test is adequate. Note the 5c Gradient between the CPU temperature and the Core temperatures on a calibrated system.

Please refer to the following thread: [Solved] Overclocking to 3.4ghz on stock cooling on i7 920 D0 ok? - http://www.tomshardware.com/forum/254502-29-overclocking-4ghz-stock-cooling

Without running Prime95 Small FFT's, it's not possible to draw any conclusions concerning your individual Core temperatures. I suggest that you run the "Sensor Test" in Real Temp to understand how your individual (DTS) Digital Thermal Sensors respond.

Comp

 

[dabesq]

Thanks Comp, I reseated my V8 cooler (PITA as expects) putting down two crossing lines of TP as shown in a video linked by a post.

It made a big difference. Using Real Temp 3.00, my four cores originally went from idle to peak as follows: 43-78 39-76 41-74 36-71.

After refitting the CPU cooler, my numbers were 38-67 33-64 36-63 32-60. Quite a change! Keep in mind that both tests were done with the side of my case off.

Once I closed up my HAF 932 case and reran the test, here is what I got: 43-80 38-77 40-75 36-72.

First, Core 0 hit 80 for only a few seconds. I think I can live with these numbers because no real world application is going to keep my CPU pegged at 100% nonstop. Second, imagine how high they would have gotten if I hadn't reseated the V8!

I think I am going to use Real Temp to monitor things via my G15 keyboard to see how temps look in real world applications.

I should note that I have an i7 960 overclocked from 3.2 to 4.0. I had hoped to go higher, but don't want to try given the heat. Also, 4.0 isn't too shabby, is it?

 

[CompuTronix]

After refitting the CPU cooler, my numbers were 38-67 33-64 36-63 32-60. Quite a change! Keep in mind that both tests were done with the side of my case off.

Once I closed up my HAF 932 case and reran the test, here is what I got: 43-80 38-77 40-75 36-72.

Obviously there's a serious problem with air flow in your case. I would persue a solution, since a 12 to 13c difference between side cover off versus side cover on is extreme. The difference on my Antec 900 is 1c.

 

[dirtdiver]

I would recommend more intake with your case cooling--if possible. Do you have a picture of your setup, I am wondering if you have too many fans at different angles pushing air around your case in a non-beneficial way..even a mid-grade case shouldn't really have that much of a temperature difference with the side on and off. Seems you lack proper airflow..

What is your standard room temperature? How many intake/exhaust fans do you have (and what size?). Possibly you have one (or more) of them mounted backwards (example, exhaust fan blowing air in, so hot air is not escaping the case).

Just a few thoughts..

 

[dabesq]

I am using the Cooler Master HAF 932 case, which seems too be very highly regarded and emphasizes its High Air Flow.

http://www.newegg.com/Product/Product.aspx?Item=N82E16811119160

I am using the stock fan setup, which were all pre-installed:

1 x 140mm rear fan
1 x 230mm front fan
1 x 230mm top fan
1 x 230mm side fan

On my CPU I have the Cooler Master V8 with a 120mm fan in it. Once would assume it plays well with the case since they are made by the same company, many people use them together, and again it is highly regarded generally.

http://www.newegg.com/Product/Product.aspx?Item=N82E16835103055

A few ideas come to mind. First, my computer is on top of a large table, but it is a dining nook of our apartment, so there isn't the greatest airflow. We keep the apartment cool generally, but that area gets a bit warmer. I can change some light bulbs and maybe add an external fan to circulate air in the area.

Second, the case fans are powered off the motherboard -- Asus P6T -- and even though it is set to Turbo, maybe I just should plug them into the power supply. There is no external speed control on them, so will full power I assume they will run at full speed.

Third, there are a couple of optional fans you can add to existing fans. They might help.

Any other suggestions are welcome.

 

[CompuTronix]

You might want to check the settings in BIOS > Power > Hardware Monitor > Fan Speed / Q-Fan Control. Make sure that your configuration is running your fans with Q-Fan "Disabled".

 

[dabesq]

Thanks! Q-Fan is on--I set to Turbo and thought that would max my fans. A little ignorance goes a long way! If I disable it, will the fans blow full speed powered from the MB just as if they were hooked straight into the PS?

 

[CompuTronix]

Yes, they should run at 100% RPM.

 

[dabesq]

Well, I changed the BIOS to disable Q-Fan and the numbers didn't get better by more than a degree or two: 42-78 38-76 40-75 36-71

Frankly, not having the hot overhead light bulb on could have accounted for that much change, because those number are about what I got early the morning after it had been off, but before I changed the BIOS.

I'm probably going to have to upgrade the fans in my case. I don't have the option of relocating the computer itself. It isn't flush with a wall, but there is not as much space and room air circulation as I'd like.

I guess I could go to the CM support forum to see if they have any suggestions.

 

[CompuTronix]

How's your cable management?

 

[dabesq]

It's as good as I can make it--almost everything runs through the back side of the MB to avoid blocking airflow.

 

[CompuTronix]

That narrows it down a bit.

 

[dabesq]

I should have mentioned btw that while my 960 is overclocked to 4.0, the voltage is not high at all--approx. 1.225.

 

[dabesq]

Contacted Cooler Master and after testing the CPU stock and still hitting 68*, was told I need to rma the CPU cooler, not the case. Does that make sense? How can I know if it is the cooler or the case or both?

 

[Conumdrum]

Heatsinks are made on a assembly line. The heatpipes could be faulty, it happens. they are hermetically sealed with a special liquid and vacumm sealed etc etc. Could of got a bad one...

Give it an RMA chance.............

Overclocking is fraught with issues, it happens.

 

[dabesq]

Quick update: I rma'ed the CM V8 and went with the Zalman 9900 instead. So far, so good, although I really haven't properly tested it yet. The temps appear a few degrees cooler at idle than they were wiith the V8. Haven't tested with the closed case yet, but I also upgraded my fans so I am optimistic the situation will be much improved. BTW, installation of the Zalman was significantly easier in my opinion.

 

[CompuTronix]

We'll be interested in seeing your results.