Phenom II X6 1090T CPU running too hot, not overclocked (yet)

[bsl20b50]

Hello all,

I am a noob with a lot of this stuff, so please forgive my ignorance.

I just put together a box with all the components as I like to run as a server to host my website. Prior to redirecting the DNS to point to my server, it ran for at least 5 days with apparently no problems. Sorry, I did not check any temps during this time. Since moving the website to the server, my CPU is always running too hot and it shuts itself down after a few hours. That is the only change I can think of. Since then I have added one small case fan with no change. My site is still showing as "under construction", but gets some traffic to live content driven by php and a large MySQL database. I get a few hundred page views a day. This morning, I booted fresh after being off all night and within a few minute the CPU was close to 60 C (likely no web traffic in that time).

The box sits in a corner of my living room and I don't think anything is obstructed. How can I bring my temps down?

Here are my components:

Case: Cooler Master Elite 310
Motherboard: GIGABYTE GA-890GPA-UD3H
Processor: AMD Phenom II X6 Six-Core Processor 1090T (3.2GHz) AM3, Retail (Black Edition)
Thermal Compound: Diamond seven carat diamond extreme performance (*this was my first time apply thermal compound myself*)
Note: I have some Arctic Silver Ceramique thermal compound that I was given after assembly that I could use if it is better.
CPU Fan: ARCTIC COOLING Freezer 7 Pro Rev.2 92mm CPU Fan For Intel & AMD CPU
RAM: Super Talent DDR3-1600 4GB(2X2G) CL9 Dual Channel Memory kit
OS: Ubuntu server 10.04
Hard Drives: 4 x Seagate Barracuda 7200.12 ST3750528AS 750GB 7200 RPM 32MB Cache SATA 3.0Gb/s 3.5" RAID 5 configuration
Power Supply: Cooler Master RS600-PCARE3-US eXtreme Power 600W ATX12V SLI Power Supply
Optical Drive: Samsung SH-S223C/BEBE 22X SATA DVD+/-RW Internal Drive (Black)
Card Reader: Rosewill RCR-IC002 74-in-1 USB 2.0 3.5" Internal Card Reader w/ USB port

Thanks,
Bryan

 

[obsidian86]

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=170&Itemid=1&limit=1&limitstart=5

 

[Timop]

Check the Heatsink and if it securely mounted?
Also, how did you apply the TIM? Is it a thin layer? Also did you clean off any TIM that was on the cooler/CPU its self?

 

[WR2]

Likely no traffic? Run your system with the internet connection shut down. Measure the idle temps just after booting up and in 5 minute intervals for about 15min.
Look at task manager and see how many processes are running and what the CPU usage graph looks like during that period. You might have some background processes running (anti-virus checks, etc).

What temp monitoring program are you using? Compare that to the temps the BIOS is reporting.

 

[WR2]

Thermal Compound: Diamond seven carat diamond extreme performance (*this was my first time apply thermal compound myself*)
Note: I have some Arctic Silver Ceramique thermal compound that I was given after assembly that I could use if it is better.
CPU Fan: ARCTIC COOLING Freezer 7 Pro Rev.2 92mm CPU Fan For Intel & AMD CPU

You don't want to switch over to AS Ceramique thermal compound.

Did you remove the thermal compound that came pre-applied to the AC Freezer 7 Pro? It would have been that gray patch on the heatsink. (it was excellent Arctic Cooling MX-2 compound by the way)

 

[bsl20b50]

Check the Heatsink and if it securely mounted?
Also, how did you apply the TIM? Is it a thin layer? Also did you clean off any TIM that was on the cooler/CPU its self?

I think the Heatsink is screwed down pretty good and secure. I cleaned off the old TIM scraping it with a credit card and then wiping it with a paper towel. Here is how I applied the new stuff. I put a little on and spread it with a credit card. It streaked and smeared and moved as I spread it, so I used a plastic bag and spread it with my finger. I had to add some TIM to get it to cover the whole area. I was surprised how difficult it was to spread the stuff and do so evenly. I finally got everything covered, trying not to make it thick, but perhaps it was too thick?

Likely no traffic? Run your system with the internet connection shut down. Measure the idle temps just after booting up and in 5 minute intervals for about 15min.
Look at task manager and see how many processes are running and what the CPU usage graph looks like during that period. You might have some background processes running (anti-virus checks, etc).

What temp monitoring program are you using? Compare that to the temps the BIOS is reporting.

With a fresh boot (I had the server off for about an hour) I have 208 processes running, 0.2% system load, LAN cable disconnected. Temp at boot was 48 C. After about 5 minutes it is 64 c.

I was checking the temps with the BIOS. I am very new to the command line interface, so that is the only way I knew to check.


The reason I re-installed the heatsink (the system came mostly assembled) is that during transit, the heatsink came off.

 

[WR2]

The BIOS temp check is usually the more accurate way of checking, so that's good.
A handy utility program that can help monitor system status is HWMonitor.
The temps it reports should closely agree with your BIOS reported temps.

48C on a fresh boot and a steady rise to 64C on an idle 2% CPU load is a clear indication something bad is going on.

When the system is shut down and cooled - you can try a light 'wiggle check' of the heat sink assembly. There should be no 'wiggle' or movement of the heatsink. That should tell you if it's firmly attached into the mounting bracket.
Inspect the mounting bracket area carefully. Check for any area where the base of the heatsink might not be flat and in full contact with the CPU. And that there is no interference with the heatsink assembly on other components close by the socket.

An ideal thermal compound application would look 'almost transparent' with just a thin film of TIM on the heatsink.
Your's might be less than ideal but from the care you did take you should be getting better performance than you are with that IC Diamond 7. IC Diamond 7 review

I'm sure you check the heatsink and the socket for any signs of gross damage (gouges, knicks, etc) due to the cooler coming loose during shipping? Had the cooler completely detached during shipping? Or just come loose on the bracket?
And that you had checked to make sure the mounting bracket on the AC Freezer 7 Pro was in the correct position.

My concern is that the heatsink or socket assembly might have been damaged. Just a small convex or concave shaping of the flat part of the heatsink might account for the temps being higher than they should.
I don't suppose you thought to do a flatness check before re-installing the cooler?

 

[bsl20b50]

The BIOS temp check is usually the more accurate way of checking, so that's good.
A handy utility program that can help monitor system status is HWMonitor.
The temps it reports should closely agree with your BIOS reported temps.

48C on a fresh boot and a steady rise to 64C on an idle 2% CPU load is a clear indication something bad is going on.

When the system is shut down and cooled - you can try a light 'wiggle check' of the heat sink assembly. There should be no 'wiggle' or movement of the heatsink. That should tell you if it's firmly attached into the mounting bracket.
Inspect the mounting bracket area carefully. Check for any area where the base of the heatsink might not be flat and in full contact with the CPU. And that there is no interference with the heatsink assembly on other components close by the socket.

An ideal thermal compound application would look 'almost transparent' with just a thin film of TIM on the heatsink.
Your's might be less than ideal but from the care you did take you should be getting better performance than you are with that IC Diamond 7. IC Diamond 7 review

I'm sure you check the heatsink and the socket for any signs of gross damage (gouges, knicks, etc) due to the cooler coming loose during shipping? Had the cooler completely detached during shipping? Or just come loose on the bracket?
And that you had checked to make sure the mounting bracket on the AC Freezer 7 Pro was in the correct position.

My concern is that the heatsink or socket assembly might have been damaged. Just a small convex or concave shaping of the flat part of the heatsink might account for the temps being higher than they should.
I don't suppose you thought to do a flatness check before re-installing the cooler?

The heatsink did come completely off in transit. I looked carefully and could not see any signs of damage to the processor. The heatsink did show some minor damage to the fins which I was able to repair with a blank CD - I straightened out the fins to their original condition. I checked the mounting when I did the install. I will check it again tomorrow morning.

How do I do a flatness check?

What is the best TIM? I may start over with a new compound. The one I used was really rubbery and hard to work with.

Thanks,
Bryan

 

[WR2]

An easy way to do a flatness check is with a straight edge. Check edge-to-edge in several locations. I check in 3 places about 1/3, 1/2, and 2/3 of the heat sink length.
Do both axis, length and width. Also check from corner to corner in both axis.
When you hold the straight edge to the heatsink you're looking for slight gaps in contact.

If you look at that IC Diamond 7 review link I posted above you'll see a lot of other good thermal compounds reviewed.
Note that IC Diamond 7 has a viscosity index of High/Thick.
Compare that to my favorites AC MX-2 and Gelid GC-2 (Low / Thin) making them easy to apply. Both are non conductive and have no cure times.

 

[bsl20b50]

Last night I removed the heatsink, cleaned the TIM off everything and checked the flatness of the processor and the heatsink. The heatsink came off easily; I did not have to break the TIM seal and it came right off. Everything looks pretty flat. One thing I noticed is that the size of the contact area on the heatsink (30 x 30 mm) is smaller than the processor (35 x 35 mm). Could this be my problem?

Of note, I also did the wiggle test and it moved left-right as well as up-down. I tightened the screws as far as they would go and this mostly eliminated the up-down, but not the left-right wiggle. I checked temps after doing this and it still got hot, but it took longer to heat up.

Then I installed the stock heatsink that came with the processor, still new with original TIM intact. The stock heatsink has a larger contact area, but the TIM covered 30 x 30 mm like the other heatsink. Testing this configuration, on a cold boot I had processor temp less than 30 C. After running the box for about 10 minutes, including some real data crunching (still didn't use too much processor power though) the CPU was consistently hovering between 34 and 36 C. After letting it run overnight, then checking it again, it got at high as 41 C, but no higher.

So here are my thoughts:

Theory 1) When the heatsink came loose in transit, the brackets may have become damaged a little bit. After running for a week or so, the inadequate bracket allowed the heatsink to vibrate looser and break the TIM seal with the processor. After this happened, temps started to rise.

Theory 2) The heatsink contact area is too small for my processor and never should have been used in the first place.

I kind of think theory 1 is more likely and I hope I can replace the heatsink with a new one. Especially since the stock heatsink is crazy loud.

What are your thoughts?

 

[Wisecracker]

This sounds more like an issue of the heatsink not making full contact across the cores --- rather than the size of the contact area on the heatsink is smaller than the heat shield of the processor (should not be an issue if the HS is snugged nice and tight).

There should be no 'wiggle'. If this is an issue then the OEM stock HSF should be re-installed. You can't screw up the AMD latch.

And don't use too much TIM.

 

[WR2]

A 30x30mm heatsink contact patch is not ideal on a 35x35mm CPU heat spreader but in itself should not have let the temps get so high. The 30x30mm sized thermal compound patch on the stock heatsink did not concern me since the thermal compound usually ends up where it's needed once a few heating and cooling cycles have taken place.

The results of the wiggle test concern me more. I'd say this is the root cause of the high temps. I like both theories as playing a part in the less than ideal cooling results.

 

[lordszone]

Last night I removed the heatsink, cleaned the TIM off everything and checked the flatness of the processor and the heatsink. The heatsink came off easily; I did not have to break the TIM seal and it came right off. Everything looks pretty flat. One thing I noticed is that the size of the contact area on the heatsink (30 x 30 mm) is smaller than the processor (35 x 35 mm). Could this be my problem?

Of note, I also did the wiggle test and it moved left-right as well as up-down. I tightened the screws as far as they would go and this mostly eliminated the up-down, but not the left-right wiggle. I checked temps after doing this and it still got hot, but it took longer to heat up.

Then I installed the stock heatsink that came with the processor, still new with original TIM intact. The stock heatsink has a larger contact area, but the TIM covered 30 x 30 mm like the other heatsink. Testing this configuration, on a cold boot I had processor temp less than 30 C. After running the box for about 10 minutes, including some real data crunching (still didn't use too much processor power though) the CPU was consistently hovering between 34 and 36 C. After letting it run overnight, then checking it again, it got at high as 41 C, but no higher.

So here are my thoughts:

Theory 1) When the heatsink came loose in transit, the brackets may have become damaged a little bit. After running for a week or so, the inadequate bracket allowed the heatsink to vibrate looser and break the TIM seal with the processor. After this happened, temps started to rise.

Theory 2) The heatsink contact area is too small for my processor and never should have been used in the first place.

I kind of think theory 1 is more likely and I hope I can replace the heatsink with a new one. Especially since the stock heatsink is crazy loud.

What are your thoughts?

Hi
the temps are surely high, atleast for amd processors.what u should do it take out the heatsink and check it physically for any cracks. i also think that u are not applying the thermal paste correctly. i think u are applying more than what is required and that causes overheating. what u should do is that remove the old paste and clean both the contacts of the heatsink and the processor. then put the paste in a small ball like shape in the centre of the processor. the ball must be around 3mm or something, meaning small. the paste will automatically spread when u apply the heatsink. once done then tighten up the heatsink and then run. if the system is overheating again then its your heatsinks fault. it might be faulty or not for your processor. hope i helped

 

[bsl20b50]

Today I recieved the replacement for my damaged fan. For the past several days, the stock fan has adequately cooled my system. When I installed the new fan today, everything snapped into place nicely. That did not happen when I installed the damage fan; it appears that the brackets were damaged as well. That is why I was not able to get the fan secured and it was overheating.

Thanks for all of the help.

 

[bsl20b50]

Best answer selected by bsl20b50.

 

[Mousemonkey]

This topic has been closed by Mousemonkey