Sign in with
Sign up | Sign in
Your question
Solved

Stability issue realted to cooling???

Last response: in Systems
Share
November 11, 2010 7:23:31 AM

Hi there,
I recently built a new gaming rig and since day one my computer continually freezes. I have found that a box fan sitting next to my desktop seems to alleviate the problem almost completely. I bought a Corsair H50 water cooler and some extra case fans thinking it would solve the issue, but it doesn't. I also tried hiding the cabling on the inside to improve airflow and still no luck. Somebody told me to check the power supply and sure enough my power supply was only putting out 20amps while each graphics card was trying to pull 22amps. So I bought a 850watt XFX Black Edition power supply that puts out 70amps. But it freezes without the box fan next to the open side. I just want to be able to get rid of the box fan and close up the side of my case!!!
Any ideas/help would be much appreciated.

Thanks,
Jon

Case- Cooler Master CM690 Nvidia Edition
Mobo- XFX 750i w/ SLi support
RAM- 2x 2gb DDR2
Processor- Intel q9400 2.66ghz quad core
Cooling- Corsair H50 processor water cooler, 2x 140mm fans, 1x 120mm fans (and box fan :bang head)
HDDs- 1tb and 2x 500gb
Power Supply- XFX Black Edition 850watt 70 amp
GPUs-2x NVIDIA 9800 GT in SLi configuration
a b ) Power supply
November 11, 2010 7:27:47 AM

Download programs called CPU-Z, HWMonitor and Speedfan so we know what your temps actually are.
m
0
l
November 11, 2010 5:08:46 PM

^+1

The best advice...get some actual temps to verify that your box is shutting down due to overheating.

If you can confirm that your temps are outta whack, try removing the H50 from the processor, clean both the H50 cold plate and the heat spreader on the processor with a cloth (not a paper towel) and some alcohol.

Once you have done this, use a minimal amount of thermal interface material (about the size of a small pea or large grain of rice) and spread it on the heat spreader of the CPU as evenly as you can using either your finger wrapped in a plastic bag (don;t use bare skin) or an old credit card.

Once you have spread the TIM, reattach the H50 and make sure that it is firmly engaged to the processor. Boot it up and see if your temps are any better.

Good luck!
m
0
l
Related resources
a c 83 ) Power supply
November 11, 2010 5:18:50 PM

Technically each of your cards is under 9 amps.

Where are your fans located. You have to single slot(they are single slot right?) video cards. Do you have a fan blowing on those? they may get hot working near each other and just blowing air back into the case.

As said above, get us some temps. You may not need to install speedfan, for me hwmon has all the same temps as speedfan.

cpu-z
http://www.cpuid.com/softwares/cpu-z.html

HWmon
http://www.cpuid.com/softwares/hwmonitor.html

Speedfan
http://www.almico.com/speedfan.php
m
0
l
November 11, 2010 6:28:00 PM

with speedfan my temp on the CPU cores are all under 41degrees Celsius, and my graphics cards are running around 60C. I have a 120mm fan pushing air thru my water cooling radiator in the rear and air flowing from the bottom upwards with one 120mm fan pulling the air into the case from the bottom and the hot air being pushed out thru the top with two 140mm fans. The computer will freeze without so much as a change in CPU or GPU temps.
m
0
l

Best solution

November 11, 2010 7:30:00 PM

If the box fan is alleviating the problem, then it is definitely something overheating, and we aren't covering all the bases.

Options:

(1) The sensors on your CPU and/or graphics card are not reporting the correct temps. This can be verified independently, of course, if you have a thermometer (preferably one of the laser pointer kind).

(2) Something else is overheating, such as MOSFETs on the mainboard or graphics card, RAM, north or south bridge chipsets or hard drive. Again, this can be best verified using said thermometer.

(3) Out on a limb here, but it could be an intermittent short on the mainboard or graphics card that doesn't exhibit itself until everything gets nice and warm and you have thermal expansion. The box fan could be providing just enough cooling to keep it in check. In this case, you will prob be stuck with the box fan unless you can identify the problem are and have it fixed/replaced. I know one guy who had such a problem...he ended up taking a heat gun to his mainboard to reflow the solder to fix the problem. Wouldn't recommend this unless you have experience.
Share
a c 83 ) Power supply
November 12, 2010 1:18:53 AM

For the hell of it, try placing the bottom fan on the side panel blowing on those video cards. As a bonus, it will also cool the chipset.

See if it helps

Next please grab some apps to test your system for stability/temps.

Coretemp. This can be set to log your CPU temps so even if it crashes while you are not looking or on a game, this will have the last temp logged. With that cooler I do not think its the cpu, but lets just make sure
http://www.alcpu.com/CoreTemp/

Prime95. This will stress the crap of your cpu and even memory with the Blend test.
http://files.extremeoverclocking.com/file.php?f=103

GPU-Z. This like Coretemp will log(from the sensors tab) the video card temperature(maybe even both, not sure, i can only afford one card).
http://www.techpowerup.com/gpuz/

What ever game you want to test will be used for the video stress tests. I will forgo furmarks as its overkill and can push a bad card over the edge.

Ok, so on to the tests

Start Coretemp. Once its up, look at all those temps. Under 75, good. Hit F4 to start logging temps in the coretemp folder

Open prime 95, but don't run torture test yet. If its asks you to join gimps, select just stress testing. On the run a torture test screen, click cancel.

Under the advanced tab of prime95, select Round off checking(this is important to tell you if the cpu or memory have issues).

You can now go to Options and Torture Test....

Select Small FFT and hit ok. Quickly look over to CoreTemp and watch the temps. Let this run for about 20-40 min, just to see temps. If you want to test the hardware, run it overnight.

If you crash here, reboot and check the log. It should have the last temp listed, Is it over 75(can even take higher)? If not, maybe your boards bios is set to shut down too early.

If you want to check this, on bootup hit the DEL key over and over until you see a message about entering setup. Under Hardware Monitor in the bios, there is normally an option to set a overheat/shutdown temp. See what it is at.

Warning....messing with the bios can leave the system unbootable. It can always be fixed, but if you are not comfortable, don't play with it.

If you are fine and no crash happens, move on to the Video Test.

Stop Prime and close it, you can leave Coretemp open, it may yet give you info on the next test.

Open GPU-Z and under Sensors, check log to file and give it a location to save.

Start your fav game and have a play.

If you crash, check the video temps in the log on reboot, if they are good. Repeat this test with the other video card selected(from the drop down menu at the bottom of gpu-z) in GPU-Z.

If you crash yet a second time, check the second cards temps in the logs. The video card should NEVER go over 105(even 70-80 may be pushing it on that card, i never had a 9800GT). 105 is Nvidia's official MAX temo

If you see high temps, Take both cards out and get a can of compressed and and blow and backwards from the heatsink to fan then from the fan through the heatsink to clear out dirt. Dirt/dust can cause very high temperatures.

Reinstall the cards and try again.

Still crash?

Test each card one at a time. If only one crashes, chances are the card is the problem.

This seems long and there are even more things to try if this fails. This is just some stuff that I would try first.
m
0
l
November 12, 2010 4:32:52 AM

Wow, Thank you guys so much for the input. Will try all of the above and get back to you
m
0
l
November 22, 2010 5:19:32 AM

Best answer selected by jon_hayhoe.
m
0
l
!