TechSpot - PC Technology News and Analysis

I'm having temperature issues, and I need help getting to the root of the problem. I built this computer back in March of 2006. I just recently purchased a nice new widescreen LCD monitor (1680 x 1050 display) and starting playing games on my computer again.

When doing some routine maintenance and cleanup, I noticed that my idle and load temp numbers seemed high.
I also had my computer shut itself off on two occasions. Once when playing an older RPG for only an hour. The second time was when I tried running the 3DMark06 test, which really bogged down the cpu and forced the computer to shut itself off.

Here are my temp numbers per Everest (idle/medium load):
Motherboard: 25C 25C
CPU: 44C 66C
GPU: 60C 67C
GPU Ambient: 44C 50C
Maxtor HD: 42C 43C

I don't overclock at all. Don't know whether I need another fan is something else is amiss.

I appreciate any and all help I can get in resolving this and making sure my computer is running at it's optimum level.


Case: Antec Sonata II
PSU: 450W Antec SmartPower 2.0 (ATX12V)
CPU: Intel P4 630 3MHz
Mobo: DFI LANParty 875P-T
RAM: 2GB OCZ PC-3200 DDR400 Dual Channel Platinum
Video Card: 256 MB eVga GeForce 6800GS AGP8X
HD: 200GB Maxtor ATA
Lite-On DVD-RW
Lite-on DVD
Monitor: 22” Samsung Syncmaster 226BW
OS: Windows XP Home Ed. SP2

Those temperatures arent all that high, perhaps a good dust out will help.

I believe the max temp for my particular CPU is 67.7C

I think it shuts down after that.

Try my suggestion and see if it helps then.

I cleaned all the dust out, but only for a minimal improvement.

Is there anyway to narrow it down to CPU, GPU, fans, or PSU? I not sure what I should check first, second, etc. How can I verify that my current fans are working properly or that my PSU is fine?

To do anymore, you will have to pull it all apart and inspect everything.

I also have and use the PC Wizard temp readings, but they gave me almost exactly the same measurements as Everest. Here are the results from the Global Performance Benchmark (if that helps at all) ran just now:

PC Wizard Rating: 2699.29
Processor Global Performance: 3588.925
Cache Global Performance: 19488
Memory Global Performance: 2411
Video Global Performance: 1440
Hard Disk Global Performance: 65


As for a reference, It is now 8:15am est. The room temp is 24C (75F). My computer has been on for 30 minutes and idling. My current CPU is 39C (102F) and GPU is 56C (133F). The temps in the initials post were from yesterday afternoon after the computer had been on for several hours.


Following advice from several posters on several boards, I removed the side panel of my case last night and remeasured the temps with the side open and also with a floor fan blowing into the case. Here are the idle temps:

Room Temp was 26C (79F)

(Case closed/Case Open/Open with Fan)
CPU: 42C 41C 41C
GPU: 62C 59C 56C
GPU Ambient: 44C 42C 39C
Maxtor HD: 43C 44C 34C



I also attempted to run the 3DMARK06 test again last night (the one that made my computer turn off last time). My computer actually completed it this time, but crawled through it ever so slowly. Here are the basic numbers from it:

3DMARK06: 2284
SM 2.0: 965
SM 3.0: 882
CPU Score: 861

Graphics Test 1: 7.998 FPS
Graphics Test 2: 8.091 FPS

CPU Test 1: 0.272 FPS
CPU Test 2: 0.436 FPS

HDR Test 1: 7.377 FPS
HDR Test 2: 10.262 FPS



Thanks all for the help! I really appreciate it. Hopefully we'll get it running up to speed in no time.

It does look as if some extra cooling will help. How many fans does your case have in it? What size are they? What speed are they? Does your case have provision for any more fans?

That CPU you have can go way further than 67°C. Unless you're reaching 80°C or so temperatures you should be fine.

The 450W Antec SmartPower 2.0 PSU that came with the Antec Sonata II tower has dual 80mm fans in flow through configuration. The manual says this about this PSU's fan system:

THE PUSH-PULL FAN QUIET POWER SUPPLY
SmartPower 2.0 features an innovative design to help decrease noise during normal use, but which still allows for superior cooling capabilities as load increases. To this end, SmartPower 2.0 features both an internal intake fan as well as a unique external exhaust fan. The internal intake fan operates when the power supply is turned on. It rotates slowly to blow out the hot air from the power supply, and is thermally controlled, speeding up as heat increases. The external exhaust fan remains off to make sure your computer maintains a quiet computing environment until the power supply senses the required temperature to turn on the external exhaust fan.


The case came with a 120mm Rear Exhaust TriCool fan, which has a three speed switch. It had been set to Low (1200 RPM). I switched it last night to High (2000 RPM).

The case also utilizes something called the Advanced Chassis Air Guide (ACAG) system. It runs from the rear of the case to the CPU to provide fresh,cool air to the CPU without the help of any fans (passive cooling). It does come with two optional fan mounts though - one 92mm mount for the CPU and one 80mm mount for the VGA card (active cooling).

Here's the pdf link for my case's user manual. http://www.antec.com/pdf/manuals/SonataII_En.pdf


I'm not sure whether I should utilize these extra two fan mounts or not? Or whether the CPU needs a better cooler (it currently uses the stock that came with the P4 630 chip) and an application of a better thermal paste? Either way, I'd need to be walked through the process, if they were deemed necessary.



My benchmark numbers definitely seem anemic and seem to indicate that my system isn't running at it's best level? Is there a better or more appropriate test or benchmark to help determine the current state of my system?

Adding fans certainly wont do any harm. A better cpu coller is also a good idea.

Also, making your cables as tidy and as out of the way as possible will help too.

Have you taken a look at the 3Dmark result browser ? You could for system with similar specifications to yours & see if they indeed score better. Your system is quite dated & 3Dmark was known to score quite low on nothing but the latest & greatest at the time of its release.

Is there a more appropriate benchmark I should be using? Or is there some software out there that can ensure my components are running at the optimum level as they should be?

With those temperatures, there is nothing to be concerned about. If you want the CPU temperature lower for the sake of being lower (or overclocking), then get a real CPU cooler, and it can drop 10-20C off your peak CPU temp. A cheaper alternative would be to purchase Arctic Silver 5, heatsink cleaner/purifier (AS), and reseat your HSF and use the AS5. That can promote a significantly better temperature.

Computer performance doesn't decrease as temperature increases (that is a confusion of ideas). Components can fail (or in the case of video cards, have visual artifacts) or be auto-shutdown due to too high of temperatures, but their performance will remain 100% whether they are at 30C or 70C.

That is, a CPU at 0C and a CPU at 79C perform the same. At 80C, the hot CPU will trigger an auto-shutdown of the computer. If it didn't and the temperature increased hot enough, it could be damaged.

If yours is hitting the auto-shutdown, I would advise reseating the HSF w/ Arctic Silver 5. If that doesn't cut it, a new HSF (made of copper), will do the trick.

Just follow Dido's advice and compare your system to others of similar spec.

Thanks everyone.

Hopefully I can put all this advice to good use and avoid any more auto-shutdowns.

Well in some cases it does have an effect on performance. Some Pentium 4s running on Socket-423 used to throttle when the CPU would rise to certain temperatures so the system didn't shut down but was quite noticeably slower than it ought to be.

Heat is not the direct cause for the performance change, though, so no, it doesn't. It is the settings of the program that configures the processor to change frequencies that has the effect on performance. Heat, however, is not responsible for performance degradation. It is an intentional reduction in speed by software, by design. The assertion that heat causes performance reduction in benchmarks or real-life applications is false. Heat can be measured by a sensor that can trigger software to change performance characteristics, but heat does not have a direct effect on processor performance, and it certainly wouldn't scale in some proportion as heat increased. In any event, it definitely isn't the OP's issue.

heat seems to be a big issue here, maybe some of the parts dont match each other, or like Rik said, take it apart and inspect it piece by piece to see if any damage has come to the parts.

How can someone go into that and fix that problem.