I have an about 9 month old built system; specs are: xfx hd 4890 q9550 gigabyte ud3r motherboard 640 caviar black hard drive Corsair 650w PSU windows 7 Temperatures in a summer room are normal im guessing, idle cpu 43-50C idle gpu: 62-63C everything is stock not overclocked or anything. I just installed windows 7 from Windows XP, and while playing the game Combat Arms, I get a system freeze and it automatically reboots by itself (in Windows XP I had to reboot it manually and Combat Arms never froze in Windows XP maybe sometimes it did). I don't know why it freezes but it's preferably my network wireless usb card wg111t I'm guessing. Is this the Windows 7 OS or is it my computer is broken/messed up(Yes there were some freezes in Windows XP during different things running)? Also when I booted it back up, it was the "f8" screen where it said something like "Safe mode may help you restore lost memory" or "System may have rebooted to not lose memory" not sure. I chose normal mode start up and worked without problems. It doesn't happen too often (First freeze with windows 7 so far). Can the auto reboot be from a CPU/GPU/PSU/motherboard failure? Or is it how Windows 7 works to make it safe? Should I worry about it? Thanks.
We need your full load temps; idle temps mean nothing. A hardware failure (even a USB device) can cause windows to crash and restart. Ram can cause such issues if it is set to a low voltage; it works at low power for the first 6-12 months, and then it needs full rated power to be stable...and often mainboards do not autodetect the correct voltage. Open Control Panel>System, then click "Advanced system settings" on the left. Under "Startup and Recovery" click "settings". In the "System Failure" section, uncheck "Automatically restart"...that way you will be able to read the stop errors.
I haven't done anything with power voltage with the ram, also at load cpu doesn't exceed 55C and GPU doesn't exceed 86-90C which are all pretty good temps. Ok I disabled automatically restart, so now where will the errors be located for me to read and stop them? Thanks.
90C seems incredibly high! I bet it smells like burning circuity! Even the latest cards should not run that hot. You can mod the fan to run faster (or even at full speed)...this free fix might solve all your problems. And yes, the change you made should allow you to see the errors, you will see them on our old "friend" the BSOD, when the crash happens. If you had the "log errors" box checked, they should be in system events.
Killerbug, no. just no. Graphics cards run at those temperatures all the time, have you never read a graphics card review?
Well at high end games and totally full load it runs that high, and normal gaming it's like near the 80s. On full load VDDC and Slaves run over 100C to 140C-150C(160C if I remember, but not sure) during stress tests. I read that it is normal to have those very high temps when doing a stress test, but during gaming even fully stressed for some reason VDDC stays much lower below the 90s or 80s for me which does prove it normal; I read that up too. (I haven't done a stress test in a long time, but that's from my memory and I read that stress tests may fry the card so I stopped using it). Thanks.
Have you looked at the logic of that last statement? A device should be able to run at full stress all the time for years on end. If it can't, then it needs cooling. That is the whole point of a stress test...to see if it can handle the stress. 80C should be OK...it is higher than average for that chip...but not too high. I would still increase the fan speed, if only to extend the life of the card. I am still of the mind that these cards run too hot...ATI & Nvidia did not re-write the physics books, electricity still works the same as it always has...and I still remember when the first Pentium processors came out, and intel claimed that they did not need heatsinks...needless to say, the chip failure rates started shooting up like crazy about a year later, plus there were tons of stability issues from day one. Then again, I guess if you change your card every 6 months it isn't a big deal if it burns out after a year...and a bad output from a video card is just a 1/60 of a second glitch on the screen; not a stop error.
Basically, yes. stress tests can only break that which is already faulty. As for the heat = failure argument, it depends how well the card is made. Most of ATI's stuff (at least the reference designs, not so convinced about non-reference cards) is built to last - my X1900XT from 2006 ran at 90C load all its life, at the age of 4 it still works fine. A large number of nvidia cards however fail under high heat, so are worth cooling quite heavily. The emergency threshold for most cards is well over 100C, if it were dangerous at any point below that, they wouldn't set the limits that high.
Nothing, I was merely explaining to killerbug that no video card would fail at the temperatures described unless it was faulty. I have no reason to suspect that in your case (yet)
I have another question for you(off topic though): If I have a 650 watt PSU, when using it does all the 650 watts go to electricity cost(in cost per Kw) or only how much is needed for the computer(about 300 watts)? Even though the pc needs only 300 watts does all the 650 function(Are all the 650 watts still in use) or only how much the pc uses while the rest of it is idle wattage? Thanks.
PSUs draw only what is needed to run the system. The rating on the label is a maximum, not how much they always use. If you attempt to draw more than that, the PSU, assuming it's a good brand like Corsair, will simply shut down due to an overload. 650W is a huge amount of power though, you have to have a lot in your PC to exceed that. PSUs are not 100% efficient, they do waste some energy in converting mains electricity to low voltage DC. Among the better units, in the US at least (the higher 230V mains of europe and other regions makes PSUs a little more efficient), they will typically achieve 80-85% efficiency for all common wattages. So if your PC needs 100W, that's all the PSU will provide, but the total amount of electricity used will be about 120W, as the extra 20W is what the PSU uses to convert the electricity. This is why PSUs need fans, as the only reason the fan is there is to remove the excess heat generated by their lack of efficiency. If PSUs were perfectly efficient, they wouldn't need fans or even vents. You will notice that small PSUs like mobile phone chargers don't have any vents or fans, but they do get quite hot, despite the minimal amount of power they convert.
I think my video card must be defective; I tried scaling back the fan...and the system crashed before it hit 70C. I know there is a big difference between the 3870 and the 4890...but I doubt there is a 40C-50C difference in safe temperature. All well...it works fine as long as I keep it cool.
The 3870 I had could get to 80C easy without any stability issues, though it tended not to get that hot unless I cut the fan speed manually.
Mine started at about 70c...it was stable by itself, but would cause crashes if I installed my hardware raid card (it would start sucking hot air from the chip on the raid card). I had to upgrade to a zalman cooler, and now it is rock solid, so long as I don't turn down the fan. I guess I just have a slightly defective chip...thankfully it works fine when it is cool, and even overclocks to the limits of my OC tools without issue.
