Sign in with
Sign up | Sign in
Your question

Green Pixels and CPU Temp Troubles

Last response: in Systems
Share
February 22, 2010 12:54:11 AM

So this week I built my first gaming rig and everything has been working fine except for some odd issues.


I went with the ATI 5870 and I randomly see green pixels on the screen. They show up in Windows and menus in game, but not while playing. I tried to take a screen shot to show, but checking the images out on a different computer the green pixels were absent. They also disappear when I unplug the monitor and plug it back in, until I move the mouse or touch the keyboard. I'm using a DVI-VGA adapter until I'm able to get a new monitor and thought this might be the issue, but didn't want to RMA the card or buy a new adapter until I could narrow down the cause. So would this be a problem with my card or the adapter?

My second problem is that my CPU temps are fluctuating a lot. I'm using an i5 750 with Hyper 212 cooler. I've tried Core Temp, Real Temp and Speed Fan with the results changing a lot. I get anywhere in the range of 18 to 40 at idle and this will change from second to second. I ran Prime95 and the at full load the temp sits at about a constant 54. Would this be an issue with the heat sink or are these programs kind of iffy to begin with?

Thank you for any assistance.
a b B Homebuilt system
February 22, 2010 1:00:50 AM

Well your heatsink seems to be fine if it holds a 100% load at that temp stably. But from 18 to 40 degrees at idle seems to suggest that power is surging through there when it isnt supposed to. More power going through will generate heat even when idle and this doesn't do your CPU any good either.

As for your 5870. Artefacts are generally found when your GPU beings to overheat and thus produces errata in the graphical processing (green dots, lines, colour changes). This might also be because of power surges through the system. What is your PSU and what is the area you are living in like for power stability etc? (Like do you live in an area rich in thunderstorms or other such activity?)
m
0
l
February 22, 2010 1:05:59 AM

I agree, CPU should be fine. The big fluctuations could in theory be explained by the CPU entering deeper states of sleep. Have you checked the CPU multiplier - bus if it is changing along with the temps? If so then there is your answer.

Regarding your green dots it sounds as either a video ram fault-overheat, or the issue that the new 5xxx ATI cards are experiencing. Try installing the latest Catalyst drivers 10.2 which supposedly resolve the most of the issues.
m
0
l
Related resources
Can't find your answer ? Ask !
February 22, 2010 1:07:43 AM

The PSU is an OCZ fatality 700w. As far as I know the area is fine, can't remember the last time we had any thunderstorms or anything of the sort.
m
0
l
a b B Homebuilt system
February 22, 2010 1:19:20 AM

darkguset said:
I agree, CPU should be fine. The big fluctuations could in theory be explained by the CPU entering deeper states of sleep. Have you checked the CPU multiplier - bus if it is changing along with the temps? If so then there is your answer.

Regarding your green dots it sounds as either a video ram fault-overheat, or the issue that the new 5xxx ATI cards are experiencing. Try installing the latest Catalyst drivers 10.2 which supposedly resolve the most of the issues.


The multiplier and BCLK frequencies will be all over the shop due to turboboost dependent on activity however this does not justify such a massive range of temps for idle. The 12v rails on the PSU could be a bit off colour (not literally, just a figure of speech).
m
0
l
February 22, 2010 2:13:26 AM

I've installed the 10.2 Catalyst driver and still getting the same problem unfortunately.

From the monitors I have running it looks like the multiplier stays at a constant 9 in both CPU-Z and the real temp i7 Turbo, although the calculated multiplier will spike to 11 some times. Now that I've been watching all these some more the range is down to 19-27. The core voltage also changes infrequently by about .008 V.

This shows all the info from the monitors, I'm not really that tech savvy so maybe I'm missing some thing obvious http://s938.photobucket.com/albums/ad221/awzmundus/?act... .
m
0
l
February 22, 2010 3:10:36 AM

awz said:
I've installed the 10.2 Catalyst driver and still getting the same problem unfortunately.

From the monitors I have running it looks like the multiplier stays at a constant 9 in both CPU-Z and the real temp i7 Turbo, although the calculated multiplier will spike to 11 some times. Now that I've been watching all these some more the range is down to 19-27. The core voltage also changes infrequently by about .008 V.

This shows all the info from the monitors, I'm not really that tech savvy so maybe I'm missing some thing obvious http://s938.photobucket.com/albums/ad221/awzmundus/?act... .


It sounds like your VGA is overheating then. If you are not overclocking at all then it is probably faulty. If you are overclocking please reset to default clocks and see if you get any strange patterns then. If not then unfortunately your card just can't overclock well.

Regarding your CPU, you say the temperature difference is coming back to more normal values. I just noticed that you just built your system, hence i would assume it is just the thermal paste on your heatsink curing over the days. If that is the case it might take a few weeks for a full cure in which time the gap in your temperatures is going to get smaller and smaller. That is perfectly normal and nothing to worry about. In fact most thermal paste manufacturers state that period in their disclaimers.
m
0
l
February 22, 2010 4:24:49 AM

Nope, not over clocking any thing. Must just be a bad card, which I was really hoping against. I'm not sure if over heating is a problem though. Running at idle the card is around 38-40, which for the 5870 is apparently normal. Guess it's time to send it back.

But that is reassuring about the CPU temps. I was worrying I might have messed up the heat sink. Thanks for the help.
m
0
l
February 22, 2010 11:32:43 PM

Send the card back then. Regarding your CPU, run a stress test for a couple of hours (Prime or something). If it does not crash at those temperatures then there is nothing wrong with the way the heatsink is sitting on the CPU. If it does crash then remove the heatsink, check if the thermal paste is evenly distributed and there are no gaps and re-sit it carefully making sure it is sitting evenly.
m
0
l
!