In need of help getting my new Core i7/Radeon 4870 build stable

[chriseid]

I'm really at a loss for what to do next. I'm a newb at building computers and I definitely underestimated how difficult it is. Please lend me a hand if you can.

Here are my specs:
Intel DX58SO
Core i7 920
XFX Radeon 4870 512MB
3GB OCZ Gold DDR3 1600MHZ



As you can see I OCed the CPU to 3.2GHZ. I have the BUS speed at 200 with the CPU multiplier at 16x and the memory at 8x. (I have the BUS speed high because the system wont boot with a memory multiplier above 8x.)

Not only have I gotten a BSOD while extracting a large RAR file but Catalyst ruitinely crashes when I try to play games and before it does red artifacts flash all over the screen.

I have tried running PassMark burnin test and I never get an error. It doesn't seem to generate a report... Should I use another program?

Again, I'd really appreciate the help of some experienced members. Thank you.

 

[spinny]

Prime95 is the most widely used program to stress-test a rig. You'll want a program such as Realtemp to be running at the same time, so you can see if your CPU is getting too hot (anything under 70C is fine). If your rig can run prime95 for a long time (easiest to test overnight, after you make sure temps aren't out of hand), that generally means your overclock is stable, and your crashes may be from something else.

A video crash while playing games preceded by red artifacts just screams overheating video card to me (red artifacts in general do). Check out the idle temps in catalyst control center, and if you can try to get a graphics load going and see how they start to rise. Another quick and easy test is to take off the side of your case and put a fan of some sort blowing on it, if you can. If this lets you play games fine or significantly longer, it's very likely you have a heat issue.

 

[chriseid]

Prime95 is the most widely used program to stress-test a rig. You'll want a program such as Realtemp to be running at the same time, so you can see if your CPU is getting too hot (anything under 70C is fine). If your rig can run prime95 for a long time (easiest to test overnight, after you make sure temps aren't out of hand), that generally means your overclock is stable, and your crashes may be from something else.

A video crash while playing games preceded by red artifacts just screams overheating video card to me (red artifacts in general do). Check out the idle temps in catalyst control center, and if you can try to get a graphics load going and see how they start to rise. Another quick and easy test is to take off the side of your case and put a fan of some sort blowing on it, if you can. If this lets you play games fine or significantly longer, it's very likely you have a heat issue.

Hey thanks for the advice man. The video driver just crashed again while web browsing (wtf?) and I checked the temp reading in Catalyst and it was at 70C. It could have come down from a higher temperature possibly... After I set the fan speed to 50% the temp deceased rapidly. Is it possible my card is defective? There is no way my card should be overheating while surfing the web right? It is factory OCed by XFX but I would think they installed an adequate cooling solution....

 

[spinny]

How hot it should be is a hard question to answer, as it depends a lot on your case, other components in it, and your ambient temperatures. "Adequate" cooling solutions for most video cards (ie. the ones the manufacturers ship it with) usually run pretty hot.

70C is fine, but you said it may have come down from a higher temp? Where you doing something graphical immediately (few mins) prior to the crash? It is possible that your card is messed up, or that you have a driver issue, but I would try to eliminate heat as this issue first since it's so easy.

See if you still have crashes with fan speed at 50%, and if you do, just crank it up (if you can stand the noise) and see if you still have crashes. If you still do (depending on what sort of temps that had it running at) then you can probably rule out heat as an issue and move on.

I would also be testing the stability of your overclock as part of this process too, since it sounds like you have not done this (prime95 with a temp monitor, preferably running small FFTs in prime). You may have to drop back down to stock settings if the graphics problem continues, to eliminate anything in there as a culprit.

 

[chriseid]

Looks like heat was probably the issue. All I had to do was manually adjust the fan speed and the system seems pretty stable - no artifacts and no BSOD. My question now is why I had to manually adjust it? Shouldn't the video driver or card bios automatically adjust the fan speed to keep the temps low? If that isn't the case are there any utilities that will allow me to set the fans to auto adjust to a certain max temp?