Searching around this seems the hot spot for friendly helpful advice. I am a noob concerning computer hardware I know just enough to get me in trouble and get lost in the jargon.
I am running a custom build PC from scratch from a few years ago. I am recently encountering a reoccurring issue of getting "noise" pixelated speckles dancing around, BSOD when running games and higher end application crashes. Coincidentally this led me down a path of getting some software that gives me some temperature feedback. A replaced and upgraded heatsink has saved my CPU, but that hasn't resolved my Graphics problem nor has it helped my video cards keep cool(according to speedfan idleing at about 60 degrees). Any advice would be much appreciated to help me resolve this issue(s) as it may be one in the same.
Case: Antec 9000 series
Fans: 2 Front 120mm (air in), 1 rear 120mm (air in), 1 Side 120 mm (air in), Top 200mm (air out). (note the Heatsink is a Thermaltake Black Widow Spin Q(has a built in fan)).
Motherboard: EVGA 132-YW-E178-FTW
Processer: Intel Core 2 Quad Q 6600 @2.4GHz
Video Card: 2x NVIDIA GeForce 9800 GTX/9800 GTX+ (SLI) Mounted to the top and bottom slots (may be to close to the power supply but I don't know what creates more heat a power supply or a video card)
Memory: 2x MUSHKIN 2GB DDR2 -1066 (I hope that is the right terminology)
Power Supply: Dunno
Hard Drives: I think are irrelevant but will include if needed.
1) A bad or underpowered psu can cause some strange issues. Exactly what brand and power is your psu?
Look on the data label for the total amps on the +12v rails. That is what powers the gpu's.
2) I think that rear fan should be an exhaust.
3) GPU's run hot, but they are built to tolerate it. for an older technology gpu, 60c idle and 100c under load is not unusual.
In a sli configuration, the top card will get much hotter if it is in a slot close to the bottom card. Older mobo designs often had that pci-e x-16 arrangement.
4) Remove the side covers from your case, and direct a house fan at the innards. If that helps much, you have a case cooling issue.
5) Since you have had this rig for a while, have you checked for dust in the graphics cards coolers or on the case fans?
6) Can you run with each card individually, without sli? Perhaps one card is going bad. You are probably still under warranty.
7) Is your graphics driver up to date?
8) A psu is designed to take care of it's own heat, not any others. Unless it is very inefficient, psu heat should not be an issue.
9) Sometimes reseating the gpu heat sinks using fresh thermal material can help.
geofelt has given some good advice already but i'll add some more. Alot of graphic card solutions don't offer cooling for the memory chips of any kind. Zalman and a few other cooling manufacturers offer complete packages with a gpu cooler and heat sinks for the memory chips on your graphics card. Maybe you can also get just the heat sinks somewhere. Make sure there is also a good airflow between the 2 cards so one isn't chocking the other with warm air.
2) I would agree and I have flipped it around it seems more logical for a continuous flow of air, although case manual recommends differently.
3) Well it is good to know that I am not frying them. This Motherboard can support up to 3 Video Cards so I have them mounted on the top and bottom slots leaving room between them. This puts my second card with only 1/4 from the power supply fan. That is why I was wondering what creates more heat a video card or the power supply. It might be beneficial if the video card is less to bump it up to the middle video card slot.
4) I tried that didn't notice much difference in temperature, but I do have that 120mm fan blowing almost directly onto the top video card.
5) Cleaned them all out when I was trying to diagnose the problem myself.
6) I think this might be the case tried running without the SLI and haven't had the dancing pixel crash yet but haven't done much playtesting yet. (or could be a SLI compatibility\ configuration (don't know anything really about how it works)
7) Yep all up to date on the drivers
8) This was in reference to Video card positioning as I expanded on #3
9) My videocards are enclosed with the exception of the fan exhaust and a air intake not sure how to go about getting into the card to have a look or if this will invalidate any warranty I have left. ( I do have plenty of thermal compound left over from mounting the new CPU heat sink.) This applys to roalds post as well about mounting a GPU cooler and heat sinks.
If I could find a way to test my cards for some defects(since I am new to this a readout of a bunch of numbers is lost on me(need a program or something to hold my hand for a bit until I gain some more knowledge)
If one is bad I will just have to scoop up one the the GTX 560s to hold this comp over for another year before I upgrade the motherboard.
Thanks so much for giving me some feedback of things to try.