Cyphre

Friends, Romans, Pimpriggers! Lend me your knowledge!

It has been a very long time out of the fray and I have finally constructed a new PC that can actually play modern games and whatnot. So I put this whole thing together only to find I apparently had some sort of graphics card failure!

ASUS P5Q Pro
E8500
4GB Corsair DD2
EVGA 9800GTX
Seagate 250GB SATA2
Kingwin 800w modular PSU
Antec Nine Hundred

Got a pretty good deal, and cost about half as much as the previous computer (which I think is still in my sig).
So I get this whole thing setup, partition, install windows, drivers, etc. Play around with it a while and suddenly my screen flashes black in frequently, with no noticeable queue behind it. I also start to notice little specs of white pixels, presumably your everyday 'noise' that would take place on a display. Then after a little while, the flashes would more or less completely black out the screen entirely (still flashing, just no display coming up). Went through the list of possibilities and updated as necessary: checked voltage, drivers, flashed motherboard BIOS (apparently the earlier launch BIOS version wasn't compatible with some graphics setups at the time), checked all connections, power, airflow, installed all subsequent motherboard drivers included in case I was missing something elsewhere, and nothing seemed to change it. Of note, I could still use the computer during the times the screen was flashing (such as closing windows and general usage of input devices). HD activity appeared normal, and soft shutdown could be easily used in the event of total blacking out of the screen with the flickering. After going down the list of possibilities, the GPU seemed to be the root of the problems if the rest of the computer was working fine. It also really really didn't like progress bars of any kind (downloads or installs), would immediately flicker to nothing if there was a window actively doing it.

So I called EVGA, as their packaging clearly indicated contact their customer support service that is oh so great. And while they were helpful and informative, there was no clear indication of an issue on the service and they said to have it RMA'd (and even recommended doing the RMA through newegg, who I purchased it through). Following week (today) I finally got the new one, slapped it in (extremely carefully), hooked it all up as necessary, booted up and everything appeared normal. Played a good half hour of Portal to see if I could get it cooking and all was well for a while. Went to mess with firefox and got a flicker. I could only sigh and hope it was a fluke, until it started doing it again. Then I started noticing the image noise, a wide array of white pixel dandruff attempting to escape my LCD. Even started getting screen tearing near the bottom which I don't think I saw much of before. Off I went to double-check my connections, boost up fan speed, and even switched which modular connection to hook the SLI power cable to. Still no change, and it is getting worse. Those same sessions of total black outs of flickering started to happen, and a soft shut down was the only temporary cure for now. Although strangely, after hitting power, the logout screen suddenyl appeared and windows installed some other BS before shutting down.

So after much heartache and waiting, is there something I am missing? Is there some inherent problem with this graphics card or chipset I'm not aware of? Perhaps some uncompatibility with other given components? Maybe even an issue with another item of the build I'm not aware of despite the ability to use the computer regardless of whether the screen showed or not?

MaTtsPc

When you see this white pixels do you ever see any dollar sings or anything? Upside down ? marks Or Anything? Sounds to me like a Northbridge or south bridge is heating up to much! And doing it.. Switch the card to the other slot on the mother board! See if that helps! Sounds like a heat issue though!


Touch the NB heat sink and SB heat sink and see how hot it is! I bet MONEY on it says the NB sink is soooo hot you cant keep your fingers on it for more then a few secs!




Matt

Cyphre

Only tiny tiny white pixels, like you'd see with bad reception on your TV, but barely noticeable (as I'm at 1600x1200). So then what could cause the NB or SB to overheat?The northbridge even has a heatpipe running through it besides whatever else to the left of the CPU; and at least appears like it would have decent airflow with its vicinity to the HSF fans intake (one of those sideways heatpipe heatsinks with the fan pointing upwards).

AntiM

It definitely sounds like the vid card is at fault. But considering the swap didn't fix it, I'd suspect it might be a power issue. That's the one part of your build that seems weak to me.

It wouldn't hurt to pull out half the ram at a time and test it either.

Dex

I was gonna mention the PSU as an issue, but since I have absolutely no experience of Kingwin, whoever they are, I don't really know...

But yea, as Mike said I couldn't hurt to try another unit. That Kingwin unit has four 12v rails, each around the 20A mark, which may be causing problems for that hungry graphics card.

This Hiper unit for example has 4 rails also, but two are rated around 30A.

MaTtsPc

Still saying its a heat issue! Have you tried feeling the NB yet? Please just try it..

Dex

Quote: Originally Posted by MaTtsPc Still saying its a heat issue! Have you tried feeling the NB yet? Please just try it..

Something the OP could rule out by taking the side door off and placing a massive fan next to the system. Or take it outdoors at night or something...

Cyphre

The two things I generally suspected were heat and power. The EVGA customer support person also though it was power, but the BIOS didn't show any anomalies (presumably because I was in the BIOS and not windows). The Kingwin does have four rails, but when I first started, I did used both SLI power cables (the card requires two six-pin power connectors) which would've come down to two rails being used on just the card (there only being two PCI-E power cables and four 'general use' cable connectors). But either way, the progressive nature of the problem doesn't seem relative to power unless it was doing it nearly off the bat. After it would become a problem (where I couldn't even see the screen), a regular restart caused even a few flickers during startup that it never would before, but waiting a while and giving it a cooldown afforded a little more time.

I did feel the NB, and it is hot, but so is the one on my old computer I'm on now and can't really tell the difference offhand, unless its supposed to physically char my flesh. If the NB or even SB was overheating, what could be the cause? Its not something I've ever considered a problem in builds since the motherboard should at least have the cooling covered in a stock setup. BIOS voltages are all set to defaults as well in all the various areas that can be changed. Is there some inherent problem with the chipsets on this board? I believe its the P45, and ICH10R I would pull the heatsink on the chipset, but the mounts don't seem very reusable or replaceable...

Also tried taking off the door, as I had to take a look to make sure the fans were all running and whatnot. Wasn't any noticeable change, and I've only been messing with it at night as its pretty hot in the daytime currently here. Something else I noted was how quiet the system was, but it also didn't appear there was much air coming from the video card (has the exhaust vent out the rear, near the connectors).

piscikeeper

and nobody asks if the monitor has been swapped as a test ?
monitor and cable should be the first check.

Cyphre

Monitor and cable are working perfectly on my old PC that I'm typing on now (as well as the rest of the peripherals). Unless there is some other DVI format that it requires and uses a different cable. Problems are still progressive though, as it'll work identical in the first minutes or half hour.

MasterP

[Stab in the dark]
Are you running any apps like speedfan?
I recently had an issue (with an old 7900GTX mind you) where speedfan would cause my screen to flicker intermittantly - uninstalled the app, worked fine
[/Stab in the dark]

piscikeeper

you could try to manually set the gpu fan speed with evga's precision. this will let you rule out the card as the culprit since it does temp monitoring also.


as another thought...............3d apps like games are fine, while 2d is having issues?
when it pixelates again, fire up a game and see if it is still showing artifacts.
if it's fine it may be time to try different driver versions.

Cyphre

Actually the issues first started happening when I first installed windows. Didn't think much of it at the time, but that was indicative of it REALLY not liking progress bars of any kind. Screen flickering anyway, didn't notice image noise at the time.

Guess I'll have to give the EVGA precision thing a try, wasn't aware it did anything fancy.
Doesn't seem to deviate in apps, once it gets to a point of flickering all the time, or nearly all, it keeps doing it until I shut down. Not sure of the distance of time, but at a point, it'll just black out entirely, though usually gave me enough time to do things like flash the BIOS, install drivers, and other things that might help. At one point I did try running a game some ways into having been on, and the progression was near identical either way; static and flickering over everything. Was just strange in the 2D environment because it would do it without much prompting or usage, with the exception of progress bars (which I suppose would constitute animation of any kind).

Dex

Going back to cables for a second, I remember when I first got my 22" widescreen, the cable id been using on my old monitor wasn't up to scratch, it did display some of the symptons you're experiencing so its worth a try using a new cable.

Fred_G

To add a bit to pisci's post, Nvidia has horrible fan controll on some cards. I am not familiar with the EVGA precision thing, but RivaTuner works well for fan controll on my cards. Seems to use few rescources, boots with XP and remembers the settings. Plus you can overclock with it once you get your compter fixed up. The Precision program may do the same.

I would remove the heatsink on the GPU and reapply some quality TIM, they can do a really crappy job at the factory sometimes! You might also want to try GPUZ to read the temp of the GPU. That might help determining if heat is a factor.

MaTtsPc

IF it were the GPU it wouldn't be the GPU core... It would be a Ram issue and still it would be right back to OVER HEATING! Something is getting hot fast!



if you leave it on.. How bad does it get? Does it get to where the hole screen is covered?

The reason why i said the NB is because if you Google that board alot of people say that they had some similar problems and the cause was the NB was so hot you couldn't keep your hand on it! Even look at the newegg reviews for problems like that!


I'm saying RMA The board quit wasting time and get that rig up and running and FOLD




EDIT: Please up the ram voltage from 1.8 to 2.0

Up the NB Voltage from stock to .1 or .2 higher
Up the HTT link .1 or .2
Then Up the SB .1 .2

Do this... And if it still does it.. Its the mother board.. For sure Id put every last penny in my pocket!


Also that board has PCI-E 2.0 witch carries 150watts through each ch! Again! Not power! Unless the board dont have the right power But then you would get Restarts if the 3.3 Rail was to low! So for sure not power!




Matt