Maximum PC

I seem to have a problem that has plagued all too many people that have purchased 4870 based cards. I love the card and I'm quite happy with how it has performed but now it is randomly crashing the entire system and the most maddening part is that the same set of situations will not always cause it to crash. Sometimes I'll be playing a 3D game like EVE Online or Prototype and the card will crashing bringing the rest of the system with it. Other times I'll simply be doing some web surfing and have some other 2D applications open and the card will crash taking the system with it. What happens when these crashes occur is that the video card stops sending a signal to my monitor and if there happens to be any sound playing it keeps looping. In order to get back to a working system I have to press the power button on the front of the case. When I get the system back up unfortunately there's nothing in the Windows logs to indicate precisely what has happened. So here's some additional information that perhaps you folks can use to help me out.



System Specifications

CPU: Intel Core 2 Duo E7400

Motherboard: Gigabyte GA-EP43-UD3L rev 1.0

RAM: 4GB of OCZ Gold DDR2-800 and 4GB of Kingston HyperX DDR2-1066 (they both have the same timings, the Kingston has to be overclocked in BIOS to run at 1066, it's going off SPD data currently as is the OCZ)

Video Card: Asus Dark Knight TOP Radeon 4870 512MB DDR5

HDD: Western Digital Caviar Black 640GB

HDD: Seagate 7200.11 1TB

DVD Burner: Lite On 24x DVD+/-RW

What I have tried so far:

1) Loaded various drivers for the video card. Currently I'm running the 10.7 drivers. I have also tried Catalyst 10.6, 10.5, 10.2, 10.1, and various other previous drivers

2) I have updated the motherboard BIOS to version F8, there is version F9 but all that supposedly does is fix a crossfire issue so I doubt it would help anything

3) I have updated all motherboard based drivers (network, audio and chipset)

4) I have replaced the previous PSU with a BFG GX-550 (the equivalent of a GS-550 except it is modular, the previous PSU is a GS-550)

5) I have stressed the video card using various games and tools to see if it will overheat and crash. I have tried Far Cry in demo loop with max details and 1920x1080 resolution for 10 hours during which the card never crashed. The same can be said for EVE Online as well. The card typically idles at about 61 degrees Celsius and the most I can push it to while gaming is around 84 to 86 degrees Celsius for the GPU. The memory portion of the card typically gets to a max of about 86 to 88 degrees Celsius and the Shader Core around 74 degrees Celsius. These temps were obtained using GPU-Z.

6) I have tested the RAM with both memtest86+ (latest version) and Microsoft's Memory Diagnostic. The RAM has passed running for 9 hours with 5 passes and 0 errors under memtest86+. Microsoft's memory test also was not able to find any errors.

7) I have stress tested the CPU using Prime95 for about 4 hours. During that time the CPU was under full load and it never got above 62 degrees Celsius. Most of the time it was going back and forth between 58 and 60 degrees Celsius which isn't bad for the OEM heatsink/fan that came with the CPU.

Every time the card crashes LEDs DE1602 and D601 as being lit.

9) I have checked to make sure all of my RAM is seated firmly and I've also made sure the video card is seated firmly by completely reseating all of my RAM and the video card. I've also made sure that the plugs going to the video card were in firmly. And while I was reseating everything I did my monthly cleaning of the card using a can of compressed air.

10) I have my PC plugged into an APC XS800 UPS which provides good clean power to the PC.

12) If I try and reboot just by pressing the reset button on the front of the case when this occurs it will reboot without video and do the Award BIOS beep sequence of long-short-short. If I have to use the reset button for any other reason it works normally. In order to get video after this issue occurs I have to hold down the power button to completely shut off the system and then press it again to turn power back on.

I'm seeking this second opinion because after looking through various websites reviews on the BFG GX-550 PSU that I have seem to be mixed. The PSU is essentially the same as the BFG GS-550 except it is modular. Both are supposedly manufactured by Huntkey and while some folks say its a good PSU others say its not so great. So, is there a reasonably good chance that if I replace the PSU with something like a Corsair or Antec PSU my problems will be taken care of or is it more likely that there is something wrong with my video card?

Have you looked in the Event Viewer, system and application logs to see what is crashing your pc?

It could be corrupt drivers crashing your system.

How did you uninstall and install the drivers?

Did you use Driver Cleaner to clean out what the uninstaller left behind before you installed the new driver?

Was the anti-virus shut down when you did both?

Did you run ccleaners registry section to clean out the registry before you installed the new driver?

Nasty

Yes I have checked the Event Viewer in every log available including sytem & application logs. There were no errors logged at the time of PC crash.

Drivers were uninstalled by removing the card from the device manager and selecting the uninstall drivers checkbox that popped up. I have then gone into the Programs control applet and removed all of the ATi installed tools and then installed the new drivers. This included installing a new ATi CCC, Catalyst Install Manager, etc.

I have used Driver Cleaner in the past but not these few times. I have found that removing the drivers using the above method has gotten rid of everything, and Driver Cleaner has not found anything additional to remove. Say what you will about ATi's drivers but they seem to thoroughly remove themselves nowadays if you follow the proper procedure.

No I did not shut down the anti virus when installing or uninstalling the drivers. Truth be told it's nearly impossible to completely shut down Norton without removing it from your system and doing that is a long process in and of itself.

I did not run ccleaner's registry section to clean the registry. I have found that registry cleaners, be it ccleaner or others, tend to cause more issues than they are worth.

I'll see about doing the whole uninstall/reinstall routine another shot and see about using Driver Cleaner in addition to see if it actually removes anything that the uninstall missed. Edit: Well I just went to driver cleaner's website and now they want to charge for the program. No thanks, this is not the type of program that I'm going to pay for especially when I have found that it finds less and less to remove nowadays.

On ATIs website they tell you to kill the anti-virus or the driver will not uninstall right or configure itself right unless you do it.

I have never had any issues using the registry side of Ccleaner and have been using it for years.

Bah! on Driver Cleaner then...use Driver Sweeper.

http://www.guru3d.com/category/driversweeper/

If the event viewer is clean, I would suspect the psu or the card itself is at fault. Can you borrow a friends psu to see if that is an issue? If you can and the same issue rears its ugly head then I would call the video card manufacture and explain what is happening and what you did to try and find the problem.

Nasty

Well I'm going to see if the card stops crashing the system and if need be send it off to Asus for repair/replacement. I went ahead and removed the ATi drivers via the control panel and then rebooted and removed the remnants with Driver Sweeper. I had found it before the suggestion was made but thanks for it anyways. I let the Far Cry demo loop for about 6 hours and it was still running when I woke up this morning and exhibiting no issues. One interesting thing to note after performing this procedure. I've now loaded the 10.4 WHQL drivers and the card actually runs a little bit cooler while idle after the procedure I performed. It now idles at around 57 degrees Celsius instead of 61 degrees Celsius. That makes me wonder if perhaps there were some remnants screwing around with the card and in the process causing it to get a bit warmer and perhaps causing it to use the improper amount of voltage. Whatever the case may be it could be that perhaps all I needed to do was a more thorough uninstall of the drivers before installing new drivers. We'll see if this fixes the problem. If the system no longer crashes then the issue is resolved and I'll keep on playing. If it crashes again with the same symptoms then I'll still have the RMA from Asus that I can return the card on.

Yes, it sounds like there were some bits and pieces floating around that caused the problems. I hope everything keeps running.

Nasty

Well a little while after putting up that post it crashed again in the same fashion. At this point I have to think that it's either the video card or the power supply. And since it would be highly unusual to have two bad power supplies in a row I'm thinking its more likely than not the video card. So off to Asus it'll go on Monday and hopefully the replacement that's sent will fix the issue.

Keep us informed as to the outcome of replacing the video card.

Nasty

Well I saw that my RMA had completed with Asus today so since the RMA page didn't say whether they were shipping me a replacement or that they had fixed the card I gave them a call. According to the rep that I spoke with when they tested the card out they were not able to find any problems with it. So I'm getting the same card sent back to me. Which of course means that once I put it in the system I'm sure the same thing will happen once again.

So any ideas as to what it may be at this point? Some friends of mine have said that it may be the motherboard since it's the only part that I haven't tested/replaced yet. Logically that's pretty sound I have to admit although I find it odd that a motherboard could cause an issue like this but it is the main board of the system after all.

if it does the same thing try this: underclock it a bit.

my 3650 did the same thing about 6 months ago, hang with no video, power off/on reset it. it would do it randomly, sometimes games, sometimes 2D. furmark could run for hours then a simple web page would hang it. or vice versa. no pattern. folding, not folding, gaming, not gaming, idle, loaded. random "no signal" nothing in the logs.

swapped the 350 antec for a 450 antec PSU as I figured power problems. nope. its on an APC 550 UPS so power is good. different drivers. cleaned the slot out with canned air, cleaned the cards contacts and power socket. reapplied ICD7 compound, made sure all fans, HS were good. memtest86 ran for 3 days straight (using a PCI vid card) over a long weekend when we were away with no errors at all. etc.

after underclocking it by maybe 5% its rock stable again. its my ICS machine and my wifes email/light game machine and its on 24/7, no hangs at all since the underclock.

not the ideal solution but I didnt have much more time to mess with it and its not my main game machine so I just left it at that as its stable and does all the things the wife wants.

its on an Intel 865 PERL (P4 era) so maybe the motherboards on its way out.. but no lockups in 6 months and I can now beat the snots out of it again (folding, furmark, games).. so who knows..

keep us posted as Im interested also.

Good suggestions, I'm also going to take another look at my BIOS settings and make sure there isn't something that's slightly out of whack. I also have a friend that's going to loan me a GeForce GTX260 for a bit so that I can use it for testing. If that card is able to work without issue and I still have the same issue with the 4870 then I can yell at Asus some more, b/c if anything that card is going to draw even more juice from my PSU than the 4870 does. Feel free to make suggestions on BIOS settings to check and any other items I should check. I should finally get to start working on it again on Monday since that's when FedEx is supposed to deliver the card from Asus.

Quick question, would it be worth my time to do a complete clean install (ie wipe the drive & start from scratch) of Windows 7 to make sure there isn't some residual crap lying around somewhere? This particular install of Windows 7 was an upgrade from Vista, I went the clean install option that was listed on the installer but IIRC that's not a true clean install.

a bare bones install cant hurt, but I would have a current image in case it takes a digger though.. just copy the image back and you have only lost the install time + a suitable test period.

I had (and still have) XP on that machine, so its not a win7 thing.

Me, personally, I'm leaning toward PSU. However, that's a giant leap of conclusion, not knowing what else is in your case, or how your case is set up. Using this calculator, just with the components you listed in the first post without fans, lights, or any other bells and whistles you may have, and without any overclocking factored in, I get 348W. That's less than 65% of the PSU's capacity, which theoretically should leave you with plenty of overhead, which makes a hard sale of it being a PSU problem. Keep in mind, however, that the same configurations ran through different calculators will give you different results.

I had similar crashes to what you're describing, (sometimes with the screen going blank, sometimes it would lock up on a frame and just display that like a screenshot), with the audio loop as you described and having to use the power button (press and hold until shut down, then press again to power back up) because the reset wouldn't work. All of this was right after a video card upgrade, so naturally I blamed the card. It turned out to be an inadequate PSU, which I actually didn't discover until the PSU burned ... well, really nearly blew ... up from my over-use of it. Once I replaced the PSU, everything ran fine.

Two important considerations about your PSU's ability to supply enough power:
1) The ratings on PSU's can be misleading. PSU ratings are stated for the overall output. They don't break down the output to +12v, +5v, etc. It's just a plain overall rating. Where most people run short of power is in the 12v area. PSU rating is measured at a controlled temperature (I think I read somewhere that the industry standard for testing is something like 25C?). As it operates at higher and higher output levels (more computer usage) it builds up heat. This is typically paired with increased heat coming off of those components that it's powering, and it's ability to supply power will diminish as all of this heat builds up. Running your stress test successfully overnight while the ambient room temps were likely cooler, then having it crash again during the day tells me a story of temperatures degrading output.

2) Electrical components, especially capacitors, degrade over time and use. This applies to both the components which will begin to require more power as they degrade, and the PSU which will have reduced output capabilities over time. To a large degree this has been improved on in the last few years by manufacturers using higher quality capacitors.

Just as one more side note: I never trust a manufacturer's testing to tell me that a piece is good. I had a bad stick of memory that tested fine under two different memory tests. The only way I could truly identify the stick as bad was to swap it with the memory in another machine. The problem followed the memory to the other machine, even though it tested as fine. Thank goodness the manufacturer did a direct exchange on the RAM instead of the test and return routine that ASUS is giving you.

I suppose it could be the PSU although I'm thinking it's still the video card. I'm none too trusting of Asus' testing either since they definitely did it in under 24 hours. I have an eVGA SSC GeForce GTX260 Core 216 in my system now (got it as a loan from a friend) as the 4870 that I got back from Asus yesterday had the same issue after approximately 15 hours of use. This was after numerous tweaks to my BIOS. If the GeForce is able to continue working for a longer period of time without the same issue occurring then I would definitely say that the issue resides with the video card as the GeForce definitely draws more juice from my PSU than the 4870 typically did.

If the Gforce survives be sure to tell ATI's customer service reps. You have more now have proven that it is the card/driver/combination that is the problem.

RMA number two has been setup with Asus. The representative that I spoke with on the phone was polite, took note of the issue that I described, noted that it would need to be tested for at least 15 hours, and this time they're sending me a pre-paid label so that I don't have to pay for shipping. Hopefully this resolves the issue and either they fix the card or a replacement card is sent to me.

Well Asus has finished looking at the card and supposedly they have replaced the chipset on it and are shipping it back to me. To be honest I didn't even know that was possible on a video card with the way the chipsets are soldered on. I guess I'll see whether they really did fix it or not though when it arrives here on Friday.

XBOX 360 BGA reball:
http://www.youtube.com/watch?v=JiY9ryVPi7M

Well I have my video card back and running my system. We'll see what happens with this one.