Visiontek HD4870 - At a Loss with this Crashing

[EXCellR8]

I'm usually pretty good when it comes to diagnosing and debunking video problems, but this one has me at a complete loss; of course, it's my own flippin' system... awesome. I can't figure out why my 4870 keeps crashing on me whilst gaming, it doesn't make any sense. I think I might just RMA the card eventually but I'd like to see what other people think about the issue first. The last thing I need is to have the replacement card do the same exact thing... which is typically my luck.

The card lasts, on average, about 2 hours before the screen goes black; the behavior after that seems completely random. Sometimes, I get a NO SIGNAL message and have to restart the system manually, and other times, VPU recover will restart the card thus returning video. VisionTek support told me that if the screen was going black that meant the card was overheating. Sometimes, I hear the fan go to 100% when the video cuts out so I'm thinking they might be right. However, I run the fan at ~60%+ when I run games so it seems highly unlikely. I have also applied OCZ FreeZe thermal compound to the core which dropped the "out-of-the-box" temps about 10c! The card never passes 50c on full load w/ the fan running above 50%. Could the vRAM be overheating?

I have tried ATI catalyst drivers 8.8 through 8.10 with absolutely no effect on this crashing. I have tried tweaking settings w/ both RivaTuner and CCC and I have even flashed the card's BIOS via thumb drive (don't tell VT). I was sure that flashing the BIOS would fix the problem but nothing's changed. The card still crashes just like it did before. The BIOS I'm using now is the ASUS TOP version.

So now, I'm debating on reflashing the card to stock BIOS and shipping it over to VisionTek for replacement/service, cuz I can't think of anything else to try. I'm also looking at BIOS updates for my mobo as my current revision dates back to... well it's kinda old. I don't really see that doing anything because it seems the only changes to the BIOS have been added CPU compatibility. Anyways, I'd appreciate as much input on this as possible. If I can't get it figured out by the end of the week, I guess I'm losing the card for awhile. Thanks in advance for any help.

As a side note, card performed just fine through hours upon hours of BioShock, Crysis and UT3 game play. This kinda leads me to believe that the card is not the problem but I guess different games use the GPU differently...

cliff notes...

HD4870 crashes in-game @ roughly 2 hours (no signal message)
running latest ATI drivers (8.10) & directX 9
card doesn't seem to be overheating
-fan is run at 60%+ while gaming
-OCZ FreeZe on core
neither ccc or rivatuner tweaks have any effect on crashing
BIOS flash did not fix the issue, running ASUS TOP bios rom
1000w PSU w/ +12v rating of 12.29-12.35

 

[kimsland]

1. Uninstall the drivers from Add/Remove programs in the Control Panel of Windows.
2. Restart the computer.
3. Run Driver Sweeper*, select the drivers that you have uninstalled and click Clean button.
4. Install the new drivers.

* For Windows Vista users, please run Driver Sweeper in administrator mode.
Right mouse click Driver Sweeper shortcut and click Run as administrator.

To download and install Driver Sweeper, please click HERE. (direct link)
Save DriverSweeper_1.5.5-setup-[Guru3D.com].exe file on the hard disk.
Run DriverSweeper_1.5.5-setup-[Guru3D.com].exe.

---------------------

Of course if the card (or video) is faulting outside Windows, when you first turn on your computer. Then it's not a Windows issue or driver issue

I'd say return it

 

[EXCellR8]

Is this Driver Sweeper any better than Driver Cleaner Pro? The reason I ask is because I've always run Driver Cleaner before installing new drivers. I will give it a try but I really can't see it helping much unless there's some nVidia residue left behind by my previous card. I must add that the new video card (4870) runs perfectly fine at start up and in 2D low power mode... never had a problem. The problem may be with a Windows system file I suppose, I've been getting an annoying "nvgts.sys" error (no bsod) in event viewer that I believe has something to do w/ my SATA DVD burner. Don't think that could tie into the video cutting out, my last video card worked fine and I had the error then too... Well thanks for the advice, and the link, I will give it a try a bit later as a last ditch effort before RMA.

 

[EXCellR8]

Possible Problem Cause?

I think I may have found the source of all my troubles. Like before, I was playing a game (Fallout 3) and my video crapped out approx 1hr and 45min into it. Same thing happened, screen went completely black to NO SIGNAL. I was actually able to restart my computer properly by "blindly" navigating to the start menu and hitting restart. This proves that the problem is with the video card.

After the computer restarted, the video came back before POST as is should and I opened up Event Viewer once Windows was done loading. There were a couple of System entries that I am pretty sure point to why my video card is failing... both mentioned "ati2mtag" as the source of the error.

[ENTRY #1]

TYPE - Error
SOURCE - ati2mtag
CATEGORY - SDTV
EVENT - 49170

Properties > Description reads: "MODE: GXO Execute BIOS Table Error"

[ENTRY #2]

TYPE - Error
SOURCE - ati2mtag
CATEGORY - GCO
EVENT - 44043

Properties > Description reads: "GCO failed to execute AtomBios"

I have been searching for information relating to these errors but I haven't found a whole lot. I think I will probably forward the content above to VisionTek support and see what they have to say about it. I know they just want to RMA the card already so ultimately, that's most likely going to happen. If anyone has any additional advice, I'd appreciate more input with this. I can't seem to find a fix for this myself so I'm still quite lost. Thanks in advance for any further information.

EDIT - looks like this "AtomBIOS" is what ATI uses for BIOS on their cards; everything is pointing to the card being faulty...

 

[kimsland]

Have a read h e r e

I had one member here at TechSpot, report that it was due to overclocking issues in your computer Bios. So possibly set Bios defaults in Bios Setup (By pressing DEL or F2 or other) key at system startup

 

[EXCellR8]

Hmm, that first link has me concerned with my RAM. I am running 4GB of DDR2 with a 32-bit version of Windows XP... I wonder if I should try to remove a gig or two and see what happens. My machine, and video card, are both overclocked but the same problem happens at stock settings. Perhaps I should look into setting the timings manually and not having them on "Auto?" Thanks for the help, I hope I can get this squared away soon!

 

[kimsland]

^^ Good idea I feel

Remove 2Gig
Set defaults
Possibly set timings manually after that

I'll wait for your inevitably reply "fixed Here's what I did..." Well here's hoping

 

[EXCellR8]

Haha thanks, I can only hope something works because I've put so much work into this computer already. I just want to use the thing now and not have to worry about it crashing. Having a powerful system that is burdened by faults is so frustrating; aggravating. Anyways, thanks again and I hope I am able to post that inevitable response soon...

On second thought... that first thread you posted a link to has to do with that ati3duag.dll which is not present in any of my error reports. I am also NOT getting any BSODs which usually happens when there is a problem caused by memory. I'm going to try removing 2GB anyways just to check but I have to admit I'm not too optimistic at this point. I'm pretty sure the computer would crash completely if the memory was to blame; like I said, the games continue to run in the background after the video cuts out.

 

[EXCellR8]

Oh how things go from bad to miserably worse in a day! I tried the whole RAM thing this morning which, of course, did absolutely nothing; the computer crashed just the same as before. I have more important stuff to deal with so I just gave up and started work on one of my school projects. While working in Photoshop, the system crapped out all of a sudden and the video card's critical heat LED lit up all nice. The fan cranked all the way to 100% and I was stuck looking at a black screen again, awesome!!

Now the really awesome part When I looked into my case at the LED on the card I noticed something else. There was a nice big drop of semi-congealed coolant smack dab on the back of the card. Upon further inspection, it seems that one of the seals on the chipset waterblock was leaking, fabulous. I took the side panel off the system and was not surprised to see that the residue didn't stop at the video card. The coolant proceeded past the first PCI-E slot and down to the second, dripping over the CMOS battery in the process.

So, I pulled the video card out and cleaned up the mess. It wasn't that bad but the stuff was a little hard to remove in some spots. There was a little bit on the video card slots and on the video card interface but I dried it all up. I also torqued the hell out of the plastic waterblock bung so that it (hopefully) wouldn't leak again. I'm not even sure when the leak started but I'm glad I noticed it when I did.

So yea that was my little project for the day, not really what I felt like doing. Believe it or not, I don't think the escaped coolant is responsible for the video card's behavior. I did however notice that the PCI-E 2.0 chip on the mobo, near the northbridge, runs absurdly hot! I can't even put my finger to it. I had glued a small-ish Foxconn passive cooler to this chip with some Arctic Allumina thermal adhesive when I installed the chipset waterblock. (The stock heatsink for this board covered both chips) I think the chip might just run hot because I remember the stock sink becoming quite hot as well.

I'm really starting to miss my AthlonXP/NF7-S setup, that was such a reliable system. I just can't win with this one, it's been nothing but problems. I don't think I can return most of this stuff either, so I'm probably stuck with it for awhile. I guess I can RMA but that's probably not going to get me anywhere. What the hell am I gonna do?

 

[kimsland]

Thanks for the update (even though not that fantastic)

Well I'll post this link, because it certainly applies here:
Have a look at ArcticSilver Instructions

But, in actual fact the Video card has its own CPU and Heatsink, which one member did also try re-pasting with positive results. An easier option would be to test with another Video card, or even remove the Video card and use onboard video instead (if exists)

Hmm. Just thought of another fabulous link, that I'd probably do as well: BenchTesting

It really seems to me, that even if you feel it may be unrelated, I believe this fault should be removed. It may in fact be a combination of two issues, one being the CPU and heat.

 

[EXCellR8]

Bench testing, while a great idea, would be extremely difficult with my setup. I have an external radiator solution and coolant lines that travel outside of the case. The link below is to a picture of my current setup. If you look just above the coolant line sitting above the video card, you will see that passive cooler I glued on over the PCI-E 2.0 chip. The chip is relatively small so I didn't use a whole lot of the Arctic epoxy. I did make sure that it cured correctly and that the sink was seated properly on the chip.

My System

That little heat sink is more than hot enough to burn the finger so I'm wondering if it could be to blame for the video problems I've been facing. I really appreciate your help in all this; thanks again for the advice.

 

[kimsland]

I see the problem, you have all these green pipes stuck everywhere :haha:

No I know you are upset, so that wasn't nice, but anyway just follow the repasting idea, probably forget the Bechtest until last resort (which is very close anyway!)

 

[EXCellR8]

^Ha, yeah the whole liquid cooling thing came back to bite me in the ***... I think I fixed the leak though. It's all good buddy, I'm trying to find some humor in it myself as it's been quite the mess to deal with lately. I think I'm just gonna use it until everything breaks, then I will just b*tch and RMA everything. There's no way I can afford any more stuff so that seems like a reasonable solution. So, until the crap hits the fan on this one, I'll be enjoying my broken graphics while waiting for a coolant line to rupture I wonder if my mobo can support 2000FSB, let's find out... j/k. lol

 

[kimsland]

Thanks for the update :grinthumb

Wait, that wasn't an update

Oh jokes, Oh I get jokes :haha:

Hang-on you're serious, aren't you?

OK have fun

p.s. That would be fun.

 

[EXCellR8]

Sorry to bump this but think I may have actually fixed the problem! It was either one of two things because I ran Fallout 3 today for 6 hours straight w/o any crashes, which I was never able to do before. Anyways, I took out a different pair of RAM sticks leaving 2GB again and installed another 120mm case fan. I kinda just drilled a couple holes in my side panel to mount the fan over the VGA vent but it doesn't look that stupid lol. I wanted to see if I could cool down that passive cooler that is above the video card by bringing in more air. My theory is that the ambient temp inside the case was eventually heating up until that PCI-E 2.0 chip failed, thus resulting in the video loss. This would probably take a couple hours to happen so that's why I couldn't play games that long. Of course that could all just be crap and the problem was actually having 4GB RAM using 32-bit Windows... lol.

I guess I'm just gonna have to wait and see if I have any further problems. I am not 100% sure that it's fixed as of now but I'll be glad if I don't have to buy another mobo or video card. If it does turn out to be the RAM, I guess I'm losing 2GB but whatever. I might just sell all 4GB and get a couple 1GB G-Skill sticks or something. So yea, thanks for the help and I'll post back if anything happens.

 

[kimsland]

Oh, my recommendation in Post#7

Thanks for the update :grinthumb, you can never have enough fans I say

Info on Fans: https://www.techspot.com/vb/topic114492.html

A picture that helps you laugh:

 

[EXCellR8]

Sh** crashed on me yet again this morning so I filled out the RMA form for the video card. The stupid little critical temp LED lit up again and the fan went crazy. I think the video memory is overheating or something because the core runs at 40c load! Hopefully they just send me a working replacement and not inspect this one because I've already taken it all apart and stuff. I'm just waiting to get a response back from the RMA dept. giving me the green light to send it in. Hopefully I have a new one by next week. I was really starting to think I had this damn thing figured out too... stupid computer parts and their nonsense.

 

[kimsland]

I agree there's no straight line
Data and power gets distributed in all directions, a little bit here a little bit there
Anyway, please keep us informed, even if it takes a week

 

[nstutt]

This is an old post and I apologise for dragging it up again but it is important for people who like me have just bought a Sapphire HD4870 card and have exactly this problem.

I was playing Fallout 3 on my system with my old Nvidea 7600 GT quite happily albeit with quite low settings and slow downs when using VATS indoors. So I decided to upgrade my card to the best Radeon currently available.

I started playing Fallout 3 and it crashed after 10mins or so with a black screen but seemed to be still playing the game without the monitor on. So this means absolutely it is a graphics card issue. I looked around online and was pretty sure my card was overheating. I tried turning down both the frequencies in the card's control panel and managed to play without any crashes. So its pretty obvious to me that the default settings this card comes with are not acceptable on pretty much any system (as my computer is pretty well ventilated). I think they have been increased to get artifially good benchmark ratings or something.

I then tried Halflife 2 and it crashed straight away even before it had loaded properly. Finally I tried Civ 4 and it crashed the same way after about an hour.

Anyway its obvious the card runs very hot, the fan doesnt seem to work very well on auto setting either ie it never kicks in until after the card has crashed! It will increase temps in your case generally and its a massive card as well. So before you buy make sure you have lots of space available and lots of ventilation. Then you will have to manually reduce the frequencies it runs at to suit your system. You may also want to manually change your fan speed to a constant 30% or so (35% or greater is noisy) to ensure your card is cooled all the time.

The good news is: my original set up when assessed in Fallout 3 was coming out at a low graphics setting recommendation. With this new card, even set at minimum speed for increased stablity, the recommended graphics setting was everything at ultra high with 2048x1024 res and 6x AA 15x AI. I can now see enemies miles away to snipe at them and can avoid unwanted fights whereas before they always pounced on me from close up. I can even watch fights going on between the AI enemies from a distance - superb. And the game is really smooth.

So in summary: buy the card, but be prepared to reduce the frequencies and alter the fan setting manually for stability. Its very fast and absolutely worth the purchase. But Radeon want their butts kicking for releasing a card in its default unusable state IMO.

Hope this helps everyone with similar problems to mine.