Something is wrong and I think it is the vid card...

[FordGT90Concept]

So I was playing The Sims 3 yesterday or a few days ago and the display just went black without warning. The monitor was still on, music was playing, ctrl+alt+del wasn't working, and I had to hard reset the comptuer.

Today, the same thing happened but it wasn't just a black screen. It was white/gray strips (really hard on the eyes):
http://img.techpowerup.org/100628/100_7641.jpg

Ctrl+alt+del still didn't work, music was playing (switched tracks even), and I made sure the system was still responsive by accessing it's shared files from another computer. Both times it happened, it was abrupt and without warning.

It has only happened so far in The Sims 3. I played and beat Singularity. It crashed to the desktop once but never locked up like this.

I expected GPU recovery to kick in, but it didn't. I also expected BSOD, but it doesn't.

I am running Catalyst 10.6.


WTF is going on? This card just got back on RMA like a week or two ago. Does it need to go back again?

[TheLaughingMan]

I am inclined to say the GPU just over heated, but your card should mud stomp Sims 3. It may be some kind of glitch with Sims 3. I would check online to see if anyone else is having trouble with that game and DX11 or 5000 series ATI cards.

Try another game you know is graphic intensive if you have 1 or a rendering test set on loop and see if the problem happens again. If it doesn't fail, it is the game. If it does, make sure the fan is speeding up with the temp rises and try going back to 10.5 if this just started happening.

[Mussels]

bios update the card. its a known weird issue with early 5870's

[FordGT90Concept]

Originally Posted by Mussels

bios update the card. its a known weird issue with early 5870's

Wha? Won't that void the warranty? Also, it is a Vapor-X non-reference Sapphire design. I don't know if it would still classify as "early."


I think I'll run FarMark for an hour and see how that goes...

[Mussels]

Originally Posted by FordGT90Concept

Wha? Won't that void the warranty? Also, it is a Vapor-X non-reference Sapphire design. I don't know if it would still classify as "early."


I think I'll run FarMark for an hour and see how that goes...

it was some issue caused by low 3D/UVD clocks having incorrect voltages/mem timings whatever. ATI keep trying to fix it in drivers, but having little luck.

my brother had similar issues, BIOS update fixed it - so give it a shot. no it wont affect warranty if you use a sapphire BIOS.

[FordGT90Concept]

Originally Posted by Mussels

it was some issue caused by low 3D/UVD clocks having incorrect voltages/mem timings whatever. ATI keep trying to fix it in drivers, but having little luck.

my brother had similar issues, BIOS update fixed it - so give it a shot. no it wont affect warranty if you use a sapphire BIOS.

Is there a guide somewhere for doing that?


Well, I started FurMark and in less than a minute (temps were probably around 55C) it did the same striped thing I took a picture of before (just different shades of orange instead of shades of gray). It went back to normal after about 3 seconds. I sat there with the camera and waited for it to happen again for about a minute and it didn't. It reached 69C before I walked away (I think it was).

I told it to log the tempatures but I can't find where it placed them.


After about 15 minutes from starting it, I came to check on it. The computer seemed to have locked up--the monitor was off and it didn't want to come back on. I switched to my server on the KVM to verify the monitor wasn't to blame and, of course, it was perfectly fine. The video card simply wasn't giving it any picture.

[Mussels]

it could well be a bad card or a bad PSU (my bet is card, of those two) but you might as well rule them out since you had problems with your previous cartd that you RMA'd.

Update the BIOS (i never flashed a 5K card before, google it), and try another PSU if that doesnt help.

[FordGT90Concept]

I did FurMark again but this time in benchmark mode. Within about 10 seconds, it went to the stripped. Stayed there for about 5-10 seconds. Flashed, then went black. I let it sit there for about 30-60 seconds and it didn't come out of it. Since I theorized the OS was still responsive. I hit the power button to make it shutdown properly. Lo behold, it snapped out of the black display and quickly shutdown. During that hole process, I bet the temp on the card never even hit 50C.

If the PSU was to blame, wouldn't the system not be responsive enough to shutdown?


The original card I RMA'd was due to CTDs in virtually every game. The fan on it was also getting loud on cold start (starting to fail).


I think I'm going to open a dialog with Sapphire. If they say BIOS flash, I'll follow their instructions. If they RMA, I'm going to demand XMA (another 3-4 weeks without the card, hell no).

[jmcslob]

i had a similar issue before my PSU went pop
I was getting screen tears and whatnot and my Pc became unresponsive..

I'd try the GFX BIOS update if it still does it try that system with another PSU if it still does it
Try updating the BIOS on your MOBO if it still does your GFX card is likely defective

[FordGT90Concept]

I don't have a spare PSU that could power my system. The PSU is pretty old so I wouldn't blame it for dying. I already opened up a ticket with Sapphire--see what they say.

[Solaris17]

yiou should try a different driver set.

[SimJett]

Sims 3 and Graphics cards has been a can of worms since day 1. The so called GPU throttling, producing "extreme" frame rates, leading to excessive temperatures etc... Nvidia cards you could Vsynch "force on" through the Nvidia control panel. ATi users had to find a 3rd party tool, "fps limiter" http://www.simprograms.com/fps-limiter-and-the-sims-3/ being one possible solution.

So if this is all old news, and temps and frame rates are / were being monitored with acceptable results, then possibly a bad card or weakening PSU. Since Ambitions, I have noticed, but did not read any of the "many" going to black screen threads on the EA forums.

[Jstn7477]

I had practically the same issue (with bluish colored stripes) when I played Crysis on my overclocked XFX 5770 board (would always seem to pass Furmark OK). I went a bit crazy with the 3D memory clocks, and it would lock up after a few minutes if the memory speed was >1300MHz QDR (5.2Gbps). I then discovered that my board had Hynix ICs that were rated for 1250MHz (5Gbps), so I backed my 3D memory speed down to 1250MHz and encountered no more issues. I'm not sure if this is your exact issue as your card is probably at stock, but I thought I'd share my experience.

[FordGT90Concept]

According to FurMark, the card is running 875/1250 which is a slight overclock (850/1250) from reference (I didn't know that). I'm going to use ATI Overdrive to bring the clockspeed down to stock and see if it still does it...

[segalaw19800]

turn down memory clock by 25mhz but keep gpu clock at that setting.. for testing

[FordGT90Concept]

I put it to 850 MHz and 1200 MHz (stock) and it lasted 45 minutes without error (as far as I know, I checked it about every 10 minutes). That was at 1920x1200 full screen and 8x MSAA. It averaged 29 FPS and the temperature bobbed between 69-70C (never exceeding 70).

I'm going to run it again with BOINC running in the background. If it starts acting up, then the PSU might be insufficient...

[segalaw19800]

sound like a plan

[FordGT90Concept]

It ran for 30 minutes and the results mirrored the first. Average FPS was 26 and max temp was 70C.

So...that answers it. They gave me a factory overclocked card that is unstable at those clocks. I'm going to push that they fix it ('cuz it's effin' ridiculous). Oh, I'm also going to mention the card they sent me was obviously a refurb having the bracket bent out of shape to the point I HAD to bend it back to get in the case.

[Mussels]

Originally Posted by FordGT90Concept

It ran for 30 minutes and the results mirrored the first. Average FPS was 26 and max temp was 70C.

So...that answers it. They gave me a factory overclocked card that is unstable at those clocks. I'm going to push that they fix it ('cuz it's effin' ridiculous). Oh, I'm also going to mention the card they sent me was obviously a refurb having the bracket bent out of shape to the point I HAD to bend it back to get in the case.

thats where the BIOS updates come in.

[FordGT90Concept]

I'll only do a BIOS update if that is what they tell me to do it. If the card can't take a 25/50 MHz, that tells me it isn't going to last long in the first place so I wouldn't complain if they want it back. It's liable to die just outside of warranty as-is.

[Mussels]

Originally Posted by FordGT90Concept

I'll only do a BIOS update if that is what they tell me to do it. If the card can't take a 25/50 MHz, that tells me it isn't going to last long in the first place so I wouldn't complain if they want it back. It's liable to die just outside of warranty as-is.

its not the 25Mhz it cant take, but that its likely just got the wrong voltagse set at stock. when you change the clocks its over-riding them with voltages set by the CCC, not by the BIOS (changing from low 3D volts to high 3D volts, etc)

[FordGT90Concept]

Althonmicro told me to flash the BIOS and I did. It is still not stable at the factory-overlcocked settings. There is probably a bad memory chip on the card.

[overclocking101]

dude, use gpu-z to dump your bios to desktop, put it in a .zip or .rar file, post it up one here, ill whip it up for you, it will hold those stock clos and then some ati bios are nothing new to me, ive done this for numerous guys on here without issues, the problem is not all cypress are created equal and sapphire,xfx,asus you name it didnt take that into account they just loaded the same exact bios on all cards and shipped them out. at least post it up and let me look at the bios ill be able to tell you the problem and more than likely fix it if you wish

EDIT: no way is it a bad memory chip if it was you would have artifacts nothing more, and it would be in 2d/3d etc, bad memory chips dont work at all on gpus

[Mussels]

Originally Posted by FordGT90Concept

Althonmicro told me to flash the BIOS and I did. It is still not stable at the factory-overlcocked settings. There is probably a bad memory chip on the card.

its unfortunate that it didnt help. let OC have a go at it with modding your stock bios, and if not, go for the warranty.

[inf3rno]

I never updated my BIOS but I had the same problem with ma HD4850. And PSU was the culprit.