I post this out of the 2 DA threads on the off-chance that other players may be having the same glitch, and that the glitch and my baked video card were related.
My E8400/GTX9800-512mb ran DA2 fine with drivers a few months old. Like a fool, I decided to upgrade to the latest and greatest a bit earlier today. After that upgrade I started to get a severe slowdown issue in which the game would stutter to a stop, then take 10-20 seconds between frames and so on. (Subsequent search suggests this is an issue coming up with the newest Nvidia drivers and older 98/8800 cards.) The second time this happened I by chance was in the middle of a cutscene and was called away from the computer. Stupidly, I didn’t heed the part of my brain that said “you might not want to leave your computer like this,” and I attended to other matters for about 20 minutes, returning to find the game still glitching.
Rolling back the driver got rid of the bug decisively, and I played for a while after that. A few hours after that, however, my graphics card has messily died. (Severe artifacts even while booting, etc.) I manage dust and airflow carefully and have had no issues with the card. Everything’s stock and non-OCed.
It could just be a massive coincidence but my teary-eyed advice to anyone with the severe stutter/slowdown issue is to not muddle through it. My advice to morons like myself is to not leave a mysterious and possibly performance related semi-hang running for 20 minutes on end. See you in a few months, PC gaming.
Correlation’s not causation but I’m (perhaps irrationally) sure I murdered it with that 20 minute long quasi-freeze-up. I guess that may only make sense in a straw breaking a camel’s back sort of way. Maybe even less, I’m speaking in ignorance. But given that it’s a known issue I wanted to warn people. And whine and cry.
Hmm, I’ve been having that problem crop up on occasion but in all but one case I let it play itself out and it recovered on its own in a couple of minutes. In one case I alt-tabbed to the desktop and shut the game down. But in a couple of cases I got “connection lost” messages, making me think it was related to that and in all the cases my video card has seemed to act like demands on it were dropping. This is with a 9600GT.
Random google hits on my now video-cardless desktop, Sark, sorry. Rolling back to an Nvidia driver from Jan/Feb or so did definitely get rid of the super slowdown glitch right away though, so even if only to improve your game experience I’d try that.
Pogo: I dunno. When I got the thing in 2008 the 9800s had virtually no advantage over the better quality 8800s. I just had a good price at my local hardware guy for a BFG 9800 GTX, and the benchmarks were as good as anything else I could afford (ie there were a few bleeding edge GT2s that were better then I think)
I guess one never believes something like this is coincidence when it happens to you. Serious glitches like that and video card failures happen to me “very rarely” and “never” respectively, so when I mishandle the former and then have the latter on the same day… Probably just the frailties of the human mind.
I haven’t had a repeat of the problems since I disabled the texture pack. It may be, at least in my case, that the problem was simply that my GPU didn’t have enough memory to handle them. At 512mb, it was below the recommended requirements for the pack and the issues seem somewhat consistent with GPU memory running low, AFAIK.
It’s like every time a huge 360 game comes out and instantly the internet believes it has a bug that kills consoles. But those consoles are basically just getting played more or fir the first time in a while and would have failed under that stress level playing any game.
It definitely sounds like your card was overheating in game, the driver was stalling and windows was restarting it causing the game to hang and then recover. In any case, you are probably just victim of the 9800’s well known hardware flaw.
Interestingly enough, Cyanid admitted to causing overheating with their Bloodbowl games in many video cards, so while “frying” may an extreme way of putting it, games certainly can tax videocards in unhealthy ways.
These big gamer videocards generate a lot of heat. If that heat is not dissipated fast enough bad things will happend. Artifacts on the screen can be the first warning (or just the game bugging), so is something serious when you see artifacts on the screen.
So, yes, is your fault.
A decent video card (in a decent built system) will run at 100% of the capacity without any problem. In fact, there are stress tests to simulate that and be sure your system is well cooled. Just yesterday i did it to my cpu to be sure the overclock i made was secure.
Games can’t run at “130%!!”, as much they can make the video card work at 100%. If your video card broke just because a game ran it at 100%, you had a problem with temperatures.
That said, there are some games that don’t use a proper ‘tick’ system so they just render as much frames as possible instead of the needed ones, that causes at 100% use of the video card while really is not needed.
Furrmark has been known to destroy cards. If you don’t enable MSAA the simple shaders used will basically be running so fast that a GPU can burn itself out, but that’s a special case where the shader cores never have to wait for the cpu or memory or other subsystems that all come into play in real-world applications. In fact, with the 6X00 line AMD introduced a special fail-safe to detect when that happens and preserve the card, above and beyond normal thermal monitoring.
I killed my video card in my laptop a few weeks back. It was a nice lesson in not trying to tax it too much with high res textures, max draw distance and highest possible screen resolution etc, especially when I went back to my old laptop and had to get used to a drop in graphical quality for a few weeks. And I wouldn’t place the blame on the games, just the fact that it was getting insanely hot (by running Crysis with very high settings).
Oh, it only cost $360 to replace, and is the laptop version of the nVidia 9800 in fact!
That should never happen if the video cards were properly engineered, though. Any GPU or CPU should come with a cooling solution that can handle any load that software throws at it. Those manufacturers had not properly tested their cards, or they were just plain cheating to save on manufacturing costs or to get better benchmark scores. Welcome to consumer electronics, gotta keep up the profit margins!
Since some experts have chimed in, can anyone see anything enlightening from these photos? The desktop booted ok for 2-3 minutes this AM, but I wanted to check inside the shroud for dust accumulation or obvious signs of trouble. (It wasn’t very dusty. I’ve never had the shroud off but I’m pretty enthusiastic with the canned air.)
Google image search seems to suggest that stock thermal paste often looks pretty messy, but it does seem to be a little thin on the chip and kind of globbed up around the edges on the heatsink.
Apparently a popular quick fix with the young folks is to strip the card down to a bare minimum and then bake it in an oven for 7-10 minutes to re-flow the soldering. The sheer perversity of it appeals, but with my luck I’d just wind up killing the mobo with my EZ-Bake video card.
Hairdryer it directly on high. I’ve fixed RROD xboxen that way.
As crazy as it sounds. Also, I’ve taken old arcade boards that are dirty and just run them through the dishwasher and left them in the summer sun to dry, no worse for the wear. That seemed crazy at first, but then I did it routinely and never had a problem. Just have to make sure they get 100% dry, so I went for a week in the summer sun.