Hello, I'm back! Thanks for the continued assistance! I'll reply to everyone now, I made sure to try every suggestion before posting again, to update on the situation.
Try to find updated drivers for your monitor.
I searched updated drivers since the one that Windows 7 installs is from 2006. I didn't find anything so I tried the next best thing, using the "Generic PNP monitor" driver. It didn't make a difference. I'm running with the generic driver now.
Are you using a dvi to d-sub adapter or anything like that? I had a problem once with an adapter. It wouldn't let me run the native resolution.
Try a different cable?
I'm not using any adapter. I use the DVI out from the monitor and my card has only 2 DVI ports. It was complicated, but I got another DVI cable to try your suggestion. Didn't make a difference. Tried both card ports just to be sure.
I didn't knew that (about PhysX) thanks for clearing me that
I am beginning to think, after your last post, that may be that game issue, now do you get issues only in that game or does happen in another one? Lets see how goes in Fallout 3, but I would suspect the monitor, if you can check every resolution in Windows and see if there is some who don't work
Haha, you're welcome! I'm glad I could share some knowledge with you too!
Reinstalled Fallout, DLCs and patches. Tried a game and it seemed to work Ok but eventually it froze on me. I forgot to open GPU-Z so I tried again, this time with GPU-Z logging. I played 1 hour without problems...
I checked temps and max GPU and MEM was 83º C. I'll try a larger game session at night to see what happens. I really wonder what causes this...
Windows resolution doesn't have problems. Went from 800x600 all the way to 1680x1050 with no issues.
When you lower resolution, temps are lower too, isn't it? I think it might be related to overheating, I'm thinking about the vram, try lowering the clocks on the memory. Be generous when lowering clocks, it's just for testing. If that fixes the problem it could be the memory, see if vram modules are properly cooled and in the worst case keep the vram clocks lower than default until you can find a solution. It's the best I can come up after you've tried everything else suggested here.
Bear in mind that games like Metro, Crysis and many many other modern games stress the GPU a lot, but due to the graphics processor being the bottleneck, memory is not necessarily working at 100%. Games like Fallout, Borderlands and Serious Sam use very old 3d engines, so they don't do as much shader computation and hence the bottleneck is created in other places, i.e textures being loaded into memory.
EDIT: Also it could be your main memory that's causing that. I had a similar problem where the PC crashed when playing some games or doing some other stressful tasks. In the end the problem was apparently that the memory modules (Corsair Dominator) required higher voltage than what the mobo gave by default. Changing the voltage to the one specified in the modules fixed the problem. The sticks should have worked on (JEDEC) default settings, but maybe because they were OC oriented modules or maybe because the mobo was a cheap one that I bought as a replacement, they wouldn't work with default vmem, not even at the stock clocks of 800 Mhz. At the specified vmem the sticks would do 1066 Mhz and then some though. Bottom line, check your main ram too just in case.
Yeah, it's the way Ageia decided for their API and Nvidia hasn't changed it. It's kinda like Java, you need the JVM/JRE installed on your computer in order to run Java code.
Hey, thanks for the long post! You are correct, with lower resolution, the temps are lower too. And another thing that maybe you're right, normally is the GPU limiting everything and older games (Or better optimized) are stressing the mems and causing the problem. If Fallout 3 cases, I was surprised to see that the GPU max load was 70 %. Memory usage stays between 400~550 Megs. I thought it was more GPU intensive.
As I said, the memory chips don't have any kind of cooling. No fans, no heatsinks, nothing. The temps are suspiciously similar to those in the GPU, so maybe the thermal diode is not reporting accurate memory temp. I didn't tried to lower the clocks yet, but I added a fan near the chips to see if it helps. It didn't made a difference but the chips gets really hot.
As for my main memory, I removed everything except for a module. I have 3 4 GB modules, only one is really mine, so I removed the other two. I'm running with 4 Gigs now, just one Kingston module. So far, Fallout 3 is working fine. Serious Sam keeps crashing. I'm starting to think that my hardware config could have a problem with the game. It's crazy, Borderlands works flawlessly and I didn't even changed anything in the game config.
Will update when I finish testing. I'm now more interested in knowing what could cause this behavior rather than solving the problem
Million thanks to everyone!