|
Recently I've been having some weird problems when I play games. I think it may be due to the physical state of my graphics card, although I've read that it might be due to the new NVIDIA drivers?
Basically I can do normal tasks such as web browsing, watching streams and movies, MS office, etc. just fine. Up until yesterday, I have been able to play LoL with no problems either. But when I load up SC2 or Bioshock or any graphically intensive game, my monitors freak out, go black, and reset, showing a notification that the drivers failed and reset. Sometimes I manage to quit the game before it happens again, and it's fine, or else it'll keep doing this and then hang. This was a pretty reliable result, and so I stopped running these games. This was fine because I didn't game much and when I did, I only played LoL.
Yesterday I've started experiencing these crashes while playing League, and it's happening pretty often now. What should I do? Do I have to buy a new graphics card?
Currently using:
MSI GTX 560 Ti
GeForce 335.23 Driver
Sample of what happens during these crash/resets (this one ended up freezing like this)
+ Show Spoiler +
Whocrashed on the crash last night:
+ Show Spoiler +On Wed 6/4/2014 5:45:36 AM GMT your computer crashed
crash dump file: C:\Windows\memory.dmp
This was probably caused by the following module: dxgkrnl.sys (dxgkrnl+0x5D140)
Bugcheck code: 0x116 (0xFFFFFA800B2FC4E0, 0xFFFFF8800FA42E2C, 0xFFFFFFFFC000009A, 0x4)
Error: VIDEO_TDR_ERROR
file path: C:\Windows\system32\drivers\dxgkrnl.sys
product: Microsoft® Windows® Operating System
company: Microsoft Corporation
description: DirectX Graphics Kernel
Bug check description: This indicates that an attempt to reset the display driver and recover from a timeout failed.
The crash took place in a standard Microsoft module. Your system configuration may be incorrect. Possibly this problem is caused by another driver on your system which cannot be identified at this time.
|
Well... I'd bet on a hardware problem after seeing your screenshot.
If you want to try a different driver, something to try is the 314.22 driver. You can't find it through the driver search page of the NVIDIA page any more as it's pretty old by now, so you need to use Google to find a download link.
There's also a newer driver, 337.88. Perhaps try that one first as it might have fixed something compared to the 335.xx you use currently.
You might also want to play around with overclocking your card. Increase the voltage used by 0.05 V for example (that increase is already a bit aggressive). That might make the GPU run stable. Decrease the memory clocks by 100 MHz for example. That's what could make memory stop producing errors.
|
I tried the newer 337.88 driver, and that did not do anything.
I overclocked the GPU like you said, increasing 50 mV and decreasing the memory clock by 100 MHz. That seemed to help a bit. I started a custom hero game on SC2 and after about 5 freeze/resets it stabilized and I was able to play through the rest of the game. I was never able to do this before; it would always end up crashing.
Is there anything else that you suggest I do? Should I further increase the voltage and decrease the memory clock, and if so, by how much?
|
Better don't increase the voltage more than this. Instead go down with the core/shader MHz.
The voltage is only for the GPU chip. I don't think it does anything to the memory chips. If that's true, you can only reduce speeds for the memory.
If you can fix things, try to find out what is causing the issues, the GPU or the memory, then only reduce the speed for that one and set voltage back to normal if you find out it's the memory.
If it really gets better doing this, I guess that's proof that the hardware is starting to break down?
Did you check temperatures? There might be something off with the cooling, one of the of the fans running slow or not at all, or the thermal paste not being good any more.
|
a) Try rolling back your drivers.
b) What power supply do you have?
c) Is your GPU fan clogged up with dirt or barely spinning? We'll need to check temps.
d) Kindly list all your other specs.
|
i5 2500K Sandy Bridge
GTX 560 Ti
Corsair 80GB SSD
P67 Motherboard
8GB RAM
WD Black 500GB HDD
Antec 550W PSU
Built the computer about 3 years ago. The graphics card had some problems a couple years ago, most likely due to my computer getting banged and moved around a few times. I sent it in for a RMA since it was still under warranty, and they sent me what I hope was a new one, which has been running smoothly until lately.
|
Any update on a) and b)?
You can use things like MSI Kombustor, GPUTemp to monitor your temperatures.
http://www.gputemp.com/
|
I haven't tried rolling back the drivers yet, because since doing what Ropid suggested it's been enough to run LoL without any crashes. When I load a custom SC2 map, it will crash/reset about 5 times before stabilizing, at which point I can play the rest of the game without any problems.
I think the GPU may be reaching the end of its life, and I'm probably going to just build a new computer soon anyways. If I can overclock it to keep it going for a couple more months that's all I really need. Between now and then the only thing that may stress the GPU is League, so as long as it can run that it's fine with me.
![[image loading]](http://i.imgur.com/MjYXxqI.jpg)
The crash/resets that were happening:
+ Show Spoiler +What was happening before:
|
I can't decide how to judge the screenshot with regards to temperatures. I also have a GTX 560 Ti, but replaced the cooler and don't remember how good the original one was, don't know if those 51 C for 0% load are strange or not. If room temperature is pretty high where you are, I can imagine that the 51 C are totally normal. It also doesn't ever get close to 100 C if I'm interpreting the screenshot right, so things should be fine.
You might still want to take a look at the card's heat-sink and see if it's stuffed with dust, also check if the fans are turning. If dust is the reason for the problems, that would be pretty nice as it would just need some cleaning.
|
FWIW, you probably don't need a new computer at this point -- a 2500K isn't obsolete by any standard. I have a 4770K and a 2400 and can hardly notice the difference in games.
Based on the graphs, the crashing starts when the GPU utilization hits 100%.
ASSUMING this is not an issue of drivers and a clogged up fan/heatsink, this to me can be 2 possible things:
a) PSU is dying
b) GPU is dying
Do you have any spare hardware lying around you could use to diagnose? If you can confirm the GPU dying, you can actually bake it.
|
United Kingdom20326 Posts
FWIW, you probably don't need a new computer at this point -- a 2500K isn't obsolete by any standard. I have a 4770K and a 2400 and can hardly notice the difference in games.
While a 2500k, especially at OC but still when at stock is a great proc, you can surely notice the difference between an i5 2400 and OC'd Haswell.
You're talking 3.1ghz at four-core load, i've been running at ~51.6% higher frequency and Haswell is like 10% to a bit over 20% faster in typical loads (10% min really, to around 20% faster clock for clock for video encoding/streaming) so i'm sure that would make a serious dent in performance for CPU bound stuff - we're talking like 1.8x performance under video encoder by that point. A 2500k, at OC, is way closer to an OC'd 4670k/4690k than a 2400, though.
Also, there's the trick you can do with sandy bridge, even on non-oc motherboards/cpu's(?) to make the turbo's run 400mhz faster, either sometimes or all of the time (depends how power limit kicks in and if you can change it) which helps get more performance if you're still at stock CPU clocks
I wouldn't realistically replace a 2500k without being a hardware enthusiast, until Skylake at the end of next year with ddr4. By then, stock sandy bridge would be really far behind skylake at overclock and hopefully, so would the standard 2500k@4.5ghz setup.
In terms of thread, i agree, temps fine etc. Hardware issue with GPU, maybe psu
Between now and then the only thing that may stress the GPU is League, so as long as it can run that it's fine with me.
I don't think this should be the case, i double checked hardware utilization with all maxed aside from Shadows in league a few days ago and saw like 60% load on my 770 with 400FPS. Load was way lower when i was cpu-bound to less than half of that FPS, later in the game, IIRC, because the GPU had less work to do because there were half as many frames and they didn't take twice as long each to make up for that.
|
To wrap this up, I bought a new graphics card and everything is great. No more problems. I bought an ASUS GTX 750 Ti for 160$ and everything runs as well if not better than before.
Thanks for all the responses. It helped a lot!
|
the problem is not the 2500K, its the mainboards that came with them, many do not work with the new graphic cards like a R9.
They would at least need to be Bios flashed to ivy bridge, which might mess up the easy sandy bridge OC.
Also, i did that with my MSI board with official MSI bios flash to get a R9 running and it killed my mainboard.
good to hear your things worked out, maybe next time i have to switch my graphics card to nvidia as well!
|
On June 12 2014 07:35 LaNague wrote:
the problem is not the 2500K, its the mainboards that came with them, many do not work with the new graphic cards like a R9.
They would at least need to be Bios flashed to ivy bridge, which might mess up the easy sandy bridge OC.
Also, i did that with my MSI board with official MSI bios flash to get a R9 running and it killed my mainboard.
good to hear your things worked out, maybe next time i have to switch my graphics card to nvidia as well!
what.. That makes no sense..
|
|
|
|
|
|
EPT![[image loading]](http://i.imgur.com/VMjQ77z.jpg)
