I've downloaded all the latest drivers but that didn't make a difference. Besides if this was a driver issue I would expect it to start immediately during boot up (not 5 minutes later) and occur when in Safe Mode as I can run all day long while in safe mode??? I haven't yet touched the BIOS code as that will be my last area to attack. However, got to be careful with this as one mistake and your PC turns into a vegetable:):)
Ok, I dunno if this'll help anyone or not, but I've been having these issues too, and I've noticed a few patterns in them.
Firstly, there have been occasions where I've had BSODs. In these cases, the sound still hangs like usual, and the message displayed mentions hardware performing a malicious function.
Not only that, but there have been times on windowed mode when I forget to underclock, and my computer manages to catch itself without crashing by restarting the Nvidia GPU. I get a popup on my taskbar telling me the Nvidia GPU stopped responding but recovered.
So that narrows it down from hardware to the GPU. Well... there is one more instance that has occured that may answer more questions.
On the rare occasion, I have been able to put my laptop into sleep mode during a freeze. I can then start it up again and it works fine, UNTIL an application that activates the Nvidia GPU does any hint of graphics-intense work. In which case, it instantly freezes again. On a full restart, however, it's fine again.
And finally, I notice that the freezing occurs most often when my laptop is in the higher temperature ranges.
What this all says to me is that the issue is in some way RAM related, as sleeping your computer generally means storing RAM and then restoring it so that your session resumes from where you left off. And so, I believe the issue is in fact with the Nvidia card's built-in memory. As to what causes the memory issue, and why this problem happens so often with this specific model of laptop, I cannot say, but this to me is the most likely scenario based on my findings. However, if I'm wrong, and someone can make a more accurate deduction from this research, then I will happily admit my mistakes.
SOME GREAT ADVICE from lots of very patient users... we all deserve medals.
This is a real b***. A machine allegedly gaming-friendly, and users suffering system instability (sometimes) when just using the OS (with no heavy G-load).
I've tried all sorts, and think I'm close to a workaround...
I don't know if reducing GPU core + memory clock results in lower voltage, or what part shader clock plays, but the underclocking fix seems to be potentially viable...
I have GT540m, wiki says default is
core / memory / shader
but I can't increase my memory clock above 1080 with nvidia system tools.
I've seen a few things (in this thread) that might also be relevant (turn off AA, threaded optimisation.... vsync even) and this is what I'd like to add:
(for me) Underclocking (nGPU, defaults 672,900) to 550,850 (using NVIDIA System Tools) is a fix... until the program "demands" more power(!!), then the gfx-crash occurs
Changing the image settings to "balanced" - not quality / 3d managed - may work
(ie keep GPU core/memory clock @ custom)
Running XBMC/Rocksmith on samsung via HDMI (always uses nvidia) - exclusive fullscreen mode can cause problems
If I can find a way of ensuring applications don't force the NVIDIA GPU core/clock back to (too?) high defaults... I think stability can be... "introduced".
One thing I don't get - opencl / webgl... the embedded intel hd chip doesn't support (but nvidia gfx does?)... even when running chrome with nvidia gpu, the browsers "sees" the embedded chip (like gpu-z on startup) and identifies openCL isn't supported, and disables...
ie chrome problems
still testing (nvidia settings) - will post back with potential solutions if I find any...
does anyone else have any issues with NVIDIA (.nsu) profiles "not sticking?"