Upgrade it or junk it - yeah, it's that bad now

[b0nes]

'Evening, folks.

I am on the verge of junking my G20AJ. I wouldn't even donate this thing to someone in need - it would be that rude. Before I progress to that stage, though, I want to take one last stab at salvaging this complete disappointment of a machine.

Background

Purchased it over the New Year's holiday of this year. I was initially extremely impressed - sexy looks, exceptionally quiet, and able to take everything I could throw at it.

Three months into owning the machine, I was forced to re-install the OS. Windows had never been able to perform a proper update; I found myself running system restores and recoveries at a rate of once every two weeks. "Silly Windows 8," I thought. "I'll just wait until 10 comes out later in the year." Unfortunately, one particular failed update completely locked me out of my system. In a situation described in detail elsewhere on these forums, I was unable to boot into even something as fundamental as safe mode. System restores would be blocked by my anti-virus. Nothing I did to try and recover the system worked, so I was forced to do a complete reset. Fresh OS install, Windows 8.1. Got all my drivers in place. Everything was fine for about a month.

Now, chaos.

The Problem

Computer threw a panic and crashed while playing Crysis 2 approximately three months ago. It would always crash at the same point, so I figured perhaps there was something up with the game itself. After one last crash, I pulled up Speccy and took a look at the temps. It came back as ~60 degrees Celsius for the GPU, and 126 degrees for the CPU. Lolwhut?

I downloaded several different CPU monitoring programs, and all returned similar nonsensical values. Best guess? They must have been summing the four core temps together. There's simply no way it was running at past the boiling point of water with everything else running fine: fans were definitely operating, everything else ran smoothly (no hiccups or artifacts of any kind whatsoever), and my cooling arrangement is impeccable. (I have it so that the A/C vent blows air up behind and underneath the unit, so cooler air gets drawn in from the intake vents on the underside of the chassis). I could even run Civilization 5 (which while not as demanding as a FPS, still puts a very heavy draw on the system - and I have the graphics maxed out to a T). Verifying temps via the BIOS returned normal values in the 25-50 degree range.

Yet, the issue comes and goes. I've gone weeks without problems, then it returns again: I will be playing a game (primarily Day Z standalone), then after 2-15 minutes I will heard the sound scramble, then the sound will return to normal while the monitor shuts off. It's as if the system is experiencing the symptoms of an overheat, but none of the usual results actually happen. There aren't any screen artifacts, no troubles running some games or other light applications (like HD video on YouTube or Amazon), and the system never truly shuts down - I always have to perform a hard boot.

Things will go haywire for days or even a week at a time, then all will be okay again. Except now, even the staple program - Civilization V - crashes the system. It's as if the problem rotates from one game to another.

Again, normal operations do not cause the system to panic. Only when a load is placed upon the GPU (i.e. playing a game) do I encounter a crash.

I have researched this issue extensively and come across a litany of different "fixes". I've tried them all, and at best they work for a few weeks before the issue returns:

- Updating the BIOS helped for a month (and ensuring all settings were set to default), then the problem returned
- Rolling back the NVIDIA drivers helped for precisely two days
- I can't say if Virtuwatt was an issue, because I don't have it installed on my system (to my knowledge). It was not part of the driver package that I installed after I had to reinstall the OS.

Analyzing crash dump files reveals the dreaded "nvlddmkm.sys" error, but nothing I have tried has been able to rectify this issue. I swear, it's worse than cancer. At the very least, it is the only tangible error I have come across in this entire horror story. Yet, nothing works.

What now?

I have had it with this system. I cannot take the unreliability any further. I've wanted to be able to use this system for professional uses (I'm an architect, so I run heavy duty CAD and rendering software), but I refuse to risk losing literally valuable files being lost because of this machine pitching a fit.

I have read (and possibly suspect) that the issue is the GPU. I have a GTX 760, and I understand that they have been twitchy. I'm prepared to invest in an upgraded GPU before purchasing a brand new machine. But I hesitate to spend ~$500 for a GTX 980 on such an unreliable machine.

So, do you think it would be worth of me to try the upgrade? I'm still under warranty, but I don't care anymore. I have no desire to send this into ASUS and experience everyone else's result: two weeks without the machine, then a week of solid performance before the issue returns. In the words of my generation, "nope."

Is there anything else I can try to recover this machine? Anything I haven't thought of?

Anything?

[xeromist]

WRT warranty service, keep in mind that the "everyone else" you are referring to is an echo chamber of people who are posting because they are having problems. That's not to diminish their grief but happy people seldom post about it online. In other words there are plenty of people who get their stuff fixed. In your case I'd agree that you have a higher likelihood of service being unable to identify your issue (being intermittent) but if the alternative is $500 which may not fix it, I feel it's worth a try.

Is there anything you can do to reliably reproduce the problem? Furmark maybe? If there's something you can give the service dept to reproduce the issue that would dramatically improve your chance of a fix.