Gigabyte GTX 1060 3GB. Games often crash. What troubleshooting can I do to help fix it?

[X82]

Firstly, apologies if this is in the wrong area.

I've thought about voltages and frequency, in the sense of lowering it (hoping against hope it could stop crashes), but I don't know the right values or really where to start.
Every single game I play will crash after a duration. The times will vary but it is always the same, play for a while, game freezes, sound keeps going for a bit and then hello desktop. I don't get any error messages and I can never find anything in event viewer either.
Running MSI Afterburner and monitoring everything doesn't show a pattern either (link below to file).

Out of desperation, I even updated the cards BIOS yesterday (it was 2 versions out) and I was hoping this would change things. Played Ghost Recon Wildlands today and the same crash happened again.
It is true that it could be anything in my case causing this, but since my problems only happen when playing games, I'm going to point fingers at the graphics card until I know otherwise.

I've ran MEM tests, all pass.

All drivers are fully updated as is the BIOS.

Using latest nvidia drivers, although that doesn't seem to matter because this problem has been occurring for over a year now and I've just gotten used to the crashes.

Here are my MSI settings, untouched after installing it.

Here is a link to the hardware monitoring in Afterburner. Saved just after a recent crash. I'm not seeing anything obvious, but I don't really know what to look for.

I am hoping that maybe there is something I can do in afterburner that can help, even if I lose frames. I don't have the money to be able to buy new components, so trying to fix what I have is the only way.
I tried to overclock it based on settings in a video (guy had the same card) and the game crashed to desktop immediately. Hell, I don't mind lowering settings and losing frames as I've said, I am just tired of games crashing. I envy streamers who can play the same game for 6+ hours and it never crashes once.

Earliest crash I got was about a minute into the game, the longest was well over and hour.

Any ideas?
To be honest, it could well be just a duff component in my machine which has been running the same Motherboard, CPU and memory since 2011. But I remain ever hopeful.

Specs for clarity

Operating System
Windows 10 Pro 64-bit
CPU
Intel Core i7 2600K @ 3.40GHz 35 °C
Sandy Bridge 32nm Technology
RAM
16.0GB Dual-Channel DDR3 @ 668MHz (9-9-9-24)
Motherboard
ASUSTeK Computer INC. P8Z68-V LX (LGA1155) 29 °C
Graphics
3071MB NVIDIA GeForce GTX 1060 3GB (Gigabyte) 57 °C
Storage
931GB Western Digital WDC WD10EACS-00ZJB0 (SATA ) 32 °C
0B SAMSUNG HD204UI (SATA )
931GB SAMSUNG HD103SI (SATA ) 24 °C
238GB M4-CT256M4SSD2 (SATA (SSD))
Audio
Realtek High Definition Audio

 

[Danny75]

What PSU (specific brand/model) and for how long have you used it?

I've had a look at the log files. While it could be the GPU I see nothing untoward in the logs. Instead what catches my eye (in two of the three logs) is the high CPU usage on all cores, and perhaps the first thing to try is to add a little CPU Vcore. Say +0.025V. Sometimes CPUs need a bit more voltage after years of use and some degrading (or motherboard/PSU degrading).

It wouldn't hurt to set a custom fan profile with Afterburner so that the fans spin more than 50% @ 1500 RPM. The GPU temp is fine but perhaps the memory/VRMs need a tad more cooling.

Can you test the GPU in a different system?


EDIT - actually, reading your post a second time, I see the crashes are silent to desktop, not BSODs. That wouldn't point to CPU as far as I know.

 

[X82]

What PSU (specific brand/model) and for how long have you used it?

I don't have the exact model no to hand, but it is a 1000w Corsair. I've had it for approx 8 years maybe less.

I've had a look at the log files. While it could be the GPU I see nothing untoward in the logs. Instead what catches my eye (in two of the three logs) is the high CPU usage on all cores, and perhaps the first thing to try is to add a little CPU Vcore. Say +0.025V. Sometimes CPUs need a bit more voltage after years of use and some degrading (or motherboard/PSU degrading).

Very interesting. I'll have a look in the BIOS. I know nothing about Overclocking, so I assume there will be a setting which will say CPU Vcore and I just bump it up slightly. I'm using an AIO cooler, so maybe I have more room to manoeuvre.

It wouldn't hurt to set a custom fan profile with Afterburner so that the fans spin more than 50% @ 1500 RPM. The GPU temp is fine but perhaps the memory/VRMs need a tad more cooling.

Good idea, I'll look into that now.

Can you test the GPU in a different system?

To be clear, I have had this problem on previous cards too. I had an ATI before this (forget the model) and it did very similar.

Going to put what you have suggested in motion and report back. Although the crashes are never "on time" so it could be a while.

 

[X82]

Update:

Firstly, I have had bluescreens. Usually the same thing actually:

SYSTEM_SERVICE_EXCEPTION ntoskrnl.exe ntoskrnl.exe+1bc8a0

When in the BIOS, I couldn't immediately find that specific setting, but I did find OC Tuner, which claimed to do all this, voltages and memory for me.
I ran it, rebooted. The system bluescreened once I opened Firefox, but that could be coincidence as my system does bluescreen occasionally.

I wasn't smart enough to capture the CPUZ settings before the OC, but here are the new readings:

Here are the BIOS settings it changed once I ran the OC Tuner:

Let us see how this goes.

 

[randal]

Good luck, just don't superglue any capacitors back on.

 

[Danny75]

So what OC Tuner has done is overclock the CPU. Not ideal right now*. Reset the BIOS to stock settings, save and exit and then go to "CPU voltage" (you can see it right above DRAM voltage in the third pic) and enter +0.025v more than what it's showing.

e.g. Right now it's showing 1.315V so you'd enter 1.340v. But it'll be less with stock settings.

And thanks for the update about also getting actual bluescreens.


* Not to mention that Oc Tuner has increased the BCLK to 103 which could affect GPU and hard drive and RAM stability.

 

[X82]

Thanks. It really helps to speak to people who know what they are talking about.
I reset the BIOS and applied the voltage boost.

Here is what the BIOS now shows.

Although CPUZ seems to show lower and the voltage amount keeps changing, unsure if that is normal, I don't know.

Let us see how this performs.

 

[Danny75]

Although CPUZ seems to show lower and the voltage amount keeps changing, unsure if that is normal, I don't know.

It can be normal yes, with power saving C-States and Intel speed stepping. When you stress it, the CPU voltage should hit near what you've set in the BIOS (1.285v).

Let us see how this performs.

Yup. One thing to get out of the way at least.

 

[X82]

Well, so far so good, but I'm remaining cautious. If this works, you will have fixed a problem which has been plaguing me for literally years.

Out of curiosity, why did you think of CPU voltage? Your first post gave the impression that CPU shouldn't be maxing out when gaming. That correct?

 

[Danny75]

Well, so far so good, but I'm remaining cautious. If this works, you will have fixed a problem which has been plaguing me for literally years.

Out of curiosity, why did you think of CPU voltage? Your first post gave the impression that CPU shouldn't be maxing out when gaming. That correct?

Not quite - the CPU can max out, if it maxes all the way to 100% all cores you might get stutter but that's it. But combined with the age of the system, and the GPU not showing signs of stress in those logs you uploaded, might as well start troubleshooting with the CPU + tad more Vcore first, is how I figured.

Too early to say though. Give it a good pounding.

 

[X82]

Ghost Recon crashed again. Do I up the voltage by another 0.025? Or is there another route?
I failed to monitor it with Afterburner, I will next time.

 

[Danny75]

I'd start messing with GPU now.

Could lower Memory Clock in Afterburner to minus 125, and apply.

 

[X82]

Just tried.

Set it to -125 for mem clock, applied and played the game. Got about 20 minutes before it happened again.
Foolishly, while I did have monitoring up, it was paused. So that doesn't help too much.
I fear we are running out of options here and it is just a case of, it's old.

 

[Disco_P]

A gigabyte gtx 1060 3GB should still be in warranty. I'd be exploring that route if it is crashing so often in games. Everyone should be able to play games for 6 hours straight without a crash.

 

[X82]

As I feared. Although I will say that I've had two cards before this, an ATI and an NVIDIA. Both did this exact thing too. Back then it was with BF4. In fact, I can't remember a time in the last 3/4 years where games don't crash like this.
I think my motherboard is just shot.
The USB 3 ports on the back of the device have never worked, half of the 2.0 USB ports on the back don't work either or say malfunction.
Even using the front headers, only 2 of the 4 ports work properly. So yeah, my motherboard is poorly. But because of the age, I can't find anything which would work with my current CPU and memory and Z68 motherboards aren't cheap for some reason.

As a last ditch attempt, worth doing anything further with voltages?

 

[Danny75]

Just tried.

Set it to -125 for mem clock, applied and played the game. Got about 20 minutes before it happened again.
Foolishly, while I did have monitoring up, it was paused. So that doesn't help too much.
I fear we are running out of options here and it is just a case of, it's old.

Yeah, while you've only tried a few things so far, without being able to test a different PSU, or test the GPU in a different system, it is going to be tricky. Some high street retailers could be approached to inquire about power supply return policy, see if they'll allow you to purchase one to test and return it for refund if it doesn't help.

Can try lowering the GPU clock to something easy like 1700MHz @ 1.0v.

Spoiler: Click the button marked in green in this pic

Spoiler: Choose the square on the 1000mV column and lower it to 1700MHz frequency, then hit L on keyboard to lock it to that frequency for now, and hit apply in Afterburner

A gigabyte gtx 1060 3GB should still be in warranty. I'd be exploring that route if it is crashing so often in games. Everyone should be able to play games for 6 hours straight without a crash.

Could come to that. But he said he'd had the same problem on previous cards, so before sending it back I'd do everything possible to determine whether it is the card.

 

[Danny75]

I think my motherboard is just shot.
The USB 3 ports on the back of the device have never worked, half of the 2.0 USB ports on the back don't work either or say malfunction.
Even using the front headers, only 2 of the 4 ports work properly. So yeah, my motherboard is poorly.

That could be a shorting issue, and you'd have to test by breadboarding the motherboard outside of the case, on top of the motherboard box or a wooden desk, with no cables plugged in except the ATX 24 pin, CPU 4/8 pin, CPU cooler fan/pump, plus a single drive and the GPU. No Power/reset etc cables from the case. You jumpstart it by pressing gently on the two Power pins with the tip of a screwdriver.

If doing this, the CPU cooler mounting should also be inspected at the back of the motherboard, to see if any metal is touching soldering on the motherboard. Or if any exposed metal on cooler mounting plates is directly touching the motherboard tray/case.

And then use it to see if any difference.

 

[X82]

Yeah, while you've only tried a few things so far, without being able to test a different PSU, or test the GPU in a different system, it is going to be tricky. Some high street retailers could be approached to inquire about power supply return policy, see if they'll allow you to purchase one to test and return it for refund if it doesn't help.

Can try lowering the GPU clock to something easy like 1700MHz @ 1.0v.

Mine is set currently lower than that.

Suggest lowering it to 1500?

 

[Danny75]

Mine is set currently lower than that.

Suggest lowering it to 1500?

Good of you to check, but that's one of the idle voltages and frequencies, so setting what I said would still be fine.

 

[X82]

Annoyingly, I managed to play Ghost Recon for over 2 hours without issue.
However, I will still change those settings and see how it goes.

Just to add, I tried the graph but when I dragged the point down and hit apply, it sprang back up to its default.

I think I figured it out, I had to ctrl drag to move them all, otherwise it would have a single drop and I don't think it liked it.