Trying to OC my newly installed 880m, keeps cashing

[JGrimm]

I have a GT70 0NC, I recently upgraded from the 670m to an 880M I flashed with Klems Bios from this site.

I was using MSI Afterburner to OC my 670 with no problems.

 

But when I try to OC my 880M with both Afterburner and Nvidia Inspector, as soon as I start a game it crashes to the desktop. I'm only applying a 100mhz OC

 

When it crashes, I get a driver stopped responding kernel error and both Afterburner and Inspector lose the card. 

i.e. Inspector won't open after the crash and gives me a net framework error, and Afterburner says no GPU exists.

 

Thoughts? Do I need to up the GPU voltage in Inspector? What is the safe limit to set it too?

 

Also when I try to run Kombuster, it will only use the Intel Graphics

Edited July 1, 2016 by JGrimm

[Guest]

You should avoid using Furmark. It has been known to kill GPUs. Choose a different test within Kombustor that is not based on Furmark.

 

If there is a way to do it, disable Optimus. If not, try setting the global profile in NVIDIA Control Panel to force the use of NVIDIA graphics for everything instead of Intel. That won't disable Craptimus, but sometimes it helps force the use of the discrete graphics when it is a hit or miss proposition.

 

If you leave it stock it does not crash, is that correct? Only with an OC? If so, try using more voltage. The TDR error you are seeing is most commonly triggered by overclocking with too little voltage. Take it up a couple of clicks. May not take much extra. If it has the same problem running stock, either the driver you are using is having issue or the GPU is messed up. If it does it without an OC, try going back to the stock vBIOS and see if it still crashes. If it does, DDU and install a different driver before concluding the GPU is defective.

 

 

[JGrimm]

Thank Mr. Fox, my card works fine at stock clocks. Runs Witcher 3 on ultra at 30FPS at 72C

 

I tired to up the voltage +50mV and got the same crashes, should I keep uping it?

 

I dont see how when the stock vBIOS lets youOC to +135mhz, Im crashing at just +100

 

Also ill get rid of furmark, thanks for that advice.

[Guest]

Well, it's good that it is working OK stock. Does the same thing happen in all benchmarks and games with the same overclock?

 

Don't keep upping the voltage. You've already tried giving it more than what should be necessary. If the same behavior is observed with the stock vBIOS you may have something going on like a bad MOSFET or a weak VRM. I just sent one of my 980M cards to a friend to be repaired. It was working fine and suddenly developed a problem with crashing and black screen lockups, so something similar may have happened in your case.

 

When your crash occurs is the vRAM clock being left at stock or is it also overclocked or only the core?

[JGrimm]

It's a brand new 880M and I had the vRAM oc'd at +150, I was able to get a stable +80mhz core +100mhz vRAM with a +75mV, which according to Inspector brought the voltage to v1.0 exactly, but then the temps shot up to 92c and the card would throttle >.< I turned the clocks off again and the temps went back down to high 70s

 

should I try just OCing the core,? I've read that vRAM oc is pointless anyway

 

I've seen example of people oc'ing this card to 1200mhz core clock, so this is rather annoying that I cant even get a +80mhz OC with out some kinda problem.

[Guest]

7 hours ago, JGrimm said:

It's a brand new 880M and I had the vRAM oc'd at +150, I was able to get a stable +80mhz core +100mhz vRAM with a +75mV, which according to Inspector brought the voltage to v1.0 exactly, but then the temps shot up to 92c and the card would throttle >.< I turned the clocks off again and the temps went back down to high 70s

 

should I try just OCing the core,? I've read that vRAM oc is pointless anyway

 

I've seen example of people oc'ing this card to 1200mhz core clock, so this is rather annoying that I cant even get a +80mhz OC with out some kinda problem.

It's not your memory. It's the core takes a lot of voltage for a minor overclock. I just stuck an 880M (actually tested both of mine) in the P750ZM and I had driver TDR crashes with an OC of more than +100 on core unless I also added a +100 voltage offset (1.100V). I tested with the stock vBIOS and the 880M vBIOS posted here at T|I. I was able to overclock the memory up to 1450 (stopped trying there) with stock voltage. Voltage adjustment is for core only, not memory. Hardware mod is the only way to change memory voltage. So, try +100 core and +100 voltage and see if that works. Using +75 voltage was not enough for me with +100 on core... instant TDR within the first 2 seconds of 3DMark 11 with anything less than 1.100V.

 

http://www.3dmark.com/3dm11/11377980

 

http://www.3dmark.com/3dm11/11378132

 

[Klem]

Dont forget about power adapter. 240, 330 Watt.

GTX 880M is very hungry and hot card, especialy with overclock.

Edited July 2, 2016 by Klem

[JGrimm]

I did not even know they made a adapter with a wattage greater than 180, learn something new everyday.

 

Well even wit my stable plus 80mhz OC my temps shoot up to 92 C immediately, so I guess there will be no OCing in this laptop oh well

[Guest]

Wow, you're using a 180W adapter? That's cutting it really close, even for gaming on stock clocks. 

 

What kind of thermal paste did you use? Where the pads still good, or did you end up replacing thoses as well?

 

Are you able to manually force the fans to run full blast? If so, how do the temps look with fans maxed out? Is it stilling on a cooler or elevated to ensure good air intake.

[JGrimm]

Yeah that's what the laptop came with, ill be ordering a higher one now though

 

I'm using Antec Formula 7 Nano Diamond, and I replaced all the pads.

 

I'm attempting to install a fan control atm.

[JGrimm]

Okay now I'm having a new problem, I can't get the voltage in NVIDIA Inspector to go back down to stock levels, its stuck at 1.012v and sends my GPU temps to 93C within two minutes of game time?

 

Im trying to set the card back to stock clocks and it wont

 

 

[Klem]

1. Completely uninstall MSI Afterburner with all settings. Reboot laptop.

2. Completely uninstall EVGA Precision X with all settings. Reboot laptop.

3. Download the program DDU (NVIDIA drivers uninstaller), run it and completely uninstall NVIDIA drivers.

4. Download and install NVIDIA drivers version 359.06 with option "Clean install".

5. Reboot laptop.

6. Go in bios setup menu and on the last page "Save and Exit" set "Default Settings", then F10, then Enter.

7. Install NVIDIA Inspector and use it for overclock.

[JGrimm]

Okay I will do that, I always use DDU when I install drivers.

 

Any particular reason why I want to use driver version 359.06?

 

Also what can I do to keep my temps down.

[Guest]

Well, one reason is that it is a known good driver and many of the newer GeForce cancer drivers are unstable. 

 

Is there a newer or better GPU heat sink for a more powerful GPU available for the GT70? The reason I ask is the 670M was a lackluster 75W GPU option even when it was brand new. Unless MSI used one heat sink designed for cooling a 100W GPU even on the lesser GPUs you might make the temperature problems go away using a more robust 100W heat sink. Even with a 100W heat sink the 880M is going to run warmer than you would like it to. Using a 75W heat sink there's almost no chance of keeping a 100W GPU cool, and especially not if it is the 880M. 

[JGrimm]

All I want is to keep the temps in the 80s, soon as it hits the 90s it starts to throttle.

 

When I bought the 880m it cam with a heatsink, not for the GT70 but it came with one. they looked to be about the same thing, 3 copper pipes leading to a vent.

The stock GT70 heatsink is slightly larger than the one the 880 came with looked more like one for a CPU than a GPU

 

I looked it up though and MSI did make a special heatsink for the 880m, it has a thermal bridge, so ill have to buy it.