Originally Posted by cookieboyeli
OK hang on a second... Set GPU-Z (update first) to 0.1 second monitoring and check log to file. Then run a bench or whatever throttles it and upload the log file for us to look at. I just want to be 100% sure it's not hitting a "micro" perfcap and getting stuck or something.
ALSO! Step 2:
GPU error checking & overclock validation. You need to use OCCT
. Nothing else will substitute.
Click the main OCCT windows once the bench starts and click nothing else for 5 minutes. Watch your temps and the error count. If you encounter any errors before 80c or two minutes then your clocks really aren't stable. If you encounter a few dozen errors by 5 minutes over 80c that's fine, you're on the edge of stability and will not have errors during anything else unless it loads like OCCT and gets to those temperatures. If you get 1000's of errors at once note how long into the test and at what temperature! This is critical for discovering what the cause of the issue is!
Step 3: Upload your bios, I'd like to take a look.
Here's the gpuz log, I just did't change the second monitoring, forgot it:
GPU-ZSensorLog.txt 338k .txt file
Playing on TW3 @1531/7500 (so even lower for the memory) for about 15', no perf caps, however you'll see an increment of these three "Memory Used [MB] , GPU Load [%] and Memory Controller Load [%]" at the same time the memory is lowered from 1873 to 1502 (and a lot of stuttering and low framerate inside the game).
Here is the bios saved with the gpuz:
GM204MUMOD.zip 136k .zip file
As for the OCCT (last time used when I had a Radeon 6870 cf long time ago) maybe I'll try to give it a shot later but I'm sure it's something else, software related
Originally Posted by Laithan
Originally Posted by arrow0309
I'm getting nervous, really
I've rolled back to the 358.70 with three restarts during the two drivers (two ccleaner run and one with the ddu in safe mode on the second restart) and I've played on TW3 @1531/7800 for about 20-25'
Then I've got the same memory downclock to 3000 (msi ab) with the lousy framerate
I've noticed that the gpu loads massively even in idle, after this memory issue
Could be something more windows 10 relating than the nvidia driver?
Don't tell me that I've already screwed my vga
This is a software issue. Uninstall all GPU related apps, primarily O/C apps but any monitoring apps as well. There is nothing wrong with your GPU, if you put it in another PC it would be fine. This issue has happened to me any many others. You just need to remove the offending program. Be sure to fully remove any trace/config settings/preferences.
It has happened to me with MSI AB. It has happened to others with Geforce experience. It can be a PIA but you'll find it.
Usually occurs when O/C profiles from previous BIOS are applied to new BIOS.
You might also get some luck using MSCONFIG to disable everything starting up and add things back one at a time.
PS. The apps themselves are fine, usually once you remove the offending app you can even re-install the same one and it will be fine.
I'm also convinced it has to be some soft. related issue but I just can't figure out what.
Already uninstalled (with the revo uninstaller) the evga precision, msi afterburner and some apple software (left) from a recent evening (trying to connect and transfer some files to an iphone).
I've also deleted the shortcuts of various oc's I've created with the nvidia inspector 22.214.171.124.
One other thing I've noticed (even before I was gonna start playing TW3) trying to oc with the nvidia inspector:
so, after the first oc (1531 default gpu, voltage and the other two sliders maxed and the memory to 7500) I was trying a second one starting from the first one but applying to the new clocks (1551/8000) was not possible as the memory slider was always remaining at its 7500 value
I'm starting slowly to literally get pissed about this windows 10 and almost wanna return to the 8.1Edited by arrow0309 - 11/7/15 at 1:44am