Have not followed this forum for quite some time, but is this a driver related issue? Also, I am not sure if this add on, Nvidia Experience, is the cause for the boost. I believe we could opt out of this previously, but not with the current drivers.
https://www.nvidia.com/en-us/geforce/geforce-experience/
I suspect it is driver related cos in my previous tests, older drivers like from the 382.xx versions (now the latest is 388.xx) seem to function ok
doubt that geforce experience is a factor because I actually dont install geforce experience
so far over there we tried
- total uninstall and reinstall of drivers
- bios update
- switching from MShybrid (which I keep on for optimus) back to discrete
in all cases the GPU still seems to go into overdrive (about a 25% overclock) when ingame settings are set to High or above, while in medium or below, the clock speeds behave more normal
the limitation of the testing is cos i only had 2 games installed (mass effect andromeda, deus ex mankind divided), we couldnt test with other games, although I did confirm that tools like furmark/unigine valley still have normal clockspeeds.
Another limitation is that all the display units at bendemeer are 1060 units so they cant do any A/B testing with their own S17
So right now i left it with them to test my hardware with their hard drive and their OS swapped in, to see if they can replicate the same issue.
My main hope is that its not hardware caused, since the GPU is non-replaceable from the S17 and I would have to RMA it, except I likely wont be able to cos the S17 model I use has been retired. I even mentioned about this before and how I dont buy parts warranty cos even you have 3 year warranty, no guarantee by the 3rd year the manufacturer still got replacement parts by then (i lucked out during the warranty period of my previous aftershock when I couldnt replace a dead 675M due to nvidia)
Although this issue was something i monitored for the past 2 months, i recently tested a 3rd title, final fantasy 14, which had the abnormal overclock
even on the old drivers (unlike the other games which was only like that on newer drivers) so now that my previous assumption of the GPU being 'fine' on older drivers is no longer valid, it raises new concern.
Frankly I hope this time round the problem can be troubleshoot easier. During my 675M incident, the stupid BSOD also only crop up when gaming, so aftershock spent 5 days trying unigine heaven on it with no luck to replicate the BSOD but I managed to BSOD on Skyrim. And then found out the 675M couldnt be replaced so I spent 1.5 years with a dead replaceable MXM GPU stuck in my laptop. If I really can be so suay and lost the silicon lottery on clevo's GPU 2 times in a row I think I will have to jump brand for the next machine haha
UPDATE
ok so they managed to replicate the same thing on an entirely separate machine and even on a clean windows OS so I guess nvidia's doing something weird to the S17 somehow that is jacking the clock speeds so high, which means I have to just keep rolling back to earlier and earlier drivers till I find one that doesnt make my games run at 90 celsius
i think when i looked up the 1070 i saw somewhere that the danger threshold is supposed to be like 97 celsius or something but 90 is still pretty uncomfortable to game with. And in my past 2 laptops (asus G73, Afteshock X15) the GPU was the first thing to die so i wanna be more conservative with temps.