[SOLVED] 8 pin gtx 1080 exceeding 225w? Safe?

[DirkDigglerr]

Question:
is it dangerous to get close to the 225w limit or a couple watts over with a single 8 pin connection? and how tolerable is the 225w “limit” for a single 8 pin connection?

My understanding is that the gpu can only get 150w from a 8 pin and 75w from the PCIe slot.

 

[Phaaze88]

1)No.

2)My current gpu has a max power limit of 300w. If I run Folding@Home with an overclock, Gpu-Z shows that I'm frequently hitting the power limit - and at times, exceeding the power limit by a few watts.
After like 10mins of this, my PC reboots. W/O the overclock, it runs for hours.

But Folding@Home isn't a typical gaming application anyways.

 

[mdd1963]

GPUs also draw up to ~75W of power thru the PCI-e slot...

Could be the mainboard's regulator circuitry initialing the hard /resetreboot for protection, or the PSU itself as well...

 

[Phaaze88]

Could be the mainboard's regulator circuitry initialing the hard /resetreboot for protection, or the PSU itself as well...

I think this is aimed at me...
This is what I've covered so far:
-Cpu power duration and limits set as high as they could go.
-Cpu Input Voltage set to 2.0v instead of 1.90v, because it actually yields better scores for some reason. Going below 1.90v soon triggers the Phantom Power Throttling, which is also triggered from point 1, due to the stock settings being too low.
-Even had to max out current limit. The motherboard only has 100, 140, and 200%. My current overclock crashes at 140%. 200% = All good. I don't know why there are so few options...
-Disabled voltage regulator communication.
-Voltage regulator set to high performance over balanced.
-Psu is short of 4 years old.
-I checked the +12v rail through HWINFO, and it never exceeded 12.1v nor went below 11.0v
-Done several 8hr runs of Asus Realbench, which stresses cpu, ram, and gpu... passes with flying colors, but the gpu is never really pushed past 260w though, except in games.

Folding appears to be the only thing pushing over my card's 300w max, regardless of medium or full power - low doesn't use the gpu at all.
If I leave the gpu at stock, I don't run into any issues and can run it for hours on end. The moment I add a 'small' bump, like +50 to the core clock - 10mins or so later, reboot!

It might be the software, IDK.
I was trying it out because it was suggested as a method to test gpu OC stability. I guess I'll have to move on to the usual methods.