Voltages Dropping: Low Performance

[Glowbit]

My PC is not performing how it should. I am getting low GPU usage in most games and some stutter. I noticed the voltage is dropping:
http://prntscr.com/61553y (OCCT Power Supply Test) With this test frames drop from 350 to 40-120, and voltage drops too.
Temps are not the problem, max cpu temp 55ºC and gpu 65º at full load.
When I run GPU stress test (Furmark), the gpu is at 99% usage and 12V rail is around 11.7,this is not a problem. But if I try to run a CPU stress test at the same time, obviously frames are lower, but voltage is lower too and GPU usage is around 30%. Here a screenshot: http://prntscr.com/61e4gw You can see how the 12V rail is at 11V.

Is my PSU broken or something? I think it's good enough. (Tacens Radix VI 850W)

AMD FX-8350 @ 4.5Ghz
MSI GeForce GTX 970 Twin Frozr V OC 4GB
MOBO: Asus M5A99FX PRO R2.0
RAM: 8GB Kingston HyperX Fury 1866Mhz
OCZ Agility 3 SSD 120GB + 2TB HDD + 500GB HDD x2
Case: Zalman Z11 Plus Red
Corsair H80i Water Cooler
PSU: Tacens Radix VI 850W
OS: WINDOWS 7 Ultimate 64 bits

 

[Dark Lord of Tech]

It's probably the very poor power supply.

http://pcsupport.about.com/od/insidethepc/a/power-supply-voltage-tolerance.htm\
Here's the ranges that are acceptable.

 

[Glowbit]

Could there be other component causing the problem apart from PSU?

 

[Dark Lord of Tech]

Could be but I doubt it the rest are top quality components.

 

[ko888]

TACENS is a company that puts out some the lowest end models from OEMs like Andyson, CWT, HEC, Huntkey, and R-Senda Electronics.

All of their PSUs have a warranty period of 2 Years or less. This is an indication of their very low quality.

Your power supply unit has very poor voltage regulation that becomes evident as the load increases.

RealHardTechX advises against buying power supplies from that brand.

 

[Glowbit]

Ok, I will test my PC with another PSU. Just out of curiosity: Which other component could cause voltage drops? Thank you.

 

[Glowbit]

I tried with another PSU, it is so estable but the performance is the same. This is playing Bioshock Infinite http://prntscr.com/68xo60 http://prntscr.com/68xoau

I tested my GPU in my friend's PC and the games were running fine. He has an i7.

 

[ko888]

I tried with another PSU, it is so estable but the performance is the same. This is playing Bioshock Infinite http://prntscr.com/68xo60 http://prntscr.com/68xoau

I tested my GPU in my friend's PC and the games were running fine. He has an i7.

Do you have graphs from your friend's PC when you tried out your graphics cards with it?

Your GPU usage doesn't exceed 65% in one of your graphs. That would indicate a CPU bottleneck.

Download and run TechPowerUp's freeware GPU-Z utility and post your test results showing the Sensors tab page. Pay attention to the PerfCap Reason when the GPU Load is at its maximum.

http://www.techpowerup.com/downloads/2452/techpowerup-gpu-z-v0-8-1/

Here's an example:

 

[madmatt30]

Are you entirely sure your over clock is stable??
The CPU voltage is low for an 8 core at 4.5ghz ,I'd be looking at 1.4-1.42v normally at that speed.
Apart from the substandard PSU for starters never trust hwinfo's voltage readings ,use amd overdrive or your boards own monitoring software ,the same goes for temps.

 

[Glowbit]

Yes my OC is stable, and the same thing happens if I leave it at stock settings.
I noticed something weird, in BF4 my fps are the same at 1080p and 720p. What does it mean? Is my CPU bottlenecking my GPU?

 

[Glowbit]

Running furmark: http://prntscr.com/69830r

 

[Glowbit]

With this PSU it says PerfCap Reason Pwr but with the other one says VRel... What does it mean?

 

[ko888]

With this PSU it says PerfCap Reason Pwr but with the other one says VRel... What does it mean?

Pwr: Limited by total power limit (i.e. the card has reached the predefined power limit that is programmed into its VBIOS)

VRel: Limited by reliability voltage (i.e. the card is being limited to NVIDIA’s default reliability voltage limit (Vrel). Manually increasing the voltage beyond Vrel puts the card at risk and may reduce the lifetime of the card.)

 

[madmatt30]

When I run GPU stress test (Furmark), the gpu is at 99% usage and 12V rail is around 11.7,this is not a problem. But if I try to run a CPU stress test at the same time, obviously frames are lower, but voltage is lower too and GPU usage is around 30%.

regarding the above - that is perfectly normal behaviour
If you want to run furmark stress test & the cpu stress test at the same time to fully test cpu/gpu together
then while running you need to drop into task manager
open processes tab
find furmark - right click,select affinity & untick all cores apart from core 0
find cpu stress test ,do the same but tick all cores apart from core 0
close both furmark & cpu stress test,then reopen them & run again

you have to do this as furmark requires at least 1 core fully free to utilise the gpu properly.