GPU getting enough power? - weird PSU setup and driver crashes - Plz Help!

Toggle sidebar Toggle sidebar
rygar

rygar

Distinguished
Aug 7, 2010
21
0
18,520
So I recently bought an i7-7700k which I O/C currently to 5.0 - it seems to be running hot (about 87c under full load) and I may need to tone it down a bit so this might be part of my issue? Going to update my profile sig with all my PC specs after this post, but it's currently Liquid cooled.

My GPU is 1 x Asus Strix 980ti - on my old rig I used their GPU Tweak software to always keep it slightly O/C - using their standard O/C settings and never had a single problem.

On this new PC, just recently I've been getting driver crashes during gaming.

My new PSU is a Corsair RMI1000i 1000 watt PSU.


Ok so here is my main question - it's a modular PSU, so the 8pin power cables it comes with for graphics cards (of which my GPU takes 2 x 8pin) have weird 8pin cables. They actually have 2 x 8pin connectors on 1 cable. So on my current setup, I have only 1 cable going from the back of the PSU to power the GPU. It connects into the right most 8pin and then loops back from there to a second 8pin I'm using - all on one cable.

As I said I've been having GPU Driver crashes randomly during gaming. I am using a Hotfix driver posted on Nvidia forums because the current driver has a bug forcing PhysX to run on the CPU.

The game runs perfectly and then just randomly crashed. I'm on windows 10 64 bit and this is the first time I've been running games in Directx 12.

Here is a pic of the PC so you can view the power cable:

http://i.imgur.com/yrG1Auf.jpg

And a screen shot from Real Temp running when Battlefield 1 crashed the second time, which is not the only (DX12) game that has crashed - typically 5 - 15 minutes into the game:

http://i.imgur.com/a64nZ3W.jpg

For some reason my signature isnt showing with all my PC specs, so here they are:

My new gaming rig / baby - 2-25-17
CPU: i7-7700k currently O/C to 5.0 (Speed Step and Turbo Off)
Mobo: MSI z270 Gaming 5 Mobo
Cooling: Corsair Hydro Series H100i v2 Extreme Liquid Cooler (Using Artic Silver 5)
Ram: Patriot Viper 2x 8GB at 3733
GPU: 1 Asus Strix 980ti
GeForce Driver: Hotfix driver 378.77 http://nvidia.custhelp.com/app/answers/detail/a_id/4411
Monitor: Asus ROG 27" 1440p at 144hz
M.2 SSD: Samsung 960 Evo 1TB NVMe PCIe
Sata 3 SSD 1: Samsung 840 Pro 256
Sata 3 SSD 2: Kingston HyperX 256
Sata 3 HDD 3: WD Black 7200 1TB for storage
Headset/Mic: Logitech G933 7.1 Surround Sound Wireless
Mouse: Razer Mamba Wireless
Additional Accessory: Razer Orbweaver Gaming Pad
PSU: Corsair RMI Series RM1000i 1000 Watt 80+ Gold Fully-Modular
Case: Corsair Carbide Series Clear 600c Inverse ATX Full Tower
OS: Windows 10 Pro 64bit - Fully Updated



Any advice would be awesome! Thanks in advance!
 
Solution
Ironsounds
Hello... 1) Inspect the Ends/Pins of the power cable you are using... Pitting or arcing and any dis-coloring of the wire insulation is a sign there is a Bad connection or Construction. Are all the PINs in their proper position and not pushed out the back of the connector?
2) Try a different PS output... you have so many +12V there B )
3) You could get another cable and run parallel 8 pin output from the PS? (depends on the circuit design... refer to Mfg'r)

Sometimes water cooling can reduce the cooling air to the MB Power circuits placed on the MB next to the CPU... and the situation of heat there gets worst with High CPU OC's... give that MB heatsink a careful touch... Are you getting proper cooling there from those bottom case fans...
Sort by date Sort by votes
Ironsounds

Ironsounds

Champion
Jul 31, 2011
10,403
4
56,960
Hello... 1) Inspect the Ends/Pins of the power cable you are using... Pitting or arcing and any dis-coloring of the wire insulation is a sign there is a Bad connection or Construction. Are all the PINs in their proper position and not pushed out the back of the connector?
2) Try a different PS output... you have so many +12V there B )
3) You could get another cable and run parallel 8 pin output from the PS? (depends on the circuit design... refer to Mfg'r)

Sometimes water cooling can reduce the cooling air to the MB Power circuits placed on the MB next to the CPU... and the situation of heat there gets worst with High CPU OC's... give that MB heatsink a careful touch... Are you getting proper cooling there from those bottom case fans? (nice Clean build by the way)

4) or as a "Simple test" shut off the CPU OC for some further GPU Tests without it on your system. B /
 
Upvote 0 Downvote
Solution
rygar

rygar

Distinguished
Aug 7, 2010
21
0
18,520


Thanks for the reply!!

So I got ahold of a guy known as Infinite on the Nvidia community forums. Based on the info in his sig he is basically a pro overclocker, ranked 7th on some leaderboards.

He gave me a lot of advice so I tried lowering the clock speeds until it got under 75c maximum. I'm a bit disappointed, but I'm stable at 4.5 -not that thats bad... :)

His biggest piece of advice now was to change the fans on to the other side of the radiator so they are pushing through rather than pulling air in. I'm still pretty determined to get it to 5 lol. Ill mess with that next and see what kind of results I can get. I may return the H100i and go for a bigger, more powerful unit so I can get the temps down enough.

Now time to test out my games and see if I still get crashing. :)

He also confirmed that using a single cable for that particular PSU is going to be just fine. He said the same thing as you about running a second cable, but that I wouldn't get any benefit from it.

I also watched a video showing the entire process for delidding this CPU and sent that guy a crap load of questions to see where I can buy the tools he used. For him it dropped his temps more than 20c. Thats nutty.

Anyway, I sincerely appreciate the advice!! :)

 
Upvote 0 Downvote
rygar

rygar

Distinguished
Aug 7, 2010
21
0
18,520
So I was looking in the wrong place. XMP had auto set my RAM to the proper speeds but during Prime95 I got an error. Same with WMDT. I backed the speeds off slightly and re-ran the tests and the error went away.

Since making that switch everything has been pretty much rock solid. It seems every time I build a new rig I never hit the speeds my Ram is advertised to hit. Matching sticks, pre-verified... it doesn't matter.

Apparently, you really need to match sticks to the mobo manufacturer and model and even then it seems like luck. If only I built new PCs more often I might remember this... lesson learned.

CPU runs great but very hot... I backed it off quite a bit to 4.5 currently and it remains under 70c at load. 4.7 = 80c. Standard ring speed of 4.2.
 
Upvote 0 Downvote
Ironsounds

Ironsounds

Champion
Jul 31, 2011
10,403
4
56,960
Hello... Yes RAM... Typically your have to Down Clock your RAM and PCIe buss communications if effected by the CPU OC procedure. Ram is very fast at what it does already, and there is no Mass Heatsink or Fan on them, like a CPU... Typically I look for 1T VS Freq for my final RAM setting, and XMP are always 2T settings.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom