Possible PSU issue? Desktop crashing when gaming in CFX mode

Shmee

Memory & Storage, Graphics Cards Mod Elite Member
Super Moderator
Sep 13, 2008
7,908
2,881
146
So next to me, I have a desktop that I use when a friend comes over to play. Here are the specs:

HAF X
Seasonic 1250W Gold X series
HP ZR30W 2560x1600
2x Sapphire Fury
Gigabyte x79 UP4
Intel 4930k ES under CM Hyper 212 Evo @ stock
4x4GB DDR3 2133 Patriot Viper
2SSD, 2HDD
Windows 10 Pro

Ok, so now that that is out of the way, here is the issue: shortly after starting a game with CFX enabled, the computer will tend to hard shutoff. My friend and I suspect PSU possibly. The weird thing is that it mines ETH or equihash fine under both Xubuntu and windows, with both cards. I am even running OCCT PSU test overnight now, and there does not seem to be any issues. Temps are fine, about 60C on the CPU and 70 C on the cards.

Could there be something else at fault other than PSU such as motherboard, or maybe PCIE controller (on the CPU I believe)? I know games in CFX can require a lot of bandwidth on PCIE, where as mining does not.

Any suggestions on troubleshooting? Or should I just try an RMA to seasonic?
 

Ketchup

Elite Member
Sep 1, 2002
14,558
248
106
I am not a miner, but the way I understand it, most of those programs only put the GPU under load. Do you only mine with one card at a time? Similary, OCCT is a good torture test, but only for the CPU, which isn't putting out near the consumption of those Fury's.

By my calculations, you are pulling just under 900 watts if the full system is under full load with all cores running at the same time (read: probably not) so unless something is wrong with the unit you should be fine there.

The easiest test might be to pull the power supply out of the rig in your sig and test it in this rig.

Also, what are you using to test your load temps, and what kind of load is on the system when you are getting the numbers?
 
  • Like
Reactions: bigboxes

Shmee

Memory & Storage, Graphics Cards Mod Elite Member
Super Moderator
Sep 13, 2008
7,908
2,881
146
Hey, thanks for the reply. Keep in mind that OCCT has a PSU test option, which runs both 64 bit linpack and a GPU stress test similar to furmark, which can be full screen to utilize CFX or SLI. Load temps are measured using the coretemp, or in OCCT. when mining the miner reports the cards' temperatures.

But you have a good solution, try an 850 Watt PSU in the same system. If you think it would be enough.
 

Ketchup

Elite Member
Sep 1, 2002
14,558
248
106
It will be close but, depending on the game, I don't think it will hurt for a test.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,200
126
Unless the PSU is actually bad, then I don't think that's the problem. I concur that this may be PCI-E related. Can you crank down the transfer speeds to PCI-E 2.0 or 1.1?
 

Smoove910

Golden Member
Aug 2, 2006
1,235
6
81
Does it happen with only 1 card installed? What driver version do you have (try updating if possible).
 

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,200
126
It could be the game, too. I built my BIL a CFX rig, with a Q9550 and a pair of HD6850 cards, and Civ V would crash, apparently every time, if he didn't disable CFX first.
 

Shmee

Memory & Storage, Graphics Cards Mod Elite Member
Super Moderator
Sep 13, 2008
7,908
2,881
146
Just an update here. First, when CFX is disabled, it does not happen. I am using the latest drivers. I tried switching the 2nd card to the bottom PCIE x8 slot. Same thing. I tried mining and running p95 at the same time. No issue there however.

However, after reading some more online, I think it may still be a very strange PSU issue. Apparently there are reports of my PSU having some issues with failing or having strange issue, especially with the first model version, the SS-X1250XM. I checked and my PSU is this, not the 2nd version, the SS-1250XM2. I figure I will give a PSU RMA a shot, as it has a 7 year warranty after all.
 
  • Like
Reactions: Ketchup

Shmee

Memory & Storage, Graphics Cards Mod Elite Member
Super Moderator
Sep 13, 2008
7,908
2,881
146
Well, on 2nd thought, after talking to some more people, I am think it is the motherboard more likely. Something with the PCIE and CFX between cards. This would explain why it does not crash during p95 + GPU mining, which should draw more power. If so, I am SOL as the motherboard is out of warranty. I will still test the PSU.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,570
10,200
126
Crank down your PCI-E standard / speed, maybe? Some early X58 and X99 boards supported PCI-E 3.0, but it was "experimental". Sometimes setting PCI-E down to 2.0 fixes things.

On some entry-level modern boards, they have "mining mode", which as I understand it, sets PCI-E to 1.0/1.1, for greater stability with multiple dGPUs.