1000hx can't keep up?

[Elganja]

So i have been having these random BSOD's lately while gaming. I've been utterly stumped, but I keep getting a 116 BSOD

So time to do some stress testing:
-When I use either #1 or #2 GPU singly I get no crashes in FurMark and the lowest my 12V rail gets is 11.74V (according to everest).
-When I use them both in SLi, I get the BSOD 116 within 5minutes. 12V rails drop to 11.30V

For system specs, please check out the link in my sig (there are 2 tabs in that spreadsheet, one for CPU and the other describes my watercooling setup)

I suspect the PSU being my problem... This PSU worked like a champ for a long time. But I think with the added fans and fan controller (each fan and pump are on their own dial now instead of sharing dials) and pmp-400's vs mcp355's I may be throwing too much at it...

I guess it is time for the AX1200 and to sell the HX1000 (pm me if you are interested)... anyone else have any thoughts on the issue?

 

[peonyu]

11.30v is within the 10% tolerance range for the 12v line, so I really doubt that is your issue.

From my experience most issues with BSOD come down to either -
-heat [easy to figure that out, run prime95 or furmark].
-Drivers, easy enough to dl a different set.
And finally RAM. RAM is my bet for what your issue is, I had a similar issue with BSODs and couldn't figure it out for the longest. Turned out that it was a single dead RAM stick, and memtest didnt pick it up. I tested each stick by itself and thats how I figured it out. Worth a shot atleast.

 

[Elganja]

11.30v is within the 10% tolerance range for the 12v line, so I really doubt that is your issue.

From my experience most issues with BSOD come down to either -
-heat [easy to figure that out, run prime95 or furmark].
-Drivers, easy enough to dl a different set.
And finally RAM. RAM is my bet for what your issue is, I had a similar issue with BSODs and couldn't figure it out for the longest. Turned out that it was a single dead RAM stick, and memtest didnt pick it up. I tested each stick by itself and thats how I figured it out. Worth a shot atleast.

will try all 3.. if passes, then each one by itself.

 

[Elganja]

btw, not heat (max 54C) and tried several nvidia driver versions

 

[theAnimal]

11.30v is within the 10% tolerance range for the 12v line, so I really doubt that is your issue.

No, the range is from 11.4V to 12.6V.

Of course software readings are useless and should be read with a DMM instead.

 

[Elganja]

at this point, since each gpu can run successfully by itself, I'm going to go ahead and order a ax1200 just to rule that out as a problem.

i'll report back in a couple days after amazon get's me the new psu

 

[ViviTheMage]

Do you have a kill-a-watt?

I have an hx1000 with 2x6950's unlocked + OC'd hard .. 2xD5 pumps, SSD + 1 mechanical drive, 16gb ddr3, no issues.

 

[Elganja]

Do you have a kill-a-watt?

I have an hx1000 with 2x6950's unlocked + OC'd hard .. 2xD5 pumps, SSD + 1 mechanical drive, 16gb ddr3, no issues.

I'm starting to think the psu is just failing, not that it is over worked.

I'm doing some longer term testing on each card individually (2 hours FurMark each) to see if it fails. This should rule out the cards and pci-e slots as being an issue. Not sure if it rules out the RAM as a possible issue (regardless, I've been wanting to try out some Corsair Dominator GT 2000Mhz ram, so it's on the way along w/ an ax1200).

It seems like I can't catch a break these days... and when I do, it lasts a week before something gets f'd up. I'm so close to saying f this shit, sell it all and just get a laptop like I used to have to game on

 

[ViviTheMage]

1000HX has 5 year warranty, might be worth looking at?

Run memtest, that will let you know if memory is failing you.

Sometimes I look at my x220 and wonder why I have this big ass $2800 rig, haha...then I remember, it was fun as hell putting it together.

 

[Elganja]

1000HX has 5 year warranty, might be worth looking at?

Run memtest, that will let you know if memory is failing you.

Sometimes I look at my x220 and wonder why I have this big ass $2800 rig, haha...then I remember, it was fun as hell putting it together.

I did put a RMA request in with corsair for the hx1000 w/ the same stuff in the OP more or less. We'll see what they say, if they replace it, I will sell it since I already have a ax1200 on the way.

I have some memory on the way also that I had been eye'ing. I'll keep the other sticks around to test at a later date.

 

[ViviTheMage]

Keep us posted

 

[Elganja]

this isn't much of an update, still waiting on the AX1200 and some new RAM (I don't think this was necessary, but I had been meaning to do this since my GSKILL RAM couldn't ever run, since day 1, what it was spec'd for. So I'll RMA this and my HX1000).

But I did stress each GPU by itself for an 1hr 45min in FurMark, 2560x1600 resolution. I tested each one by changing what GPU the monitor was plugged into and disabling SLI. Whatever GPU the monitor was plugged into, was the GPU that was used. This simultaneously tested the PCI-e slots they are plugged into as well.

GPU 1

GPU 2

as you can see, they each work like a champ when run independently from each other. Heat is not an issue nor is anything else that I can see. I tried to get a log of them in SLI, but it crashes immediately and the logs capture nothing. Just as an aside, I tried 2 different SLI bridges as well.

 

[Meghan54]

Again, even Everest's readings are not to be trusted, like any software readings for voltages. The only way you can be sure is, as was mentioned before, to take readings with a DMM.

Otherwise, you're trusting a lie.

 

[Elganja]

Again, even Everest's readings are not to be trusted, like any software readings for voltages. The only way you can be sure is, as was mentioned before, to take readings with a DMM.

Otherwise, you're trusting a lie.

I'm more concerned between the delta of sli and nonsli. In anycase, the new psu will be here tomorrow

 

[OVerLoRDI]

Beautiful machine by the way

It would seem that it is power related considering what you are describing. Only other thing to test is ram, although the fact that the cards work fine independently leads me to believe that it isn't the ram. Logically your problem is a) power issues, either PSU failing, voltage issues or b) SLI being wonky. Given your software voltage readings, I'd also come to the conclusion you have, something is wrong with the PSU. Yes software isn't to be trusted, but if the difference is consistent, it points us in the direction of the power system.

Your problem is definitely not the lack of power, the 1000hx should be enough. My system in sig lacks Fermi cards, but still they do drain a respectable amount of juice, for a psu weaker than yours.

 

[pcunite]

It seems like I can't catch a break these days... and when I do, it lasts a week before something gets f'd up. I'm so close to saying f this shit, sell it all and just get a laptop like I used to have to game on

It is tempting to buy a prebuilt system to save on the headache of getting a bad part. I went through this the last time I built. That is why I recommend Puget Systems for those who don't wanna QC for asus, etc ...

 

[Elganja]

popped the new psu in (didn't put new ram in yet)... prelim results good. Running for 4 minutes so far in FurMark (will let it go for 1hr 45min)

 

[Elganja]

psu was the problem for sure. I ended up stopping the FurMark test after 30min and declared the system stable (for ****s and giggles I also did a few passes in LinX after I installed the new RAM... the system passed that test like a champ too).

thanks for everyone's help an suggestions