©ASUSTEK COMPUTER INC. ALL RIGHTS RESERVED.

nickolp1974 · ‎08-12-2014

Hi been doing some testing on some 780 classy's and wondered if you could help??
Been having a few hard shut downs when pushing both cards and a message which is to fast to read properly on the bios splash screen but it's on the lines of the psu's ocp was triggered due to instability,
My psu is the SF12000 plat.

My testing

Other day worked out max clocks for a voltage of 1.36v on cards independently then switched em both on and both were set to what i found there clocks to be and got a good score.(firestike)
Now did the same with 1.4v the good card managed 1520/2027!! And the other is hitting a wall at 1416/2027 which it will do at 1.36v so fired up both at those clocks and hard shut down with that message about ocp/instability, so dropped down the good card to 1500/2027 @1.39v and it passed but with a lower score than the run with both at 1.36v, tried again, lowered clocks and volts and score improved but still not beating the original one even though clocks are higher. During the bench there are no signs of instability.
I may grab a power meter and test to see what its drawing.

What do you think is causing this??

So running I have

CPU 1.47v
Gpu 1 1.4v
Gpu 2 1.3625v
2x 18w DDC
15x high powered fans
AQ6
4x ram
1x SSD
RIVBE

Sig is now out of date but most of the info is above.

i9 9900K, ROG Gene XI, G.Skill 4266 and one big bucket!

jeepcoma · ‎08-12-2014

Try putting your CPU back to stock settings and re-run the test. With an overclocked CPU and two healthy OCed GPUs, that's a lot of juice. I'm wondering if you're pulling more that 1200W for short periods of time that are triggering the protection (which is exactly how it's supposed to work).

When I first had my rig set up, I went straight for OCing and didn't get much other tuning done at the time, I would easily blow past 1000W according to my UPS, which is very granular and updates every couple seconds. I am running a 4960X and 2x Matrix Platinum 7970s on an AX1200i. Point is, for brief periods of time, I'm sure it was using more, even if the average consumption was lower. Tom's Hardware has shown this to be the case with some of their latest GPU reviews; transient consumption can be something like +25% more than what the rolling average consumption would indicate.

Anyway, it's an easy thing to test. Also I'm assuming you have all the power sockets connected on the MB (two for the CPU at the top, and a molex for the GPU at the bottom, in addition to whatever the individual GPUs require). Depending on your PSU's construction, you may need to balance the load appropriately across rails as well so no individual rail is being over stressed.

nickolp1974 · ‎08-12-2014

Running a superflower 1200w platinum an i think its on a single rail, i posted here just in case its a motherboard issue but most probably with the gpu's so high its more than likely pulling to much power/spiking. Yes every connection possible is connected, will more than likely grab another psu to power one of the cards and hopefully that will solve it.

i9 9900K, ROG Gene XI, G.Skill 4266 and one big bucket!

WillyK · ‎08-12-2014

Unless your PSU is faulty, I don't believe lack of power is your problem. On my R4E, I'm running a similar configuration with i7 4960X @ 4.7GHz 24/7, GTX Titan Black (OCed), 2 x DDC, 16 x 140mm fans, 4 x SSD, 12 x SAS HDD, Areca 1882ix-12 RAID card, and a bunch of other PCIe cards (all slots occupied). My AX1200i is reporting about 800W power consumption in average, and it's barely passing 1000W on peaks. I can't imagine your rig is spiking over 1200W.

I've experienced similar shutdowns before (long time ago). It turned out that pushing the OC too far was causing the problem and the OCP messages were just a red herring (the PSU was fine). You probably should make sure that your PSU sample is not failing on higher loads within specs anyway (just to eliminate that). I'd also have a look for possible mechanical problems (shortages, bad cables / seating, etc.). If everything is fine, then I'd guess you're probably pushing the OC a bit too far (the BIOS messages are not always pointing at the true root cause).

nickolp1974 · ‎08-24-2014

To follow up this thread, decided to purchase another psu and a power meter, made myself a dual psu starter cable and plugged it all in, went on to test firestrike with my known max clocks for the cards which are 1520/2027 & 1416/2027, this test passed without a glitch and pulled 1544W from the walll!!
Also tested with one psu as i had the meter and bassically anytime it went over 1420W it triggered OCP, so now when benching i have to use 2 psu's!! Highest i have seen so far is 3D11 with 1592W!! Amazed at what these cards can pull!!

Original psu was superflower 1200w plat. And have added a superflower 1000w gold to power one of the cards.
Cards in question on 2x 780'classified's.

i9 9900K, ROG Gene XI, G.Skill 4266 and one big bucket!

Possible RIVBE/psu OCP problem