• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Need some knowledge please...

Dirk Diggler

Dirk Diggler

Dirk Diggler

Soldato
Joined
6 Jan 2013
Posts
22,170
Location
Rollergirl
I'll try my best to keep this as short as possible, but sometimes the hot air is unavoidable! :)

TLDR: Please prevent me from killing another CPU :(

I've put together a benching rig as a little side hobby sort of thing, the hardware is:

5930k
Asus Rampage V Extreme
16GB Corsair Vengeance DDR4 (2*8)
EVGA 1000w Gold PSU
Custom water loop
Radiators and fans galore
Windows 7 64 Bit

I prefer GPU benching to CPU benching, so I've amassed a small army of GPU's to get stuck into and I've kept the OC on the 5930k sensible, 4.4Ghz @ 1.25v and at a push 4.6Ghz @ 1.35v.

First up was benching a single GTX555. I've took 1st place on m ost of the 3D benches on HWBOT for this GPU, but Heaven DX9 is alluding me. So, I push the card to the edge (910 / 2000) and I push the CPU @ 4.6.

  • 1st run, driver crash. That's OK, I'm expecting that.
  • 2nd run, dialled the OC back a bit but it crashes again at the same spot in the bench.
  • 3rd run and I've dialled back a bit more, but same crash at same spot.
  • 4th run and I've put the card @ stock settings to see what's going on... same crash at same spot.
  • 5th run doesn't happen because... something's died. :(
I'm stuck in a boot loop, and after some Googling I fear the BIOS is corrupted. Here's some reading of where I got with that: https://forums.overclockers.co.uk/threads/rampage-v-extreme-sos-boot-loop.18776147/

After trying everything with the BIOS, including two replacement chips, I end up picking up a Gigabyte X99 board in the MM so I can test the 5930k. I fire up the new board... and boot loop. So, it's RMA time for the 5930k. Intel provide a stunning service and provide a 4 day turnaround, new chip in hand.

Superb. I install the new 5930k into the Rampage V and we're off and trucking again. The BIOS resets to default and I also get a message about IME updating, and then it's back in business.

I drop one of my 780 Classified's into the mix as I've gave up with the GTX555 and I'm looking for some 3 way action :eek: with the 780's. I've stuck the first one in so I can gauge it's limit on overclocking, so I started with +150 +300 and I have left the CPU & RAM at stock for now. Firestrike Extreme stress test and ....Driver crash. I'm not too fussed with this as it's normal right? Dialled back the OC a bit and... crash. Uh Oh, this sounds familiar. I put the card to stock and yep, driver crash.

The only way the 780 will pass the stress test is if I go complete stock and don't even move the voltage and temp/power targets in Afterburner.

Can the card really be that bad a sample, or am I in danger of cooking another CPU? Could this be co-incidence?

There's no way Intel will entertain another RMA. Can a GPU bench fry a CPU when the CPU is at stock? Of course, I could try with another card, PSU, MoBO etc.. but I thought I'd fish for some advice first as I really don't want to kill another 5930k.

Sorry about the long winded post.
 
Have you tried a different PSU maybe? Do you have a DMM to check the rails on the outputs of the PSU? Have you checked the voltage readings inside the UEFI/BIOS?

Also disconnect everything that draws power, that is not needed, e.g .fans extra hard drives and so on, just have the absolute bare minimum.

Probably take your block off, and run it with an air cooler too. :)
 
Thanks again. It totally makes sense to be the PSU, and I just feel a bit lucky not to have lost another CPU. I was running 780 SLI yesterday and my power meter was reading 750w draw when I was benching them.

With all the care taken with overclocking to ensure temperatures are within limits etc.. this has been really frustrating. How do you even allow for such a thing?

The PSU is within warranty with EVGA. How would I prove it's the power delivery without risking my hardware? I've got a dummy plug that I use for bleeding air from my water pump. Could I test all voltages within that plug?
 
I'm not concerned about the length of time to turn the RMA around, this is a benching set up and I'll probably not be doing too much of that during June/July due to heat issues.

I'm more thinking how would I prove to EVGA that it's a power delivery issue? I do have a SuperFlower 1200w PSU that I could slot in and test the system with, but not sure what info EVGA will be looking for in terms of the RMA.

I appreciate that you may not personally know the answer to that, so the question is aimed also at anyone who may have RMA'd a PSU for this reason.
 
I had an issue on an RVE and i cant remember exactly but it killed the CPU, i'd suggest using a different bios revision and make sure LN2 mode is disabled. IIRC after checking in the bios when i was having issues a certain voltage setting was sky high, the boards auto rules for the offending setting sent it way past a "safe" level for ambient use.

Thread i did a while ago which may interest you with regards the 780 classy's https://forums.overclockers.co.uk/threads/gpu-psu-power-usage-thread.18619153/
 
Update: the Dr Power II has arrived and I've taken the PSU out and did a quick test. It appears that all voltages are OK. The +12 is showing +12.2 but that'll be within margin of error?

I presume the test I did above is only telling me the voltages are OK, but isn't telling me how the PSU would cope under heavy load? I've got a Super Flower 1200w spare so I'm going to replace the EVGA 1000w with that and see if the system can hold a SLI benchmark.

Will update thread later.
 
You won't kill it in a benchmark no. The only thing that realistically will kill it is severe temps (105c+) or very high voltage. Possibly unstable power from the PSU yes.

Temps realistically won't do anything as it has thermal protection to stop anything from melting and cpu's are designed to run just under that point for their life.

Either run the latest or one of the earlier bios with the RVE. I have tried a 1500 version and it nearly killed my CPU there and then. Basically put 1.9v into it at startup (halted at overclocking failed).

I would say likely it's the card. Wouldn't be a bad idea to test it in a separate system?
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom