Maybe somebody will found out this info helpfull. About 2 weeks ago I begin to experience strange problems on my mining rig: * GPU #1 got redicolous temp. reading 511°C and hashrate come down to 0 * mining rig freezes * after restart it work fine for few minutes * If only one GPU was working, than it was fine At this point I begin with exaust testing: PSU XFX and GPU #1 - everything working fine for few days; conclusion: GPU #1 is OK PSU XFX and GPU #2 - everything working fine for few days; conclusion: GPU #2 is OK PSU XFX and GPU #1 & GPU #2 - rig was working only for few minutes than rig freezes; conclusion: PSU XFX is faulty! PSU changed with Antec, further testing: PSU Antec and GPU #1 - everything working fine for few days; conclusion: GPU #1 is OK (again) PSU Antec and GPU #2 - everything working fine for few days; conclusion: GPU #2 is OK (again) PSU Antec and GPU #1 & GPU #2 - rig was not starting, PSU's current protection triped immediately; conclusion: PSU Antec is faulty also! Temporary solution: 2 partialy faulty PSUs connected in parallel are able to provide enough current to power my mining rig. Permanent solution: buying new PSU, probably Corsair RMi 1000W I opened Antec PSU and checked few capacitors on the end of 12V branch just before output wires. Very disappointed about quality. Leakage of electrolyte was obvious and this PSU should be high end. I changed faulty capacitors but problem still persist so fault is somwhere else also, conclusion: this PSU is only good for garbage. It's only few month over 2 yeras old - ts ts ts. Poor Antec, poor me.
That sounds bad, but it is good that you got this working. I'm curious, were you using multiple rail PSUs when the rig would freeze or not start?
It doesn't matter, because PSU's are not able to provide enough current (power). I try all available rails, same result. Now I use one 12V PCIe rail from PSU#1 (XFX) for GPU #1 and one PCIe rail from PSU#2 (Antec), for now this is working, but at any moment one or both PSUs can die for good and game will be over .
@Johnshpon3, I tried to PM you to ask if you'd like to buy a PSU from me, but it wouldn't allow me to send the message to you. Let me know if you're interested - I will take DCR or USD!
johnshpom3@gmail.com I'm in EU...from other countries we need to pay costum and VAT so it would not be competetive price.
Oh yes... I'm in USA. While my price would be competitive here, you're probably better off getting a new one after your country's fees and taxes.
Which VGAs do you have on? Which PSUs are they? How many +12V rails do they have? Did you have any spare Amperes left on +12V while those VGAs were on load? I prefer not to load a PSU over ~80% of its capacity for it to run 24/7. Anyway from your tests and since you can make both VGAs work at the same mobo at the same time it's almost certain that the PSUs have the problem. I suppose their warranty is over to RMA them? If you don't have any other problem you can keep them running in parallel but most possible their efficiency is getting lower so you'll may pay lost watts compared to a new PSU.
I have 2 Sapphire R9 Nano. PSUs are 750W so in theory should be plenty of spare Aperes. My measurment shows that one R9 is using about 0.7A @230V. DC was about 15.5A per GPU what I found little bit to much, it should be about 1A less but this could be measurement fault because not enough space for measurement clamps. This PSUs was previously used for BTC mining and it was runing 24/7 for maybe 2 years at 75%. I assume they done their work for good . I agree with you that efficency is not high with 2 PSUs in parallel but I not care much for few € every month. When first PSU will lost his sole I will buy new one, till then I'm happy PoW miner with 2 faulty unefficient PSUs .
On multi-rail PSUs, the maximum draw from any two rails are not always independent and at the listed values. There is generally a diagram with a PSU that shows these interactions, meaning not all setups are possible for all PSUs even if the rating of a PSU and its' rails declare otherwise.
Good point, my friend. You directed me into right direction, I think. I did some more researches and I think I found out where main problem lies. I will try to explain in next few words. According to some specefications an 8-pin PCIe rail should not draw more than 150W (12.5A @ 12V DC). My readings on my GPU was well over 15A and that should not be case for normal operation. GPU is high end and rated at 175W with only one PCIe 8-pin connector, so additional power can be supplied only by PCIexpress 16x connector on motherboard which is rated with max 75W. So, GPU mounted direct on mother-board and with only one PCIe additonal connector would work fine. And here is catch 22. For mining we use risers PCI 1x which can not provide extra power for GPU. In case of R9 Nano all power would be provided only by PCIe 8-pin connector and it exceed max. power for almost 25%. This is just not enough to trip current limit protection on affected PCIe rail, but it's enough to fried PSU over longer period. My conclousion is that I need to join 2 PCIe rails together in parallel and connect them to GPU. To support my theory I should mentione also that both PCIe connectors on both GPU's was partially melted and wires was visible damaged by heat. If I would wired two PCIe rails together before, I'm 99% sure that both of my PSUs would survived my DCR mining odyssey. I will surely use described wireing for my next new PSU and will publish some photos here . Conclusion and my recommendation (valid only for R9 Nano, or similar high-end GPU equiped only with one PCIe 8-pin connector and used with risers for crypto-minig): Rewire two of your PSU's output PCIe (12V DC) rails together in parallel and connect it to GPU's 8-pin connector. DISCLAIMER: Be sure that you know what are you doing!!! Failing to do correct wireing will result in destroying GPU, PSU, mother-board or all mentioned components! P.S.: My concern is that R9 Nano can die also after some time of mining 24/7 at 100%, because it's design is just not meant for that kind of (ab)use.
I'm using exactly this one. Apparently it is not enough (in my case ). Hmmm...idea...I will check how much current is going into GPU through this powered riser, but should not be much, because all power consumption in my case was (is) through 8-pin PCIe.