Never assume the cables from an old PSU will work with a new one. Much of the time they won't, and if the unit doesn't have the protection this unit did, you'll burn parts up.

Anything by OklahomaWolf: www.jonnyGURU.com

I don't have the gear to test EMI, but the scope will give me 20-30mV of background noise at the load tester without the unit being tested even being fired up whenever those big Enhance OEM units are online.

The big ones I use that don't throw out EMI: all EVGA/Super Flower 1600W units, Corsair AX1500i, Coolermaster MIJ 1200W, Seasonic X-1050. Also running an Andyson N700 and Seasonic Prime Titanium 750W in the same room as the load testing gear - those don't do EMI either.

1800 watts in that size housing at that price from a unit with a fake 80 Plus Gold cert? Yeah, I'd pass on that one.

Yeah, that's completely irresponsible engineering right there. Too few conductors with too few connections to the hardware. And with wiring like that, I seriously question their "1600W" unit is capable of anywhere near that mark over the long haul. If it is, it's a fire hazard because the cabling simply can't support it.

Stuff like this sickens me. This crap is going to burn someone's house down.

FYI, those Tokamak units throw a ton of EMI back into the power lines. I just pulled one out of my farm because I couldn't review power supplies with it running at the same time. Too much extra noise in the oscilloscope. They kill AM radio for a block, too. I have another 1200W Enhance Titanium doing the same thing which will also come out of the farm soon. Other than that, they are good units. I'd have just kept using it if I wasn't a power supply reviewer, but I hate shutting the rigs down just for load testing so I'm migrating to units that don't do that.

Confirmed - you da man!

Been waiting for a manual payout over 2 hours now, balance and unconfirmed has not increased since that time - issues still going on?

Decent unit for consumers with undemanding systems... not so much for 24/7 crypto mining. It may be shutting down due to overtemp protection, depending where you have it located. Go for something at least rated at 50 degrees at full power... staying with Corsair, the RMx, AX, and HX lines would be my suggestion.

Nanopool is paying out for me.

Watch out for those Y adapters, folks. They often won't take a lot of power either.


That's only one possibility. They're the same type of Mini-Fit Jr connectors used on video cards. It's either poor contact inside the connector, too much current running through too thin wires, or a combination of the two causing things like this.

The higher the current draw, the greater the risk of failure. Absolutely none of my mining cards are overclocked, or even running as high as stock speeds in many cases. I have had zero connector failures on any one of my mining PSUs over the last 3 years because I'm very careful about this stuff. I prefer to give up a little mining profit so I can be absolutely sure nothing starts a fire in that room.

Currently running one Seasonic X-1050, one Corsair AX1500i, one EVGA P2 1600W, one Super Flower Titanium 1600W, and one Thermaltake Titanium 1250W. The Corsair's been running the longest.

The Seasonic is only temporary at the moment, but it's not a newcomer to my farm. It ran 24/7 for a good year on a four card Tahiti machine with no issues because I wouldn't push the cards hard enough to cause issues. Right now it's running a Hawaii (two 18AWG cables), a Tahiti (one cable), and a Pitcairn (one cable). All cards downvolted as far as they'll go with reasonable speed.


That's really the best way to do it.

That would be me. I've got two of those Rosewills - the review unit will be pushing a 6 card rig as soon as I figure out what I'm doing for a motherboard.

Seasonic has been 18 AWG on all cables for a while now. I'm trying to gently prod them into upgrading those PCIe cables. They're amazing PSUs, but not what I run in the dedicated mining rigs because they're not really intended or built for it. That said, I have a Prime Titanium 750W running the office rig, and it does mine. It uses two single connector cables to push one downvolted 290X to get around the wire gauge issues. I've also got a Seasonic 1050W mining a three card machine temporarily until I un-lazy myself and get that Rosewill in there. Again, using as many connections between the PSU and GPUs as possible.

I actually can't remember the last time I saw peripheral cables bigger than 18 AWG, actually. Even my Super Flower 1600s use 18. That said, I usually run two risers per cable and have never had an issue provided I use them with cards that don't take a lot of slot power. For cards that do pull a lot from the slot, I get the risers with actual PCIe power connectors on them. You don't want to pull 90W out of one Molex for 24/7 use.

I review power supplies. Most run 18 AWG wire on the peripheral cables, regardless of OEM... never assume they'll handle much power.

This, right here. SATA and Molex cables and connectors were never intended to supply the amount of power needed at the slot for some of these RX cards. This is not a power supply issue, it is a power distribution issue.

That PSU is at the end of its useful life, full of 10 year old electrolytic capacitors. One card might be ok. I would plan on replacing it ASAP regardless of what it's connected to. Worry about more cards after you've done that.

My four Pitcairns are all doing around 75h/s. 2 270, 2 270X. One of those is a 4GB card, the rest are 2GB. I run them at 1100/1500.

My Tahitis and Pitcairns just woke right up. Thanks, Claymore!

That depends on the consumer PSU... most of the high end ones are industrial grade. I use all consumer units... then again, I review these things and have access to the really high end stuff that costs most folks a fortune. For most people, the server units will be more cost effective. I wouldn't even look in the consumer market if I had to actually buy PSUs.

I also try not to run any units that have output devices under the main PCB - these usually require heatsinking to the housing, and if you have no airflow to said housing they tend to heat up more than they otherwise would. I prefer units that have the output parts and heatsinks directly in the fan's airflow, and have a way to disable any fanless modes. Running these things 24/7 at high output is demanding - they must be kept cool and well ventilated. And at 240V, if possible.

Three reference 290s with Stilt bios (one Elpida, two Hynix), one Elpida 290 Windforce flashed to a 390, one Gigabyte 7970, one Gigabyte 7950, one Asus 270X TOP. 16.1.1 using 16.10.3 dlls, Win 10.

The machine is still not hashing at full speed - I'm suspecting a bottleneck at the CPU, the slowest Celeron I could get for it. Board is a Z87-G45 Gaming.

My 7 card machine has been stable for a good couple hours now on 0.6 at Flypool using two instances. Four Hawaii cards on one, two Tahiti and one Pitcairn on the other. Huzzah!

0.6 is a bit more stable than 0.5 for me, but less so than 0.4.2. Having a lot of machines crash out on writing the Tahiti bin file - some of them do this over and over and then suddenly start mining for a while, others never start mining.

The 7 card Tahiti/Pitcairn/Hawaii rig continues to be a problem on anything but 0.4.2 - 0.5 and 0.6 both run to GPU driver failure and the machine must be rebooted at that point. 0.4.2 ran on that machine all day and night yesterday with only a 1000 second reset timer. Not getting max hashrates out of any one card on this machine. Currently trying 0.6 on it with the Hawaii cards on one instance and the rest on another instance. We'll see how this goes...

Unlock the other core if you can. One of my all Tahiti rigs uses the same CPU, and it was a real slowpoke with this miner until I realized it was the CPU holding it back and I hadn't unlocked both cores yet. That machine is now mining much faster.

Not a problem on my Intel rigs, though the 7 card rig isn't mining to its full potential yet. I suspect that one could use more CPU as well - it's a mixed Tahiti/Hawaii/Pitcairn rig. Incidentally, that one is more stable with 0.4.2 than 0.5. 0.5 will mine for a minute or two and then the cards drop out one by one.