Your not changing HBM2 voltage. If your changing memory voltage via wattman your probably not doing anything(if anything your changing VDDCI). Vega's HBM2 voltage is locked at 1.35v for highest state. There is not much headroom to over volt it much more, since they are already over-volting it from the HBM2 spec of 1.2v.

Please contact one of the distributors in NA and Europe I posted above if you want  1-5 they both have enough.

Would probably have to add you to next batch, but send me a PM ill know for sure by Friday how many I have left from current batch.

haha give them a break? these things are designed to run 24/7/365. As long as your chip temps are running under 70c your fine. No point in running solo with only 2 machines...variance in block rewards will be super crazy (i.e. you could get "lucky" and mine 5 blocks in a single week, or 2 blocks in a month).

If you really want too you can point them at my pool in my sig. I was using it to mine segwit blocks a few months ago to help push segwit on Litecoin, but currently don't have my miners on there. You would be "solo" mining on it and all rewards would go to you.

Pro hashing is super unreliable. They advertise good rates, but in reality your mining on your backup pools 50% of the time.

PM sent

If your looking for just 1 or 2, please contact Scott at HolyBitcoin.com (he is getting a shipment for NA small orders).

If your in Europe, contact Bitshopper.de, he will be getting a shipment next week.

I developed a M.2 adapter specifically for GPU mining, and currently have a batch I'm producing for anyone interesting. The adapter connects directly to M.2 slot, and straight to GPU riser using USB socket.

There was only one working version of the M.2 to PCIe 4x  adapter that was used for GPU mining, and that got sold out, so I built this better version instead. This effectively eliminates the M.2 -> PCIe slot -> USB 3 connection, and has a way more stable PCIE connection to the GPU.

These have been tested with multiple boards, GPUs, and GPU risers and so far no one has reported any issues.

Pricing is $9-14 depending on how many you want (most of the batch is sold out, so if you want to order alot you'll prob have to wait for another batch).

Please keep in mind that if you already have 6-7 GPUs connected to the board, this adapter DOES NOT guarantee that more GPUs will post by connecting them to this adapter, since that is limited by the BIOS on how many GPUs can post. Also some board physically disable some M.2 slots if a PCIe lane is occupied (or you need to enable the PCIe M.2 slot in BIOS).

These are coming in next week, I only have a few left that have not been sold already if anyone still wants them.

Get yourself a clamp meter

2 TH has already been added to the network for the initial July batches...and they just started shipping them out...lol

Bye bye profitability from here on out.

Just tested aeon on the 1400...does about 700 h/s so not bad (8 threads). I cant enable large page size on the home version of windows 10 im running though.

EDIT: figured out paging...now doing about 800 h/s on 6 threads

I hope your not seriously thinking to cool down 10 units in an enclosed room by A/C alone. If your doing it by A/C at all in the first place your doing it wrong. You'll end up with as much A/C power to cool then as they put out.

Ill help you out:

1. Cut a hole in your wall
2. stick an attic fan to it
3. sip on your espresso

Driver is up on my github page btw for anyone that does not trust pre-compiled binaries...guess I should throw up the binary released on there as well.

https://github.com/jstefanop/bfgminer/tree/futurebit_driver

BTW stay tuned for some cool stuff

I got this thing up to 2.25 GH/s on Pascal...looks like out of the box with the shitty drivers this is the best bet currently. Makes sense since Pascal is purely compute and does not use vegas gimped memory controller.

My GF is currently using it as a hairdryer though...320 watts @1600mhz AND undervolted to 1090mv 

This is not impressive at all. A 580 clocked at 1200 can do 800-900, thats with half the cores and a fraction of the clock speed.

Why do people like you, that I assume has been around crypto for at least more than a week, make ridiculous posts like this? Where in the history of crypto has profitability EVER stayed the same for more than a week??

Even if you could buy one TODAY it would never ROI in 55 days, because difficulty will double over the next month or two as bitmain unleashes their new batches. The earliest you could get one without paying 6k for one is at the end of September, and ROI will be the same as an S9 by then.

Like with everyting else with crypto, the only people that had 55 day ROI were the people that placed an order when the first batches came out in April/May. Just like GPUs those days are over as well.

lol you wish...currently doing 400 sols @250 watts undervolted on zawaws kernel.

Maybe it can do 500 sols with claymore ASM kernel, but what I'm starting to see here is that they intentionally cut HBM bandwidth in half in the initial driver. Looks like they are intentionally doing this to keep miners away from the cards. Thats the only explanation that makes sense, since there is no reason why these CANT do 800-900 sols theoretically.

Yep, tests are from genoil's miner...had to compile it from scratch with some tweaks to make it work properly.

I find mining just more enjoyable than trading. Plus I actually believe in crypto and the ecosystem at large. Securing the network while earning coins for  "free" is way better than traders that are just in it for profit. Im still holding eth even with this downturn.