The proto rig is rock-solid after 24 hours of running at the new chip.conf.
These asics are curious critters - if you have a bad chip on a board, software will disable the chip in config. This causes adjacent chips to see a little more power and therefore get a slight overclock. For example most chips are producing around 1.5 - 1.6 GH/s actual nonces. But some chips are at 2.1 GH/s. In this way, even with some duds the full rig is producing 410 - 415GH/s actual accepted shares. Even Slush stats have caught up and show average hashrate over 400G at the pool.
I think this is unique - it seems most mining hardware is rated at a certain hashpower, but true output is typically less than promised.
Any chance you can post a CGMiner (or whichever miner you are using) screenshot of a 400gH/s product in action ? Genuinely curious, and excited with this product offering.
Also, as I understand it, the products becoming available will be allocated to your 100TH endeavor first ?
Does this mean that August orders will not ship until you have all 100TH online for yourself, or will the products be divided between customers and your mine if it looks like production delays may prevent your initial 100TH allocation arriving on schedule ?
Cheers.
I posted this video a few pages back of the screen output from our rig:
http://youtu.be/naW5uGHLbZE - added it to the OP.
The order for 100TH and August product is all one batch, and its being produced now. The first 4 full rigs shipped to Tytus today. I expect the rest to finish assembly very soon (though specific finish date has not been provided by factory). We have no reason to expect production delays at this point, or that we won't have enough product.
Logistically there two challenges for each effort:
For 100TH, assembling, configuring and running all those rigs will take time. We are trying to cut down on that by: preassembling racks, prewiring them with PDU, switch, patch cords, power cords & power supplies. We are pre-imaging SD cards for all the rPIs for both 100TH and customer orders. I have all the racks, PDU's, switches, PSUs, patch cords, power cords, SD cards, rPi's and a 16x SD duplicator all in hand.
For August orders the main bottleneck will be testing for bad cards. We will try to fire up as many as we can to test simultaneously. We can see fairly immediately if there is a card with bad chips, so hopefully we don't need to test for very long - maybe only a few hours, or until the unit shows it can come up to operating temperature with no bad behavior. Then out the door they should go.
I have two different ship teams and a build team for 100TH.
