Hardware:
We intend to provide the crypto community with a low cost, low power and high performance alternative to GPU mining. We are looking to produce a board based on either the Xilinx Ultrascale+ or Intel Stratix 10 lines. Performance to power ratios for cryptocores we have developed have ranged from anywhere in the low 3x multiplier to as high as 100x GPU performance to power efficiency. We are cautiously optimistic that we will be able to deliver these devices at a cost / performance ratio greater than a GPU.
Software:
We will provide end users (customers) with software and bitstreams to allow them to operate their device in a 'plug and play' fashion similar to that of GPU mining. The bitstreams provided may or may not include a fee as it is up to the developer to set the fee. As time comes closer to launching orders / sales we will list some performance numbers that we have achieved for bitstreams that we intend to allow usage of with the device. I expect that the best performant cores and designs will come out of community designs from community RTL developers.
Allmine Crypto Shell:
Our FPGA devices will have our encryption keys burned into them and as a result we will be able to distribute secure encrypted bitstreams and software. The goal is to provide a development environment similar to that of the "aws-fpga" / Amazon F1 service. Community RTL developers would be able to access the software and compile time scripts / libraries necessary to securely compile their code into our shell environment. We will be able to distribute these community driven crypto cores and collect a developer set fee on behalf of the developer.
05/29/18 - We are going to delay pre-sales at this time. We have calls scheduled with both Intel and Xilinx for this week. News to follow.
Hi,
After reading all post in this topic, I understood that yours proposes derivate the idea of whitefire900 posted in the other topic.
I have some questions, I'm sorry if some seem a little obvious but I'm not a technician in that area:
1. How long will the equipment be guaranteed?
2. What is the deadline for delivery?
3. Where will these devices be specifically produced?
4. Is there any expectation of unit price with fees?
5. What is the minimum order to have a discount?
6. What will be the form of payment and the means that can be used?
7. Can they be shipped to Canada?
8. In the review of the initial post of this topic you wrote: "Our FPGA devices will have our encryption keys burned into them and as a result we will be able to distribute secure encrypted bitstreams and software."; In addition to enabling automatic operation with the software you have distributed via Allmine Crypto Shell, does this mean that devices can only use validated firmware within their de-development environment?
9 If you give up using this mining equipment to your encryption keys that are recorded on the devices have prevented other uses?
Thanks for your attention.
I had answers to these questions but a few things have changed. I'll answer the questions I feel comfortable answering at this point.
3 ) USA
4 ) Original price we were going to launch at is $3995. This will change over the next couple of weeks.
5 ) 1 unit minimum order
6 ) Crypto, Credit Card, Paypal (CC/Paypal only with manual review / authorization), and Bank wire
7 ) Yes -- first batch -- no problem
8 ) No, it means that the only encrypted bitstreams you can use that make use of the efuse key can be ours. Once our key is burned into the device it's permanent and irreversible. You can still use the devices for any unencrypted designs... Or designs that make use of other keys (BBRAM). We will be allowing any community developer who wishes to design secure code for use on these devices through our encrypted platform. They will be able to publish their bitstreams and the bitstreams will be available for anyone who's using our boards to download and operate via our software.
I should be really clear here, we will be colllecting a dev fee on behalf of these devs who release code. The dev fee will be set by the dev. We will take a small portion of their set fee for maintenance / operation of the software and shell platform. 9 ) The only other use prevented is loading secure / encrypted bitstreams. This limitation applies anytime you burn the efuse. If you wanted to use some other secure / encrypted bitstream, that provider would require you to ship your device to them or buy a device from them that is loaded with their key in the efuse. After they do that, you would again be limited to only using encrypted bitstreams provided by them.
.. I'll also add, our margin on these will be fixed. We're not planning a "how much you got?" sales method like baikal, bitmine, and others who regularly shift their pricing according to profitability. My goal right now is to get the cost / performance ratio to be better than a gpu. I'm not saying that we'll be able to sell them for as low of a price as a GPU -- but that when you divide cost by hashrate you'll get a number that is better than that of a gpu. For example, if I purchased a GPU to mine lyra2z it would cost about $400 per Mh/s. If I purchased a FPGA to mine lyra2z it would cost about $115 per Mh/s. We would like to get all of our algo numbers to be far more cost effective on fpga than gpu. On top of this, we're expecting an even greater power to hashrate efficiency. A gpu may use 100-200 watts per 1Mh/s of lyra2z. The fpga would use 6 watts per 1Mh/s of lyra2z.
These devices could be used to provide some ASIC resistance against 'secret' asics. Once we start releasing efficiency numbers you'll start to get an idea that any ASIC developed at greater than 28nm will be at best be on par efficiency wise as the FPGA. There are numerous groups that are making 130-65nm asics in secret and mining on them. I would not be surprised to learn that a group has already redesigned their high level cryptonight asic and are preparing to tape out for CN7. We need to close the efficiency gap between high level ASICs and general purpose hardware that everyone is using for mining.