Happy to announce I have signed on two additional distributors that have VERY limited stock for the initial pre-order.

klintay from eyeboot serving the Asia region: https://www.eyeboot.com/futurebit-moonlander2-litecoin-miner.html

Bitmart serving the South Africa and surrounding countries: https://www.bitmart.co.za/product/moonlander-2-5mhs-usb-scrypt-miner/

I also am working with someone in Australia to serve Austrilia/New Zealand so stay tuned for that!

Please DO NOT order from here if you do not live in these regions. These guys are set up to serve all the people that expressed interest in these regions but could not order from my western distributors.

SHHH I haven't made an official announcement yet :p

Shipping/general update here: https://bitcointalk.org/index.php?topic=2125643.msg22879343#msg22879343

Short version is that I finally have ASIC ship date to me on on the 30th, so production and shipments will start that week, so all of you will have your Batch 1 orders in hand by mid Nov.

Quick update for you guys!

So finally got ASIC ship date from packager. They are expected to be delivered to me on the 30th (there was about a 1 week delay due to the Chinese holiday all week last week). Production and shipments will start the week of the 30th, and I expect the full batch to be shipped out by mid november.

Since I decided to include a fan with these, I reworked the thermal design for it around the fan. There is no point in including a large passive heatsink with a fan, so am now including a pin type heatsink designed for fan induction airflow. This has done wonders for efficiency and max these can be pushed. Looking at about a 10c reduction in temperatures for all frequency ranges, and allows a max clock of nearly 1GHz (I was going to cap the clocks at about 800mhz with the old heatsink). This allows for a hashrate of nearly 6MH per miner (even though I would not recommend running it past 800mhz for stability/longevity)!

Below is a pic of the new thermal design...its only 20mm thick from base of PCB to top of fan, and the whole thing is 10mm shorter due to the fan not being stuck in the back.



Here is a pic of me trying to make the voltage regulator blow up...after nearly 3 days it was still going (that was running on a hub controlled by an original Rasberry Pi btw)




Windows binaries are also done, as well as Pi 1/Zero, and Pi 2/3. I will be posting instructions and driver links probably next week for you guys to download and making sure the software runs on your systems (so these will run on Windows/Mac/Linux/armv6/and armv7/8 systems).


Were just a few weeks away and can't wait for you guys to get your hands on these!

Yea you have to make sure M.2 slots are configured for PCIe...some motherboards have the option to switch between SATA/PCIe, and some M.2 slots don't have PCIe support at all.

This is exactly what I did, somehow Newegg agreed to a instore credit with a 100 restocking fee, and picked up two vega 56s for the same price   

Vega FE is a joke and the 64s/56s use the EXACT same pcb and die. Even it's double stacked HBM memory is worse ( upper stacks actually use separate slower timings). Unless you can use the 16GB it's a waste of money.

Yea its already been done..a GPU attached to a USB riser

There are no ASICs for Ethash since the bottleneck is the memory throughput and the 2GB+ dag size, and currently GPUs have the fastest memory of any compute device that can store that much data in memory. You could theoretically make a very efficient eth ASIC, which would essentially be a GPU with the cores optimized for ethash, but since the 2GB+ ram needed would already cost 10-20 watts, at most you could probably reduce power consumption from ~120 watts to maybe 25 watts for the same hash power and probably even more cost than a GPU. The power savings alone would not be worth it, since ~30MH for a single eth ASIC would never ROI the development costs.

Scrypt is actually a "similar" algorithm to Ethash, but how its structured allows you to make tradeoffs in memory size vs compute, and since compute is cheap scrypt ASICs have essentially reduced the memory requirements per scrypt core to 32kb. The ASIC used in Moonlander 2 has 64 scypt cores, which comes out to ~2MB of essentially L2 cache memory on the die. Just this 2MB alone actually makes up >50% of the ASIC die. You can't make this memory tradeoff with Ethash, so you can quickly see why a pure Ethash asic with its 2GB of memory needed is nearly impossible

nope...he has just developed mods for Vega on linux side that allow timing/volt mods that allow same kind of speedups you see with BIOS mods for polaris etc, and his big clients don't want anyone else using them

Im doing the same stuff, pretty much just modified drivers to do the stuff BIOS mods did (since you can no longer BIOS mod vega with its security chip). Just injects the voltages and timings during runtime instead (kinda like what windows guys are doing with registry edits).

If i wasn't so busy with my USB asic id probably release a how to guide...the changes are pretty straight forward, and you just need to recompile the linux kernel.

Maybe wolf's overlords want to gag me too?

Im working on scaling up production...im a one guy shop, and I don't have the resources to risk making a massive batch that I can't deliver. Which is why the first batch was relatively small, but as soon as the first batch is shipped out and everything goes smoothly ill be able to ramp up fast.

Have to keep in mind that the main bottleneck are the ASICs...I secured as many ASICs as I could afford, and all of that came out of my pocket with zero pre-order money. For second batch ill have more capital to work with and be able to order a larger number of chips.

If you didn't get one there will be another chance!

Yea you could do this with no problem...ill be providing bfgminer binaries for all major OSes and ARM builds for pi's. You can just point bfgminer to nicehash pool and mine with this on there.

Please keep non-order related questions to the main thread instead of here though. Thanks!

I hate to repeat this every couple pages, but the #1 point of these is not all about ROI. They are low cost miners anyone can afford to buy and a great way to learn about crypto and how mining works. I specifically build these in order to get new people into the ecosystem, that would have never bothered because A) they cannot afford 2-5k USD mining ASICs, or B) can't even run one because of crazy power and noise requirements. Alot of people also gift these as a way to get people that have never even heard of crypto into crypto.

I say this all the time, but new people and veteran members alike get stuck into the ROI and higher price mania and forget that the only way your going to even ROI on a 2k machine, or expect your 54 dollar litecoins to turn into 200 dollar litecoins is by getting even more people interested, buying, and using crypto. A new user is not going to buy 100 Litecoins, or buy a L3+.

Whats better? The same old 10,000 people using litecoin and about 1,000 huge mining farms mining it all in an endless cycle, or 100k new people all mining with these USB miners and all getting interested and buy/using Litecoin?

Hope you get it now

My original post was never comparing USB 2 ports....most people have USB 3 ports these days, and in its current config the stick can do almost 4 MH on USB 3 power. I was mostly stating the fact that you can stick this in your normal USB 3 port without extra hubs or anything else and it will run at nearly full speed.

EDIT: nvm i see where the confusion came from...I had mistaken listed USB 2 in my OP and meant USB 3.

25mm width, total length from front of USB port to back of fan 85mm, hight from bottom of main heatsink to top of small heatsink ~34mm

The final design has been locked in for production next month, and I'm happy to reveal it you all! PCB is a cool matte black design, with the same anodized black aluminum heatsink as the original moonlander. I decided to include both memory voltage adjustment in addition to core voltage, as my tests had an efficiency gain of 10-20% for ASICs that can handle lower memory voltage!

As a super thank you to all who pre-ordered I decided to include a 25mm fan into the cooling solution as well, so no need to come up with clumsy external fan solutions to keep these cool. A top side small ASIC heatsink will also be included to further increase thermal efficiency (it will be black not blue). With this setup I've achieved 1.6W/MH all the way up to 800mhz, which is 4.6 MH of hash power for just 7.5 watts! Pretty impressive for a chip thats designed to do 3.8 MH at the same watts.

Check out the pics below!



[/url]

I am still looking for a reputable member/reseller in the Asia region that would be interested in bulk distributing these in that region. At minimum you would need to have some sort of e-commerce store up, be able to accept crypto payments, and have some sort of verifiable seller reputation. I have had hundreds of PMs from members in these areas, but I'm unfortunately swamped right now and can't service small orders for Asian countries.

Shoot me a PM if your interested for future batches (I have a couple dozen units from this batch I can reallocate as a "test" run).

Guys please at least read the first post before asking basic questions. No one will receive anything until end of October at the earliest. And thats if ASICs actually get delivered on time and production goes smoothly. At that point my distributors will get the first moonlanders, and they will ship to you in the order of payments. If you weren't one of the first to place an order I would not expect anything until November.

With that being said, all major components are ordered, PCBs are being fabricated, and ASICs are still on schedule for mid-october delivery. So I don't foresee any delays at this point.

Keep in mind to mine these at a decent hashrate  (4MH), you only need 5 watts (1A per port). The original moonlanders you needed at LEAST 7-10 watts to mine anywhere near 1MH so these are way more USB port friendly, and alot more hubs *should* work with them.

Yea I would not, nor have the capability to build a huge ASIC...plus it goes against my philosophy of spreading out the hash. And yes if I do build a non USB ASIC it would be a "pod" style in the 5-20 asic range

Wait how "long" have you been around? Obviously not long enough to remember 2014-2015. Thats when 5k-10k scrypt machines ended up being resold for a fraction of that price a few months later. Those were fun times...picking up 250MH ASICs for 300 bucks back when diff was 40k

So I've been playing around with memory voltage on the production board(Moonlander 2 has a separate memory voltage controller), and was originally not going to include an adjustment for memory voltage (was going to be fixed), but I've been able to make some significant powers savings by tweaking memory. So looks like you guys will have full voltage control over both memory and core voltage This frees me up from having to choose a "high" memory voltage set point to ensure it working across all chips, and you guys can tweak even lower if the ASIC allows.

Ive been able to get the low end (3.4 MH/s) down to just .75 amps! Thats 3.75 watts and 1.1 watts/MH!

For comparison to the L3+ , if I could build a 150 chip version, that would be 510 MH @ ~ 560 watts!