Mining distribution: A new contender joins DigiByte, and Odocrypt ASICs
One of the things that MultiAlgo affords DigiByte over single-algo is a greater distribution of the block rewards, predominantly due to the broader away of hardware permitted for mining, and finding that valid nonce.
This is not just about a better / broader scope of distributing it to individuals, but also about decentralizing the mining and subsequently the block creation, minimizing the chance of any 51% attack. This applies to all PoW blockchains, not just DigiByte, though this is something DigiByte excels at.
Recently it’s come to my attention that HuobiPool have begun mining DigiByte on Odocrypt, an FPGA-focused algorithm made for DigiByte that morphs itself every 10 days in a bid to minimize ASIC creation chances.
HuobiPool already are finding approx 4% of all DigiByte blocks, which equates to around 20% of the Odocrypt mining hashrate. Good for them, that’s quite an accomplishment in just a few short weeks. This is exactly what we want to see with DigiByte, more people joining and participating, across all algorithms.
However this brings in to question the previous accusations that there were ASICs available on the Odocrypt algorithm. IF there were ASICs already available, mining in secret, it seems odd that a new player can participate and have such a high hashrate so quickly.
Now there may still well be ASICs, that’s not something that can be ruled out, which is why we have a pull request for DigiByte for inclusion in the next version to “tweak” Odocrypt. The idea behind this being:
- If there are no ASICs, then modifying Odocrypt with Odocrypt+ will not matter and it’s easy to update the bitstreams for Odocrypt to utilize Odocrypt+ on FPGA hardware
- Or, If there are ASICs, then mission accomplished, these changes should brick any existing ASICs for Odocrypt out there, returning it to being FPGA-mined
However this also begs the question: If HuobiPool can mine DigiByte and attain so much of the hashrate in just 2 weeks, are there really ASICs on the algorithm? Or is it just FPGA hardware being pushed further / harder / faster?
When Odocrypt was first launched on Testnet, mining was done with the Cyclone V and DE10-Nano FPGAs. They attained approx 35MH/s and 37.5MH/s. This means that at present, they would recoup approx USD$0.5 per-month at current rates. Given they cost USD$100 -> $200 for the hardware, you would have to spend 30+ years presuming the hashrate never changed, just to make back the cost of it.
What happened to those two miners? Improvements from other mining hardware, leaving them behind.
We had the likes of the Atom Miner which came along and left those figures for dust, then the BlackMiner Mini, then the BlackMiner F1+, all pushing harder and harder, with a greater hash-per-watt and a greater hash-for-dollar-upfront.
I’ve seen grumblings in the Mining channel on Telegram for a while with the presumption there may be an ASIC for Odocrypt, and it’s plausible, and this may simply be a second one? However it’s also plausible it’s the next generation of FPGA hardware and this is why the pull request has been made.
FPGAs by nature are reprogrammable, so tweaking the algorithm slightly in such a way as is proposed (Odocrypt+) should remove all doubt.
Again though, this is not something that will happen overnight, and feedback on the pull request is greatly appreciated.
What about the distribution though?
Well this change should also improve the distribution, as one miner is currently finding around 60% of the blocks for Odocrypt.
However coupled with this is the proposal for RandomX and ProgPoW, which would further incentivize home mining through CPU and GPU mining, while also accomodating larger ‘corporate’ mining. The intention with these is that all of the aforementioned stats will improve with them, which is why I am a proponent of them.
Keep an eye out on the GitHub for a future pull request for the algorithms implementations for further discussion around their implementation / inclusion. Such discussion points will likely include:
- Do we even want the algorithms?
- If so, do we want one, or both?
- What algorithms should they replace?
- If we’re activating more than one, in what order should they be activated?
If you want to participate in this discussion in the future, I would encourage you to go along to the DigiByte github:
As of 2020, development has now moved to the DigiByte-Core organization, under the digibyte repo You can find more over…
And click on the “Watch” button up the top-right hand corner when you’re signed in:
This will notify you of everything, or, you can even go in to Custom and choose the following:
This should then notify you of discussion (It sends you an email) around these new pull requests and when the future pulls occur to talk about algorithms and the likes.
So stay notified, have your say, and let’s make DigiByte 8 the best release we’ve seen yet!
I look forward to seeing more feedback in the comments :-)