ASIC-friendly proof-of-work discussion

[jwinterm]

In previous discussions the focus was on merge-mining with Monero or implementing a CPU-focused proof-of-work algorithm as sort of a main-net testbed for Monero. However, there seems to be significant interest in switching to a hardware-implementable proof-of-work among some Wownero/Monero/Aeon folks, however the consensus seems to be that for Monero at least a change to something like SHA3 would be years out on the horizon at least, and in the meantime tweaks to Cryptonight will continue to be made or RandomX will be implemented perhaps in the fall of 2019 or spring of 2020.

This thread is for gauging people's interest in moving Wownero to SHA3 or another hardware-friendly PoW, as well as discussing the potential costs and timeline for an open-source design (and possibly production) for a hardware implementation of such an algorithm.

There are several coins that have used SHA3 in the past, including Maxcoin from 2014, so I think there are CPU, GPU, and even FPGA miners out there. Xilinx even seems to have something akin to a proprietary SHA3 ASIC already available. There are also Verilog implementation available here and a VHDL (hardware) implementation available here.

So, what would it take in terms of time and money to go from the available hardware designs in VHDL to actually having silicon parts made, put on PCBs, and interfaceable with mining software? Is there any opposition to pursuing this from people involved with Wownero? Could it be possible to crowdfund something like this ahead of a fork implementing SHA3 as PoW for Wownero, even on a prototype kind of scale?

[ghost]

Related issue for developing an open source CryptoNight ASIC https://github.com/monero-project/monero/issues/4533

[caokun320]

no!

[Satori-Nakamoto]

Well, someone's gotta do it. Might as well be AEON :laughing:

[caokun320]

Well, someone's gotta do it. Might as well be AEON 😆

asic ASIC-friendly not good way,BITMAN can first mine and dumping to peoples. 51% attack will be very easy. look DASH coin. if you think design new proof-of-work algorithm is hard,you can use RYO's algorithm, they work very good.

[caokun320]

buy the way,the pool wow.aiwakuang.cn is my pool, i like this coin ,if ASIC-friendly,this coin will Good for nothing

[ghost]

@caokun320 the idea is to have a transitional period where Wownero will remain CPU/GPU mineable for a number of months while a simple SHA3 ASIC-friendly PoW and free open-source firmware/hardware design is developed for the general public. This would enable ordinary hobbyist and entrepreneurs to build their own ASIC/FPGA, companies like BITMAN would not have a monopoly over the technology.

[caokun320]

@ caokun320的想法是建立一个过渡时期，Wownero将在几个月内保持CPU / GPU的可开发性，同时为公众开发简单的SHA3 ASIC友好PoW和免费开源固件/硬件设计。这将使普通的业余爱好者和企业家能够建立自己的ASIC / FPGA，像BITMAN这样的公司不会垄断这项技术。

but wow community too small,if asic friend,maybe one people control more than 51% hash, gpu miner leave ,WOW dead i think team more think about other,not this adic-friend

[timolson]

Wownero itself is far too small to support hardware development, so even if you switch to an ASIC-friendly PoW, that doesn't mean ASICs would appear. The economics aren't there.

That being said, I think Keccak is an excellent choice for an ASIC-friendly PoW. It is both efficient in hardware and simple to implement. When considering using specifically the SHA-3 variant of Keccak, it depends on whether you want some kind of wider market for the miners, or whether you want the coin's ASIC miners to be specific to the coin. If ASIC's are not specific to the coin, then there is less incentive for the miners to be good citizens. Also you have some risk of "general" SHA-3 hardware switching over to attack the coin, then going back to whatever they were doing before that. So maybe you want to use a particular configuration of Keccak-f[C].

The "Bitmain monopoly" argument doesn't hold weight, IMHO. Now that crypto markets have sufficient valuation, we have seen many competitors enter the miner market and currently it's WhatsMiner--not Bitmain--who has the best product and technology on the market. Specifying something easy to implement like Keccak helps ensure the lowest barrier-to-entry and most competition.

Finally a quick note about FPGA's. Generally they are far too expensive to be economically viable for mining, but since Keccak is especially fast in hardware, it might be the one exception. We tested the Athena VHDL code, and something like this $410 Altera 10 would get about 140 Gbit/s for Keccak-1600. Of course it would also need a logic board and system around it. Note that the Athena VHDL is from the SHA-3 Finalist round and some minor changes were made to Keccak before SHA-3 ratification.

[ghost]

Faster block verification is a huge win in the long run, which is probably more important for wownero, because of the long emission ...