PoW for Mainnet? cuckaroo29 vs cuckatoo31+

I see “Floonet and Mainnet will accept cuckaroo29 and cuckatoo31+”

What is the use case for cuckaroo29 vs cuckatoo31 on Mainnet? E.g. if I have a 1080ti, why would I want to run cuckatoo31+?

If I read other forum topics correctly, 29 is for GPU miners, 31+ is ASIC friendly and should be used by ASICs (and not GPUs)? Is this correct?


They are launching both? Interesting

I would assume the idea is that “cuckatoo” removes complexity form the problem, making it a better proof of work algorithm if it is actually NP; but its newer so the “proof by absence” that backs the complexity of any NP problem is weaker. If people are allowed to mine both and there needs to be a quick and dirty fork to disable cuckatoo, the network should still survive. Or something along those lines.

Note: my quote “Floonet and Mainnet will accept cuckaroo29 and cuckatoo31+” is from the top of this page: https://github.com/mimblewimble/docs/wiki/GPU-Mining-Stats

Cuckaroo29 is ASIC resistant and will be tweaked every 6 months to remain the domain of GPUs. It will cease to be in 2 years.

Cuckatoo31+ is the long term PoW that is meant to attract ASICs. Every 2^k years, the smallest remaining graph size wil be phased out over some months. So sometime in 2020, the PoW will have become Cuckatoo32+ which will be out of reach of consumer GPUs.

Is anyone aware of an ASIC available at launch or shortly thereafter?

As I understand no asics are known they’re anticipated to come later. Nobody knows for sure if they’ll be here within a year or never, or in a month.

So all the hashrate on GPS31 are from cards like the Nvidia rtx 2080 ti, and other cards with 11gb of memory?

How long have the details about this Cuckatoo31 algorithm been in the public domain?

Since about 4 months: https://forum.grin.mw/t/cuckoo-cycle-weakness-and-possible-fix-cuckatoo-cycle

Is that the right link? I get ‘access denied’.

The link works fine for me.

do you reckon anyone has built an asic in 4 months?

No, I doubt anyone made an ASIC yet, and we don’t see it in the hashrate, either.

The hashrate is roughly 50,000 GPU’s if they’re all 1080’s. Certainly under 100,000 GPU’s mining Grin currently, and if we price them at $200, that’s only $20m worth of hardware currently mining Grin. It will cost you minimum of $5m to even think about making an ASIC.

As it stands, Grin is actually just too small for an ASIC market to develop, even if ASIC’s got 100% of the block rewards.

That analysis can change quickly, of course! We just had Genesis and a bunch of GPU miners suddenly made public. Coin markets change fast in the first few months…

1 Like

First public announcement on a Grin ASIC at Introducing the GRN1 - A Cuckatoo31 ASIC from Obelisk