Let’s talk about ASIC mining

Here’s another important data point of another coin we can learn from: SiaCoin. They announced long ago that they had decided ASIC mining was better for security than GPU mining, for reasons that they explained at length in this blog post:

(I’m not entirely convinced by all of their reasons, but I certainly think there’s truth to at least some if not all of it.)

Since then, they launched their own custom mining hardware: November Obelisk Update. As we noted in June, Sia firmly… | by David Vorick | The Sia Blog

After that, there was apparently a situation where Bitmain announced a miner for Siacoin and then the Siacoin devs wrote “If Bitmain takes any action to harm the Sia project, we will soft-fork to invalidate their hardware.”

I haven’t had time to follow the whole thing closely. Anybody know of current developments, more details, or errors in what I wrote here?

I disagree, If a coin has one miner then it’s up to miner to put which transactions in. In my experience conflict of two ideology can simply destroy everything for one side. What if that miner is living in a jurisdiction which imposes for all transaction to do KYC and then simply kick out certain users.

Well, suppose we make Z-addresses extremely efficient and usable (starting with Sapling!) and then we (later, like one or two years later) deprecate t-addresses and then (later, like another one or two years later) the community consenses on no longer accepting payments from other people’s t-addresses, thus effectively eliminating t-addresses.

At that point, all transactions would be encrypted, so even if there were just one miner, it wouldn’t know which transactions to exclude or which users to kick out.

Hm, that’s interesting. It’s still circumstantial evidence, but it circumstantially suggests that Zcash ASICs are not (yet) widely deployed. :relaxed:

This is an interesting possibility — an open source ASIC miner. It seems like the sort of thing that would fall within the Zcash Foundation’s purview.

5 Likes

Here’s an idea of going back to the Myriad mining idea, what if the blocks were daisy-chained asic gpu asic gpu? (If that could be done). Would prevent either side from ever finding more than one block at a time, it would split the reward 50/50 and potentially moderate the difficulty for either side (assuming they exist)?
I saw some articles written by the same person about why obelisk should and shouldnt soft fork though i didnt get to read them yet

Disagree that such a change could be included in either of those two Network Upgrades, for reasons of business continuity for our partners (wallets, exchanges, OpenBazaar, Purse, Shapeshift, etc. etc. etc.) and project management (our current team is completely allocated for pulling off Overwinter and Sapling successfully and we can’t add to or change the plan at this stage).

It could definitely be done in a subsequent Network Upgrade, which could activate as few as three months after Sapling activates if we get started soon deciding on it.

Fantastic data collection and analysis. :relaxed::heart:ⓩ Thank you.

1 Like

What i mean is something else, Even thought everyone uses z-address, If there’s one miner having monopoly. Then it can simply say anyone who wants to get their transaction in the blockchain they need to share with them enough data to get the links together and even their personal information. So zcash will be some service like paypal. Don’t you agree?

Open Source ASIC hardware should ensure that there would be enough hardware to go around to those who want it. (I would like to see them be PCIEx16 ASIC cards…just sayin)

1 Like

With that daisy chain idea, I was thinking it could be done on the transaction level as well, it may not affect mining so much with maybe a more real time difficulty adjustment and you could get the same block attack resistance if the number of transactions in each block ( I know it varies with #s of z and T transactions per) was kept always an odd number, that way whoever started mining the block would also finish it and the next algorithm would begin the next, each block would be half Mined by 1 and half Mined by the other
Edit - it would be like 55/45 split each time but back and forth

I like this idea of letting ASIC’s battle with other ASIC’s and GPU’s compete against each other. There are coins that have multiple protocols you can mine against with different equipment and get equal weight in processing transactions. Like a few have mentioned - Myriadcoin and others (see What is Myriadcoin and Multi-Algorithm Mining? - Coin Brief )

Instead of arguing about asic/no-asic I vote for the daisy chain / multi-algo strategy. The open source ASIC idea is OK, but that turns into a battle of who has the cheapest production cost.

Just as a follow up after reading some more recent replies, I think what @zooko is saying makes a lot of sense.

ASIC miners have single-use boards, which means they take a bigger risk and depend on Zcash succeeding. GPUs are mutli-purpose consumer graphics cards. They can be repurposed, sold or can mine something else.

A publicly available ASIC would be a benefit, where a privately-held ASIC would disproportionately distribute coins to a small few.

I think it actually makes sense to either change the mining parameters every N months - to prevent private ASICs, or to make a publicly available ASIC (Bitmain, whoever) that is open source and free-to-build for any ASIC manufacturer (many ASIC developers all competing for consumers).

It makes complete sense that GPU miners here would be afraid of having one less coin to mine. ASICs do very little to hurt mining decentralization as long as anyone can buy them. Zcash hashpower is already centralized under flypool, but that’s a different thread.

If ASICs are inevitable as many here think then we should get ahead of the problem, accept that they are real and see if they can be made widely available instead of private. Disclosure: I own a bunch of GPUs that would need to be repurposed so this change would hurt me as well, but it could be best long-term.

2 Likes

In such a scheme, unless there is a PoW with strong ASIC resistance, there is no place for GPU miners.

You can buy a MyriadCoin ASIC miner here, for $2700 : http://www.baikalminer.com/product09.php

Here are some open-source designs for SHA256 mining.

Nice projects but the designs did not prove competitive. Bitmain is the market leader with its proprietary silicon.

What incentive is there for Bitmain, or anyone in their position, to open-source their design?

I don’t think this supposition holds.

While cryptography can help protect against the censorship and surveillance of individual transactions, it doesn’t help the system survive against total censorship i.e. blanket denial of service.

A blockchain where many independent decentralized miners can produce the next block, to keep the system functioning, is far healthier than a blockchain where only a single centralized miner can produce the next block.

A single centralized miner is vulnerable to third-party force and coercion, as well as short-term motivations (profit from short-selling) and resource challenges (power cuts). Surely it is anathema for a cryptocurrency to place so much trust in a single entity to not disrupt or destroy the system?

6 Likes

Yes, I see your point. Interesting change that would block a known issue with Bitcoin miners.

Other groups are actively discussing this option as well. However, a point that has come up here and elsewhere is that task specific hardware CAN be regulated, while general purpose hardware cannot. This is a serious issue that cannot be ignored.

1 Like

There’s no incentive, It’s competition, if you don’t care about the coin itself.

The threat I’m thinking about is gang of fiat money distributors on coins like bitcoin. What if china, US and EU decided to heavily regulate miners. Which force the miners in there to get more information about transactions through another protocol. “Power Corrupts”, If that happens probably wallets too will integrate that protocol in their program.
You might say what’s the issue with that. That’s fungibility. There’s incentive for governments to keep fiat money fungible. But crypt-currency, We should protect it.

But if we have more distributed mining this threat will get weaker. Since there’s always war in this world and conflict of interest can save us.

Good news is all of these hardware have lifetime. So There’s still time to help decentralizing mining.

1 Like

right on target @bitcartel !

I was laying in bed when i thought of it, i know its probably a band aid and really isnt viable until “unsolicited” or previously unknown asics begin outperforming gpus or maybe not even at all
It would keep gpu mining relevant, as asics hashing strength grows, gpu difficulty would decrease to keep up
and doesn’t lock us out of changing that when gpus do eventually out-date (it is inevitable) (or is it?)
Just brainstorming
Edit- if anything it could just buy some time to maybe spread out the zcash developed Asic chip to as many other manufacturers as possible