Let’s talk about ASIC mining

ZenCash intends to change equihash params to 144,5

6 Likes

Nice, I thought they might fork. I guess Zcash will be the only one not to fork.

That is very true.
I doupt there is any real intention to fork off ASICs now or on the future. I would respect if they say it clearly rather than giving unrealistic plans.

I guess ethash is really inspirational their algorithm has an increasing memory requirement as you go.
And I see also ASICs are greatly failing in such an algorithm.
They have have made ASICs but its size so huge and very electricity consuming and just has the hash rate of roughly 4 ~ 6 GPUs mining rig.
I see no reason to think this way from now on.

1 Like

Thank you so much for your thoughtful post! I liked everything you wrote, and I agreed with most of it. It seems like you’ve really put a lot of effort into reading this whole thread and thinking about it and writing that post. Thank you.

I’m really interested in whether a dual-mining system — half GPU-friendly, half ASIC-friendly — would both:

a. Keep both kinds of miners engaged, and
b. Provide more robust defense against some sort of error or attack that disrupted the consensus algorithm

Also, if we had a dual-mining (GPU/ASIC) system, then it would be a natural step to progress to triple-mining GPU/ASIC/Proof-of-Stake!

10 Likes

I really like the economics and computer science of masternodes — anyone can put up sufficient stake to get the “master” flag on their full-node and then we can do more efficient and reliable distributed protocols with this smaller number of higher-quality full-nodes.

What I don’t get are the computer security of masternodes — there’s no way for the consensus algorithm to verify that the masternodes are performing their job correctly and reward them, so that means it would be possible for someone to cheat the masternode system and get more and more of the reward without actually providing more and more of the service.

Since the consensus algorithm can’t detect this, then it would be up to some organization to watch the masternodes and that organization would have to have the power to kick out masternodes who it thought were cheating… Overall, not the direction I want to go.

3 Likes

Yes, I agree with everything you are saying, except Bitmain is selling their packaged ASIC’s to other manufacturers, not their ASIC designs (If I understood you correctly).

1 Like

Who is mining? Bitmain right. They are batch testing their ASICs.
So I will tell you what happen when you fork.
You will simply drive out stealth mining and if they are the half of the harh rate then you will get ride of a big centralization threat.

What happened to monero when they forked? Nothing they actually got ride of a big threat.

1 Like

Bitmain is known to use TSMC, GlobalFoundaries, and Samsung as their foundry. Either way the foundry just produces the silicon for Bitmain, the ASIC is 100% owned by Bitman. So exactly what Myth are you trying to correct?

2 Likes

Why don’t you have a rotating POW where two different POW schemes can be running concurrently on the network where one is set to expire when it as been ASIC compromised (ASIC’s are a cancer to any crypto network) and a new one introduced with the other one already there if it is possible…can be researched. Your network will then be ASIC resistant and will not experience sudden drops of hashrate when transitioning. Can think of POW schme as an ongoing network security issue where ASIC’s are a compromise of the network POW scheme. i.e. I could create the best ASIC and not sell to anyone and will in effect control you network, if the network gets valuable enough this is what ASIC’s will do so companies with best ASIC fabrication processes will own your network if the POW scheme does not adapt.

2 Likes

Texas Instruments does not fab ASIC’s for Bitmain. As bitcartel posted for you, Bitmain states they clearly use TSMC.

Yes this is true, but that is not how this industry works. The are quite a few open foundries around the world. Some are better suited than other to do ASIC fab. So why would anyone want in house foundry capabilities??

So you have been involved with ASIC development? How many ASIC’s have you developed? I have only been involved with 9 so far, but recently started on my 10th.

So what are you saying? You believe TSMC owns or in some way contributes to the Bitmain ASIC’s? I thought you knew about ASIC development??? Do you not understand the difference between the ASIC design and the foundry? The foundry does nothing but fabricate the customers design, in this case Bitmain. Bitmain owns 100% of the ASIC, TSMC just converts the design into silicon. Yes the design and masks need to be tailored to TSMC processes, but other than that they own nothing.

This is how most silicon chips are fabricated. Nvidia uses TSMC and Samsung, AMD uses GlobalFoundaries and TSMC. Intel will almost certainly use TSMC as their foundry when they enter the discreet GPU later this year as their in house fabrication is tailored to CPU’s and they don’t have the expertise with GPU fab that TSMC has.

Edit:
Samsung on the other hand with certainly use their in house foundry for their ASIC’s and discreet GPU’s.

1 Like

Thanks, if that keeps ASIC’s out then I know which project and development process I will support in the long term over ZCash. I spend mining from day 1, I even specially got into mining because of ZCash (you can find my post before launch). I spend the last year building my small GPU farm in favor of ZCash ad helping out locally for people to get their own GPU’s running. At this point I am very pessimistic about the stance ZCash is taking, and will take a deeper dive into ZenCash and their approach.

4 Likes

I like the idea of a multi GPU/ASIC algo. It could work like a early bird (community) reward and you would get the best out of both worlds. But how to build that? Is there a proof of “I mined ZEC for the last block with a GPU”-proof? Could you create a ZEC corp mining pool, where you can proof that and or distribute >100% to early birds?

Anybody, is there a possibility for a pool to distinguish between ASIC hashpower and GPU?

well yes, we have every reason to be pessimistic right now, even after foundation taking action to investigate and get the result, it is back to the company to decide not foundation, and i’m personally only hear zokoo ‘decision’ so far, he seems didn’t take any ‘feelings’ or ‘loyalty’ value into his calculation and it is set even before the ASIC shown up, and no other team member give statement as loud as zokoo yet. But i will personally wait for more ‘sign’ before i take a step and not going back.

1 Like

The ASIC’s and GPU’s would use different algorithms to distinguish them

It appears ZcashCo want to make Zcash a corporate coin like “Ripple” they should clearly come out and say this as the community have no say and are slow walking any decision (which means they are not forking due to …(insert reason you can think of), even though their initial mission statement claimed ASIC resistance for their Equihash POW.

The community is responding by delisting Zcash as it is a money grabbing scheme, Changelly and ShapeShift no longer exchange it, not sure if it is permanent yet.

1 Like

the Myth that they are a chip producer/manfucator. There is a huge difference if you design a chip or if you produce and manufactor it, at least for me. Obviously not for you.

Thank you. Changelly kept telling me they were having a problem with Zcash and would not say what it is. And that’s why they pulled it. But I don’t believe that, over a week later and nothing. Been exchanging it there for year or so now I cannot. zooko keep the good work up. In time Zcash will be worthless!!

@chucky732 @bentusison and others if you want to speculate about future value of Zcash there is a thread for that: Price Speculation

Otherwise the speculation that “Zcash will be worthless” or “the value is going up/down/whatever if they don’t fork” are off-topic for this thread.

And there is no evidence that Shapeshift or Changely have paused thier conversion services for anything to do with ASICs or otherwise.

1 Like

Hi Zooko,

Thanks for posting that link. I understand that this must be a very hectic time. I am a bit confused as to why you replied to my speculation post, in which I was just thinking out loud, rather than my post where I asked direct questions, and responded to your request for ideas on what would be a good example of the potential of forking away from asics when you onlu have asics on the network. electronium will be the closest. I will ask my questions again and hope for a response.

The landscape has changed in the past few days and I know that the foundation will be checking this stuff out, but both your opinion and that of and zcashco’s is still important and relevant.

Regarding the link you posted, do you still stand by that report? It seems to have a different tone to the one the company is portraying now.
The report outlines and ignores a number of risks, which was my point. I was asking for the follow up risk analysis, but this is for a different thread, I don’t want to get side tracked with it here. (I am not knocking the report at all)

Would you please answer my previous questions:-

  • How is zcash going to cope with a need for protocol changes after gpu’s stop mining the coin and only asics are on the network?
  • What incentive is their for the gpu miners to come back to your coin to prop it up after a hardfork?
  • What incentive is there for someone to buy a next gen asic when you made their last asic invalid?
  • For the coin to be dynamic it needs miners that have the ability to make changes as the coins develop. How do you propose to keep this flexibility whilst having only asics mine the coin?
  • Are the ideals outlined in the zcash-evolution blog post still the ideals of zcash co?
  • Are the concerns raised in the conclusion of fixing-zcash-vulns still concerns? How does this align with asics?
  • How is zcash going to mitigate the issues caused by asics to bitcoin? specifically the issues I outlined in my previous posts.

I am not just trying to bring problems to the table. This I guess is more for the foundation. I will type this up properly and submit it for evaluation.
[does this belong here? should I chop the post up into two?]

I am sure this has been thought about/brought up, but maybe it is time to think about it again.

this is very over simplified for brevity

  • Get the algo audited by a company that has knowledge of ASIC’s, FPGA’s, cryptography and cryptocurrency. For example, bitfury.
  • look to embrace FPGA’s - work with a major crypto related bitstream producers. Go back and look at the old bitcointalk posts, bitstream developers are as plentiful as gpu miner developers.
  • Open to tender the contract to develop the inital bitstream for a Xilinix/Altera/whatever fpga. - There are methods and incentives for this. open source the results.
  • Open to tender the contract to develop the board layout. - again there are various incentives for a company like bitfury to do this. - open source it.

Work with “board manufactures” and develop a list of “approved manufacturers” A bit like the way ASUS design their own GPU boards, but there is also the reference design (which in this case would be the open source one)…

Heh, this just reminded me of an old bitcointalk forum thread where people were making their own fpga boards by using a skillet to attach the fpga to the board. when was the last time someone did that to make a gtx 1080ti.

But this is all speculation. It needs to be investigated. There are disadvantages to this and I have made quite a few assumptions.

full disclosure, I am in no way affiliated with bitfury. However, 5/6 years ago one of bitfurys founders helped me with the board design and development of my sha256 fpga. These guys really know what they are doing. They are by no means the only people with this ability/capacity, I am using them as an example because of their proven track record in the fpga/asic crypto space and their commitment to the ideals of crypto. Also at this time mistfpga has no intention of developing an equihash fpga.

Thanks for your time.

3 Likes