The State of Cryptocurrency Mining by David Vorick (Sia)

[gentlemand]

https://blog.sia.tech/the-state-of-cryptocurrency-mining-538004a37f9b

A fine overall read here from someone close to the front lines.

TLDR - unless you're Bitmain or someone similar you're probably eventual toast.

[d57heinz]

https://blog.sia.tech/the-state-of-cryptocurrency-mining-538004a37f9b

A fine overall read here from someone close to the front lines.

TLDR - unless you're Bitmain or someone similar you're probably eventual toast.

Excellent article.  It’s long but well thought out.  Thanks. Looks like if we start getting FPGA with a flexibility of about a 5-6 on your scale we could see them surviving so in the end centralization will go to those with more money than brains. Or ones that look to do harm later.  This system was designed on trust.  I’ll be damn if I trust bitmain to have over 51% control on any network.  They got their money now they could give a shit about what happens.  They keep selling the shovels and those that keep buying don’t even realize they are the ones dooming the entire system.  It’s a sad state of affairs.  Be a huge game of cat and mouse no different that what we have today with fiat.  There’s a reason satoshi needed it to be one pc one vote.  There also would have to be a way to prove you don’t have more than one vote.  Has anyone discussed a way to have a proof of decentralization.  Whereas the value of the system is actually based on the concept that it can’t be easily subverted.  I feel we are losing it with all coins currently in existence.  This will have to be addresses or we might as well all buy ripple. Or just keep with fiat and have blockchains that run on a few raspi. What’s the point of wasting all the energy if decentralization doesn’t matter to anyone here?


BR

[-ck]

Yes it's a good article and it's accurate. What's frustrating for those of us that lived through the conversion of bitcoin from CPU to ASIC is that the rules and issues have never changed, and we see the same mistakes being made over and over by both new manufacturers and newcomers into the game trying to capitalise on a market they think they understand. Bitmain have got the mining world by the balls and that is not going to change by anyone trying to take them on by the same rules they play by. I've been warning people since the first ASICs got announced but alas no one listened, and then I just gave up trying to warn them... No one seemed to want to believe me and for whatever reason very few every considered consulting me despite my obvious position in the mining world.

[gentlemand]

Bitmain have got the mining world by the balls and that is not going to change by anyone trying to take them on by the same rules they play by.

So given unlimited brain power and funds what would you do to try and change things? Or do you think it's too far gone regardless of what anyone does?

Mining is an area I understand very little of, but I'm surprised it wasn't foreseen that behemoths like Bitmain would rise and something wasn't done to attempt to address it before it happened.

[-ck]

Mining is an area I understand very little of, but I'm surprised it wasn't foreseen that behemoths like Bitmain would rise and something wasn't done to attempt to address it before it happened.

Satoshi predicted it. He simply said people should be nice and hold off doing the inevitable... The rest of us predicted it too but there isn't actually anything we could do to "address" it as you have said. Bitcoin is a diffuse collaborative and there are no weapons to defend us against transient centralisation of power and hashrate. Proof of work will inevitably ALWAYS have some element of this happening at regular intervals. Changing the proof of work algorithm will NOT, as the article said, make any difference at all; a new asic arms race will simply replace the old one.

Probably the biggest mistake everyone makes is buying stuff off bitmain only to find themselves competing against bitmain... duh.

[NotFuzzyWarm]

Surprisingly a very good, competent, and on-point article/blog post. Maybe I'm just jaded to have been expecting less but -- damn. Very accurate.
Original blog post

My end of manufacturing insight come from my company being a supplier of equipment made for one thing - mass production of chips. Specifically, laser micro-machined ceramic heat spreaders that power chips are built on. Anywho, each system can produce over 10 million components per-month, biggest customer has over 100 of our latest systems. I get what the author is talking about and he is spot on about costs and equipment availability to make critical parts that are needed to actually produce your product. Very often there is no 2nd source for those components.

Despite such huge production capacity our customers machines are booked for months in advance for, call it limited run 'open availability', along with solid years-long dedicated use contracts for their huge long term customers. In short they run 24x7x365 and the world supply of some critical 'widgets' does have a limit.  The fact that Bitmain can (nearly) buy out a suppliers yearly production is no surprise.

On the ASIC scale, surprised the author has not mentioned Intel, Micron, and others cpu and memory makers developments with CPU on-memory chips that has been going on for at least 2 years now. General term I've read is 'Intelligent Memory'. Originally aimed at very-large database storage and analysis applications to eliminate the round trips between CPU cores, outside memory, and back again.  Each memory stick has a set of cpu's on it to pre-sort data, sort of like a programmable gpu with huge local memory cache.  Sounds like it is perfectly adaptable to flexibility mining algos. Being in fact mainly a DRAM storage memory stick, call it maybe a 5-7 on the author's ASIC scale?

Folks, do take note of the Author's mention of Halong and their dealings.... worthy of another thread methinks?

Again, very good article and very rare I give praise!
Merited.

[philipma1957]

Lots of good info.

I agree with much of it but I need to quote two paragraphs in a bit of time on a pc rather then the phone I am on.

[Flying Hellfish]

Yes it's a good article and it's accurate. What's frustrating for those of us that lived through the conversion of bitcoin from CPU to ASIC is that the rules and issues have never changed, and we see the same mistakes being made over and over by both new manufacturers and newcomers into the game trying to capitalise on a market they think they understand. Bitmain have got the mining world by the balls and that is not going to change by anyone trying to take them on by the same rules they play by. I've been warning people since the first ASICs got announced but alas no one listened, and then I just gave up trying to warn them... No one seemed to want to believe me and for whatever reason very few every considered consulting me despite my obvious position in the mining world.

Ya I never understood it.  Since ASIC's day 1 people have LINED up to give interest free loans to their competitors months in advance for a completely unproven product.  I have mentioned quite a number of times the simple math of buying a unit from Bitmain (or anyone really) and it gives them the ability to put multiple units of their own on line.  I was even basically called a nut job conspiracy theorist for suggesting such a ridiculous idea.  I asked a million times why the fuck is everyone so horny to become an investomer?  Why take ALL the risk of an investor but never be able to reap the rewards of an investor instead your BEST CASE reward is being a fulfilled order? Why are you funding your competition with interest free capital??  But it really wasn't a popular opinion (probably still isn't LOL) and I took my far share of heat for it!

Your're so right it's not a new concept. Some folks over the years have commented on the "near zero sum game" design of the system.  We all know that those with the lowest capital and operational expenses will always make the most money.  Again as obvious is that a company manufacturing miners will have the lowest possible capital cost's to start a mine.  Add to that some of the lowest operational cost's (manpower and electricity) are in the manufacture's backyard and you can see that in time they will be the only ones left mining (if nothing changes and coins retain fiat value).

And even after all the outright scams, the borderline scams, the failed business attempts, delay after delay, 10's of millions of dollars "lost" and every other excuse the wild west here had to offer it's still amazing to me how many people come here with a get rich quick attitude or a complete lack of basic understanding of how POW mining works.  But such is the capitalist way I guess and a new batch a wide eyed awe struck newbs will roll in here every time coin price spikes and BTC is in the news.

[philipma1957]

Gpus were good for two and a half years.

They drew in a lot of people long warranties with decent service make them very attractive to anyone with lost cost power.

The next run up will be gpu driven again as there are methods that prevent asics.

Best method is work hash rate limit per ip.

Zcash  2x 1080ti = 1400 sols

So cap hash rate to 1400sols per ip this means multiple ips for bigger farm and only small asics could compete with a two card gaming rig.  Still attracts a very large amount of miners.

As much as asics did well for bitmain. The big year of 2017 had a lot of gpus.

A limit to hash power at diff 1 from an IP address works for gpus.

Cross the limit and increase the difficulty 10 fold.

Cross a second limit and increase the difficulty 50 fold.

That stops asics fairly easy.

So at the sia guy how about that?

Would have made your coin truly asic resistant.  But no you thought building an asic miner was smarter.

In the long run there can be less then one of five coins using asics.

And if developers had simple tiered Mining by hash per ip the financial desirability to asic mine is lost.

I have 140th for btc if my difficulty was 100x because I crossed a preset level I would have stayed with lessor gear.

I have 10gh for ltc if my difficulty was 100x because I crossed a preset level i would only have moonlander usb sticks.

I have 13000 sols for zcash if the cut off was 1400 sols I might have three locations with two 1080tis.

Point is stopping asics is easy.  The method above works.

It simply punishes large sized hash of more efficient gear like asics .

Just think if the difficulty for the 10000 sol asic from bitmain was 20x that of the diff for two 1080tis.

I get what you tried to do.  And you pointed your efforts in the wrong direction.

My method stops asic Mining for any coin that wants to stop asic mining.  No endless forks just punish high hash gear.

[-ck]

A limit to hash power at diff 1 from an IP address works for gpus.

IPV6 gives every person on earth quadrillions of their own IP addresses. You'd have to limit it to a subnet for IPV6...

[philipma1957]

A limit to hash power at diff 1 from an IP address works for gpus.

IPV6 gives every person on earth quadrillions of their own IP addresses. You'd have to limit it to a subnet for IPV6...

Sure but at 1400 sols or 700 sols. Per ip no big hash machine works very well .

Or as you say make it a subnet.

[Thetaj]

I'm going to state the obvious here but..............

Mining has always been a zero-sum game. The only question is whether Bitcoin mining will be lucrative enough that newer manufacturers will be willing to fight the old established entities for the a piece of the pie.

Honestly if you are not willing to play dirty against bitmain, you already lost.

[philipma1957]

I'm going to state the obvious here but..............

Mining has always been a zero-sum game. The only question is whether Bitcoin mining will be lucrative enough that newer manufacturers will be willing to fight the old established entities for the a piece of the pie.

Honestly if you are not willing to play dirty against bitmain, you already lost.

nope not so   simple

dedicated asics = by far =  zero sum or close to it.


gpus  with a vast array of other uses are much further away from zero sum concept.

I am sorry that you don't see the difference between a 2 x 1080 ti rig  that was purchased to game more then any other reason and 500 asic equihash miners at 10,000 sols a piece

one is a dedicated zero sum game the other is not although there are zero sum aspects to it.

That gamer will now buy a one card rig when the new video cards come out. Simple adjustment for that type of miner.

The truth is the 2017 boom was done on the backs of gpu expansion and asic leaching.

For me I am okay as I have a power deal that up to 20k-watts is pretty much unbeatable.  If mining was to last 15 years I would do well and the 20k-watts of cheap power will only grow for me.

But with bitmain doing the sell gear and sell gear and sell gear  and sell more gear I don't think the mining game lasts for 15 more years.
I had a lot of fun with it.

[moonstruck]

A limit to hash power at diff 1 from an IP address works for gpus.

IPV6 gives every person on earth quadrillions of their own IP addresses. You'd have to limit it to a subnet for IPV6...

Sure but at 1400 sols or 700 sols. Per ip no big hash machine works very well .

Or as you say make it a subnet.

A big mine would simply buy a block of thousands IPV4 or millions IPV6 addresses and create software to seperate the hashrate up to the maximum hashrate per IP. Something that would be easier than you'd imagine and would mainly cripple small mines that couldn't afford the needed hardware & internet connection, it would only push more power towards the big miners.
This would turn into a proof-of-IP sooner or later.

I like the thinking of how to decentralize mining but in this case the execution of the idea is not good.

[Thetaj]

nope not so   simple

dedicated asics = by far =  zero sum or close to it.


gpus  with a vast array of other uses are much further away from zero sum concept.

I am sorry that you don't see the difference between a 2 x 1080 ti rig  that was purchased to game more then any other reason and 500 asic equihash miners at 10,000 sols a piece

one is a dedicated zero sum game the other is not although there are zero sum aspects to it.

That gamer will now buy a one card rig when the new video cards come out. Simple adjustment for that type of miner.

The truth is the 2017 boom was done on the backs of gpu expansion and asic leaching.

For me I am okay as I have a power deal that up to 20k-watts is pretty much unbeatable.  If mining was to last 15 years I would do well and the 20k-watts of cheap power will only grow for me.

But with bitmain doing the sell gear and sell gear and sell gear  and sell more gear I don't think the mining game lasts for 15 more years.
I had a lot of fun with it.

While I understand what you are saying Phil, I think you are missing the point.

GPUs are NOT meant to be mining! It is something that they conveniently were able to do better than GPUs. The inevitability of ASICs or rather, machines whose sole purpose is to "mine" coins is as sure as sunrise and sunset. It is just business sense. If you think that GPUs are good investment because of their resell value, then you are pretty much limiting yourself to at most a very small mining endeavor. In that case you are losing the benefits of scale. Can GPU mining be profittable? sure it can, but it will never be as profitable as ASIC mining......ever.

If you have cheap electricity, then great! I have cheap electricity as well, hell I think most of us serious miners have cheap electricity. But what we do with that electricity counts way more. And any calculation will tell you straight away that given the same Electric price and timespan: ASICs are going to out-perform GPU rigs by at least 2-3X per dollar invested. Thats just facts Phil. Even accounting for resell value of GPUs (which if you haven't noticed are tanking as we speak because as I said, ASICs for everything) If you are long crypto in general, it is far better to buy ASICs now, especially with newer chips and slowing development.

[2112]

No one seemed to want to believe me and for whatever reason very few every considered consulting me despite my obvious position in the mining world.

Changing the proof of work algorithm will NOT, as the article said, make any difference at all; a new asic arms race will simply replace the old one.

Personally, I think you are just too pessimistic about the whole issue, possibly because you were ignored for so long despite your openness.

Picking winners in the ASIC game is relatively easy if one isn't afraid of occasionally changing the source code. When properly designed, it even doesn't need to be hard fork, just the version number of the PoW function needs to be explicitly recorded.

At the moment I don't have time to write a longer discussion, so for now I'll repost what I wrote in another thread. We'll see which of those new threads will get most intelligent discussion.

a non-trivial portion of the more complex parts of the instrucion set

Yeah, that is the key.

As a miner, this frightens me because when this era of "flexible ASICs" arrive, GPU miners will definitely be obsolete. Add this to the threat of Ethereum going POS, it seems like the odds are stacked against us regular home-based miners. This might be a signal that now is a good time to liquidate mining rig assets and just directly invest in coins.

The thing is that it is relatively easy to write hash function that are very ASIC-proof or FPGA-proof.

Bytom folks are a good example. Their goal was not to be general-ASIC-proof but to make sure that the ASIC that is fast at implementing their hash it their ASIC. So they wrote a hash function that uses lots of floating point calculations exactly in the way that their AI-oriented ASIC does. The hard part of understanding Bytom's "Tensority" algorithm is finding exact information about the actual ASIC chips that are efficient doing those calculations.

But the general idea is very simple: if you don't want your XYZ devices to become, play to their strengths in designing the hash function.

For XYZ==GPU start with GPUs strengths. I haven't studied the recent GPU universal shader architecture, but the main idea was to optimize particular floating point computation used in 3D graphics using homogeneous coordinates, like AX=Y, where A is 4*4 matrix and X is 4*1 vector <x,y,z,w> where w==1. So include lots of those in your hash function. In particular GPUs are especially fast when using FP16, a half-precision floating point.

For XYZ==CPU made by Intel/AMD using x86 architecture, again start with their strengths. They have unique FPU unit with unique 10-byte floating point format and unique 8-byte BCD decimal integer format. Additionally they have dedicated hardware to compute various transcendental functions. So use a lot of those doing chaotic irreducible calculations like https://en.wikipedia.org/wiki/Logistic_map or https://en.wikipedia.org/wiki/Lorenz_system . Of course one could write an emulation of those formats using quad-precision floating point (pairs of double-precision floats), but it will take many months.

During those months you have additional time to research more strengths of your GPUs or CPUs. Use them in a hard-fork to assure that the preferred vendor of your mining hardware continues to be Intel/AMD/Nvidia.

[fanatic26]

This whole speed limiting per IP is a ridiculously bad idea. It is so cringe worthy I cant even believe it was brought up and is being discussed as a remote possibility. Even the average miner with say 2 6 card rigs would have to own 3 IPs just to make like $250/mo mining. That would extend GPU ROI out so far it would be pointless to mine anyway.

People just need to understand that ASICs are part of the natural flow of crypto. Everything starts general purpose and as it ages it is refined and laser focused on doing only what needs to be done. People only protect GPUs as the "proper way to mine" because they are greedy and dont want their poor GPU investments to lose out even worse.

Pretty much everything about an ASIC miner is better than a GPU rig. Less failure points, less expensive hardware, higher hashrate, the list goes on. The only arguement that can be made about GPU versus ASIC is GPUs are quieter. I manage 10k ASICs and 10k video cards so I have more experience on both sides of the field than pretty much anyone else you are going to meet. GPUs are just a terrible investment when stacked up against ASICs. As much as nobody wants to hear it, these are just facts.

[philipma1957]

This whole speed limiting per IP is a ridiculously bad idea. It is so cringe worthy I cant even believe it was brought up and is being discussed as a remote possibility. Even the average miner with say 2 6 card rigs would have to own 3 IPs just to make like $250/mo mining. That would extend GPU ROI out so far it would be pointless to mine anyway.

People just need to understand that ASICs are part of the natural flow of crypto. Everything starts general purpose and as it ages it is refined and laser focused on doing only what needs to be done. People only protect GPUs as the "proper way to mine" because they are greedy and dont want their poor GPU investments to lose out even worse.

Pretty much everything about an ASIC miner is better than a GPU rig. Less failure points, less expensive hardware, higher hashrate, the list goes on. The only arguement that can be made about GPU versus ASIC is GPUs are quieter. I manage 10k ASICs and 10k video cards so I have more experience on both sides of the field than pretty much anyone else you are going to meet. GPUs are just a terrible investment when stacked up against ASICs. As much as nobody wants to hear it, these are just facts.

Wrong because you look at big scale. Which is what I am trying to get across a widespread small gear person creates interest . If 400,000,000. Gpus earn more then power and 200,000,000 are in your large 1k to 10k gpu farm and 100,000,000 are in a 50gpu to 1k gpu farm.  That means 100,000,000 are deployed in small setups. 100,000,000 gpus can earn as I type for the little miner. That is world wide distribution.
I have built and sold a ton of 2 card gaming machines.  They mine 6 to 12 hours a day to get a cheap gaming machine.

Destroy that widespread base and the zero sum aspect of asics come in.

Right now there are more then 1000 coins. Most are meh. But the volume of them create trading trading creates value for exchanges.  Asics simply flatten that out and coins will shrink.

Many coins sole value was or is they could be gpu mined at a profit.

And at thetaj I am not talking scale I am talking 50,000,000 worldwide gaming rigs owned by 1 person

Or 50,000,000 x 1


Not 5,000 x 10000 gpu  big farms

both are 50 million gpus.

and I 100% agree bigger miners want what works best for them.

In my case I will stay at a 20kwatt level for 2 years and maybe go to 40kwatt then 80 kwatt if buysolar and i can do the expansion of solar energy we want to do.

80kwatts is only 800th  of s-9s  which really isn't much at all

60 s-9's  in the scheme of mining is meh.

but  it means acres of paid for solar panels that have a 25 year life which is the real plan.

for buysolar and I  asics fpga gpus all work if they simply don't break.

My defense of gpus is not about my gear as I only have 16x of 100x gpus left.

I am all L3+ T1 and avalon 841

my defense of gpus is my belief that knocking off millions gpus as a mining possibility is bad for crypto coins .  Not for me.

[-ck]

Here's the thing though, the Bitcoin will likely be surpassed by another crypto-coin and another and another etc as people with the knowledge start their own block of coins buried in the digital ether....

And that's bullshit that dreamers always keep going on about just because they want another coin to make them rich by capitalising on first mover advantage again. Guess what, the first mover advantage is over. Bitcoin isn't going away.

[Thetaj]

my defense of gpus is my belief that knocking off millions gpus as a mining possibility is bad for crypto coins .  Not for me.

And on that I agree with you. But we are moving away from that now that Crypto (especially Bitcoin) is becoming mainstream and actually quite relevant in society as a whole.

I do believe that the way forward is to commoditize Bitcoin mining into appliances and other forms of electrical "subsidies" for consumers. Honestly, that is really the most plausible route to decentralized mining that I can  if that is what we're aiming at. But I digress, right now we are in the sweet spot of Bitcoin mining (regardless of what people say about profitability etc). The mining curve vs. Adoption rate (growth) is at its peak right now. After this, price growth will flatten out, and mining profitability will shrink after the next halvening enough so that by the time 7nm chips become mainstream, it probably won't be fun to have a dedicated mining facility anymore.