Resurrecting the Champ: PoW to become Bitmain/Buterin resistant

[DooMAD]

and the referenced external post https://blog.sia.tech/the-state-of-cryptocurrency-mining-538004a37f9b

Fascinating read.  Seems there's even more nuance and intrigue than most people realise.  The takeaway is that if general purpose hardware can do it, custom hardware can do it faster.  So adding more complexity or a larger number of randomly drawn algorithms isn't going to help:

At the end of the day, you will always be able to create custom hardware that can outperform general purpose hardware. I can’t stress enough that everyone I’ve talked to in favor of ASIC resistance has consistently and substantially underestimated the flexibility that hardware engineers have to design around specific problems, even under a constrained budget. For any algorithm, there will always be a path that custom hardware engineers can take to beat out general purpose hardware. It’s a fundamental limitation of general purpose hardware.

So, in effect, we should take the opposite approach and lower the bar, not raise it.  If ASICs are inevitable, they should be as widely available as possible.  Make it easier for a greater number of manufacturers to create ASICs, not harder. 

That's the argument most Libertarians take when it comes to firearms, isn't it?  Everyone should have one so that no one can take advantage?  It's not an argument I agree with when it comes to guns, but I think it fits nicely here.

[2112]

So, in effect, we should take the opposite approach and lower the bar, not raise it.  If ASICs are inevitable, they should be as widely available as possible.

That is the general idea. Some already been arguing that the common CPUs and GPUs are ASIC, where the Specific Application that they are optimized for is a well known von Neumann architecture or 3D visualization pipeline. So the ball is on the software engineer's side and they need to find how to fully utilize the strength of the devices that everyone and their dog already have.

Make it easier for a greater number of manufacturers to create ASICs, not harder. 

I would reword it to the effect that we don't really need many manufacturers, we need many alternative uses and prospective users for the hardware used for mining.

The software engineers designing PoW algorithms are mostly too focused on how to spoil other's game instead on how to improve their own game.

Anyway, Bitmain responded to the above article:

https://blog.bitmain.com/en/bitmain-sia-state-cryptocurrency-mining/

which will definitely interest the readers.

[aliashraf]

Would it be worth pursuing the obverse strategy also, i.e. try to target the weaknesses of ASICs when designing the hash algorithm?

That approach in reality becomes: play to the weaknesses of the education of the cryptocoin developers. The sad reality is that nowadays most of the computer science graduates have no idea about logic design and architecture beyond the ubiquitous https://en.wikipedia.org/wiki/Von_Neumann_architecture from 1945. Moreover, after learning what the typical fixed-program ASIC does they are still mentally stuck in the next decade: https://en.wikipedia.org/wiki/Mealy_machine (1955) https://en.wikipedia.org/wiki/Moore_machine (1956).

You really should read the other referenced threads, e.g. "ASICs mining game"

https://bitcointalk.org/index.php?topic=3788591.0

and the referenced external post https://blog.sia.tech/the-state-of-cryptocurrency-mining-538004a37f9b :

Sia dev, David Vorick, seems to me just like an ordinary hypocrite, a person who attempted a crack against his own aglo, get bribed to do so, probably. The referenced article is a part of his mission and of Bitmain's canonical propaganda these months, discouraging ASIC resistance attempts in PoW domain.

Monero did it and there is nothing Bitmain can do about it with all the resources and talents accumulated in their corporate. And yet cryptonight 7 was just a minor tweak to the algo made in a rush.

There is no flexible ASIC, it is cheap journalism, there is no bar sliding from 0% flexible ASICs to 100% flexible ones(!), it is just a pile of hype and propaganda invented by Bitmain to sustain its dominance in btc mining and expand it to other coins.

Ethash is a solid ASIC resistant algorithm, as Vorick has admitted in his embarrassing article, and I don't believe Bitmain has been able to do much about it and E3 is an architectural attack that enjoys the chaos in ram and gpu market nowadays. It is completely possible to have a practically ASIC proof PoW algorithm. Some ideas has discussed here and there is a lot to discuss more.

Saying that a cpu is an ASIC optimized for Von-Neumann range pf problems, or a gpu for 3D algebra, does not change anything, even for a bit!

Intel,  AMD, Nvidia, ...  chips are optimized chips for a very wide range of calculations, wide enough to make them usable for almost any application. An attacker can do whatever s/he wants with a specific application but when it comes to a complex enough problem, an ASIC resistant PoW algorithm, nobody can go further than a state of the art gpu unless s/he manages to become a competitor (and a winner) in gpu market as well.

[DooMAD]

but when it comes to a complex enough problem, an ASIC resistant PoW algorithm, nobody can go further than a state of the art gpu unless s/he manages to become a competitor (and a winner) in gpu market as well.

Nope.  The whole idea is precisely that it won't need to perform in the same way a GPU does, so they don't have to worry about the "very wide range of calculations" a GPU would have to deal with.  That means they can engineer around the problem and focus solely on the calculations they need to worry about to do the "work".  That's why ASICs are faster to mine with than state of the art GPUs.  They are dedicated to one singular purpose.

That's what Vorrick means when he says:

For any algorithm, there will always be a path that custom hardware engineers can take to beat out general purpose hardware. It’s a fundamental limitation of general purpose hardware.

It's like the difference between a high performance, yet road-legal, sports car versus a Formula One / NASCAR / Le Mans car.  The latter ones are generally going to be faster because they're purpose built for racing and never have to worry about traffic calming speed bumps like a normal sports car has to cope with.  However, you can bet that if Formula One / NASCAR / Le Mans changed the rules to say the race courses could have speed bumps, the engineers would immediately plow untold sums of money into developing a car that would still be fast over speed bumps.  Hence, why making PoW more complicated isn't going to slow them down for long.

[cellard]

As widely discused before, it is only a matter of time when ASICS (or just specialized hardware in general) will always arise no matter where you go with the PoW configuration. Just pick any algorithm, and I don't see how these fast as hell machines will not start appearing eventually again, so it's more can down the road kicking.

I still see the "random PoW change lottery" approach as the only way to sort of spread around the advantage, so in some cases, some people will have more hashrate, in other cases, others will.

Im not sure how viable is this and if this would help decentralization in practice. What if someone has a massive advantage in a certain hashing algorithm and we are stuck with someone clearly dominating for a while? they could use that time to perform an attack... one would need to make some sort of simulation model before venturing in such a thing.

[aliashraf]

but when it comes to a complex enough problem, an ASIC resistant PoW algorithm, nobody can go further than a state of the art gpu unless s/he manages to become a competitor (and a winner) in gpu market as well.

Nope.  The whole idea is precisely that it won't need to perform in the same way a GPU does, so they don't have to worry about the "very wide range of calculations" a GPU would have to deal with.  That means they can engineer around the problem and focus solely on the calculations they need to worry about to do the "work".  That's why ASICs are faster to mine with than state of the art GPUs.  They are dedicated to one singular purpose.

That's what Vorrick means when he says:

For any algorithm, there will always be a path that custom hardware engineers can take to beat out general purpose hardware. It’s a fundamental limitation of general purpose hardware.

Combinational logic design, is no magic. For instance, a floating point multiplication operation can not be optimized by means of a magical specialized circuit better than what a modern cpu/gpu is optimized for, all that can be cracked by ASICs is the controller unit, the ALU is already optimized in cpu/gpu technology.

For control unit, a memory hard algorithm like Dagger-Hashimoto (Ethash) will enforce fetch operations to become the bottleneck and no optimization will help the hypothetical ASIC design to go beyond what a modern gpu is capable of.

David Vorick is doing his job as a trojan to convince people about inevitability of ASICs by naive claims about 'fundamental limitation of general purpose hardware', I don't take it as a serious technical assertion, instead imo, it is a weak and ridiculous claim for winning a multi billion dollar war.

It's like the difference between a high performance, yet road-legal, sports car versus a Formula One / NASCAR / Le Mans car.  The latter ones are generally going to be faster because they're purpose built for racing and never have to worry about traffic calming speed bumps like a normal sports car has to cope with.  However, you can bet that if Formula One / NASCAR / Le Mans changed the rules to say the race courses could have speed bumps, the engineers would immediately plow untold sums of money into developing a car that would still be fast over speed bumps.  Hence, why making PoW more complicated isn't going to slow them down for long.

Sorry, but you are performing worse than Vorick, using analogy (the worst reasoning technique ever) and playing with common sense (the worst playground ever) you are trying to convince people that PoW is deemed to be cracked by some magicians overseas.

[aliashraf]

As widely discused before, it is only a matter of time when ASICS (or just specialized hardware in general) will always arise no matter where you go with the PoW configuration. Just pick any algorithm, and I don't see how these fast as hell machines will not start appearing eventually again, so it's more can down the road kicking.

{bolded by me}

It is not "only a matter of time"  but also a matter of flaw. A well designed algorithm, being significantly memory hard while it is mostly utilizing ALU and vectorized calculations would resist ASIC attacks permanently.

"I don't see" how it is possible to see real  machines that can perform extensive multidimensional parallel calculations with a multi gigabyte memory footprint that are cheaper and more efficient than  a modern gpu. I have to emphasis, a real machine and not a magical chip made in china!

[aliashraf]

As widely discused before, it is only a matter of time when ASICS (or just specialized hardware in general) will always arise no matter where you go with the PoW configuration. Just pick any algorithm, and I don't see how these fast as hell machines will not start appearing eventually again, so it's more can down the road kicking.

{bolded by me}

It is not "only a matter of time"  but also a matter of flaw. A well designed algorithm, being significantly memory hard while it is mostly utilizing ALU and vectorized calculations would resist ASIC attacks permanently.

Just like you but on the contrary side, "I don't see" how it is possible to see real  machines that can perform extensive multidimensional parallel calculations with a multi gigabyte memory footprint that are cheaper and more efficient than  a modern gpu. I have to emphasis, a real machine and not a magical chip made in china!

[DooMAD]

Sorry, but you are performing worse than Vorick, using analogy (the worst reasoning technique ever) and playing with common sense (the worst playground ever) you are trying to convince people that PoW is deemed to be cracked by some magicians overseas.

I don't need to convince them of anything.  They can see it happening with their own eyes.  All the reasoning in the world isn't going to change the fact that multiple algorithms now have ASICs designed to mine them.  I'll change my mind when, or if, an algorithm is proven to be uncrackable.  

I also don't understand why you're fixating on the geographical locale.  Would hardware that isn't manufactured overseas be more comforting to you somehow?

For instance, a floating point multiplication operation can not be optimized by means of a magical specialized circuit better than what a modern cpu/gpu is optimized for, all that can be cracked by ASICs is the controller unit, the ALU is already optimized in cpu/gpu technology.

Cool, so should I expect to see your algorithm being used in loads of major coins soon, then?  Clearly you've got it all figured out.  An entire multi-billion dollar industry is about to be disrupted by your revelatory interpretation.

Y'know, that, or your idea won't amount to anything.  It's bound to be one or the other.  

[aliashraf]

Sorry, but you are performing worse than Vorick, using analogy (the worst reasoning technique ever) and playing with common sense (the worst playground ever) you are trying to convince people that PoW is deemed to be cracked by some magicians overseas.

I don't need to convince them of anything.  They can see it happening with their own eyes.  All the reasoning in the world isn't going to change the fact that multiple algorithms now have ASICs designed to mine them.  I'll change my mind when, or if, an algorithm is proven to be uncrackable.  

I also don't understand why you're fixating on the geographical locale.  Would hardware that isn't manufactured overseas be more comforting to you somehow?

Now you are using another worthless technique: induction. It will happen because it has happened before!

First of all, it has not happened yet, actually after years of investment, the most important gpu mining algorithm, Ethash, has not been cracked by ASIC (Bitmain's E3 is not an ASIC as i have mentioned earlier) and Monero has managed to mitigate the attack in few days. We will see how easily Z9s will become obsolete similarly.

For instance, a floating point multiplication operation can not be optimized by means of a magical specialized circuit better than what a modern cpu/gpu is optimized for, all that can be cracked by ASICs is the controller unit, the ALU is already optimized in cpu/gpu technology.

Cool, so should I expect to see your algorithm being used in loads of major coins soon, then?  Clearly you've got it all figured out.  An entire multi-billion dollar industry is about to be disrupted by your revelatory interpretation.

Y'know, that, or your idea won't amount to anything.  It's bound to be one or the other.  

Of course I have good proposals for this problem and I will announce them whenever I found it useful and yes I'm 'nothing' compared to Jihan and his army of engineers and propaganda agents  , but guess what? I'm winning this war because the gpu mining industry is far bigger than Jihan's Bitmain and they will support me and guys like me eventually.