It's always possible to accelerate math with function specific silicon, the market just doesn't bother for monero because the leadership tends to rotate the hash algo quite frequently.
A state actor or central bank wouldn't be producing monero asics to generate revenue though they'd be doing it to covertly perform a massive reorg or mine empty blocks.
Login to reply
Replies (17)
BITMAIN
View Bitmain
It is not always possible, there's absolutely no basis in this truth for this. It maybe possible, it may not be possible, the certainty in your statement is false.
Yes, it is possible to create a RandomX ASIC, but by the time you have optimised for every possible RandomX function, you have built an entire CPU. Probably an inferior one, at vast expense. But it will run DOOM, so not a total loss.
That one "RandomX ASIC" commercially available turned out to be a bunch of RISC-V CPUs taped together in parallel lol
They are RISC-Vs, I believe.
CPUs built to run in parallel
Don't think that counts as possible lol. Thats the entire point of RandomX so that you have to do this to parrelellize it.
Literally every single time someone has created an "asic resistant" algo over the last 15 years it has turned out to be false.
Bitmain X5 is a randomx asic that already exists.
Enlighten me on how it works.
DYOR, I don't care enough. I've seen this over and over again, asic resistant algorithms are in the same category as free energy machines.
Or let me enlighten you: it doesnt, it's just a good cpu
Weaponized ignorance.
Every generation of ASIC-resistant algos has been more resistant than the last.
RandomX is so close to Turing-complete that no viable ASIC has emerged less complex than a CPU.
This is what a 100% win rate looks like
When it was first released it was apparently twice as efficient as a normal CPU at randomx.
In any case, I would not bet against a massive black budget being able to build a more efficient randomx solver.
I know theoretically you would need to create much of the CPU, but in reality "life finds a way" every single time. Perhaps this time its different, but even then it just means the enemy needs to buy or rent general purpose compute which is readily available.
The enemy occupies the administrative "high ground", and always has.
The enemy can vet who can buy SHA256 ASIC mining rigs, quite easily, and likely does. The enemy can control who gets approval for a utility-scale electricity interconnect to power said mining rigs. (And takes fricking years doing it sometimes.) The enemy can see who is running any big mining pool, and "liase" with them. To prevent misunderstandings. And prevent processing ML/TF/CSAM/protesters/splitting bills. You want to stay legal, don't you?
The enemy finds it very challenging to control who can buy general-purpose compute, as Iraq, Iran, DPRK and most recently Russia could tell you.
It was tried, still cheaper to just use existing CPUs
Not really, and none of this changes the math. It's multiple orders of magnitude cheaper to permanently kill monero than bitcoin.
And its multiple zeroes easier to permanently capture Bitcoin than Monero.
Both these problems are real.
One of these problems is solvable.