Post
Topic
Board Development & Technical Discussion
Re: Pollard's kangaroo ECDLP solver
by
yoyodapro
on 21/09/2021, 18:23:31 UTC
" Google announced it has a quantum computer that is 100 million times faster than any classical computer in its lab. "

...except that even Google's classical computers are all hosted in datacenters with stupidly high specs and speeds, a magnitude faster than even enterprise servers I can rent from any old reseller or cloud provider. [I can rent 24-core servers at some places].

Hold that thought there.

Say you have a "classical computer" that's really just an aggregation of a bunch of dedicated systems similar to the ones powering their clouds. Now if you put enough of these together (as they probably did - into a cloud of course - remember that private clouds are also a thing), of course you will be able to make a classical computer that outperforms any other system.

Now considering that Google already has some of the most powerful classical computers, it's a no-brainer that they will also have one of the most powerful quantum computers as well.

Are they going to pass that on to end-users? Of course not. We're talking about internal hardware that powers things like Google Search and Youtube. They would never sell or even lease the hardware to third-parties, because they have no use for all that collective raw power.

Now a dumbed-down version of their QC that's much slower than your 100 million-x fast Google QC (we are talking around 100M x the speed of a PC) - that's more viable to be used in commercial settings (and hacker-oafs who attempt to use this to search the full 160-bits of BTC addresses, or the 256-bits of private keys).

Now 100M log 2 is about 26.5.

Assume the best single-PC build can crack 50 bits like a champ (and just for kicks, assume a cluster of 1000 of these can do 60 bits).

That means your rad, spiffy "x100M" commercial QC can only crack 76.5 bits feasibly (86.5 bits in a cluster of 1000). Nice, will solve you a few stubborn puzzles like #64+ and earn you a couple thousand dollars of BTC, but nowhere near enough to threaten the security of secp256k1 or even HASH160.

I would +1 Merit this but I have no more sMerit to give Sad

Very good explanation, also I did post the whole Quantpy Beta thing as a joke, obviously its infeasible to simulate anything close to 63 Qubits with a classical computer.

However there are a ton of quantum simulators out there but to have the same effect as a quantum computer it takes alot of ram to simulate. 2.9TB is not far fetched for a classical computer to run a very basic quantum simulation. This is primarily for testing purposes and even something as simple as bit flipping requires a tremendous amount of processing power, the speed increase in my testing doesnt even compare to the most basic of BSGS programs.