Quote from: kTimesG on Today at 12:26:18 PM

Quote from: RetiredCoder on Today at 08:40:23 AM

When I solved #125 I gave up because it became boring. I thought someone else would solve #130, but nobody did it, so after a year I started again and I spent about two months to solve #130.
Yes you can use old DPs to solve next puzzle, it helps a bit.
I will try to finish solving #135 in 2025.

Sorry but you are beginning to lose credibility (unfortunately not the first sign).

Let's pretend it took 60 days to solve the 129-bits puzzle, and somehow you were also very lucky and only needed 2**64.5 operations (basically half of the time and lower than your supposed 1.23 factor). Let's also pretend that everything was running 24/7 with no pauses, no interruptions, no overheads from storing and matching DPs, no network congestions, so an ideal scenario.

Let's also pretend that you never needed to compute new starting points (this is the biggest question mark I have on you and your algorithms, since it's unfair not to be taken into account).

2**64.5 = 26,087,635,650,665,564,425 EC point additions
60 days = 5,184,000 seconds

That gives 5,032,337,123,971 EC point additions/sec, from start to solve, in an utopian context.

Let's now pretend your super fast kangaroos runs at 10 Gop/s on a RTX 4090.

This gives a required number of 503 GPU cards, all of them running non-stop 2 months, yielding and sending DPs to a central server.

Let's take something like 0.8 $/h cost of running a single RTX 4090.

The cost of running the GPUs for 60 days at that rate would be 576.000 $.

Until one week ago, you didn't take any profit from earlier puzzles, so clearly this wasn't a reinvestment, and there is strong chance the prize may have been less than the costs.

So, why are you doing this? And I'm not talking about solving the puzzles.

I paid 0.25-0.3$ per hour for 4090. The rest of your calculations is correct (roughly).
I got more than I spent, but mostly I do it for fun and prestige, I like world records