Thank you for all the congrats.
I started this particular search January 11 afternoon. I had access
to the network of computers at Mathematical Institute of Wroclaw
15 computers with 64-bit Athlons
15 computers with double core 64-bit Pentium D 805 (running 2 threads, but both
threads combined were just slightly more productive than a single
30 32-bit Athlons 2500 and 15 Durons 900 - but 32-bit computers were
very slow on my program, so you can count them all perhaps as about
5 64-bit machines.
All were under Linux, with my own C-code. They were in students labs, so
they were working only partly during the day and mostly at night (however
sometimes few of them were not accessible at night too).
I estimate the storage memory used by the program was in the range 100-200kB,
so should fit in cache, at least on better machines.
It is hard to tell what was the exact cumulative time, perhaps
4,000+-1,000 hours of a single 64-bit machine would be a reasonable
Now, the truth is that I was quite lucky too. While running the program
I got 4 AP22's, rediscovered one known AP23, and the AP24 was the 6th
progression of length at least 22. I think I did deserve to find AP24
this year, but getting it after a week instead of many months required a
lot of luck.