Нам надо договориться, чтобы не считать одно и то же.
I'm not sure how valid that is when we are using different techniques, but ok.
I usually have 6 processes running. I am currently working on full minimization for T(44,7), T(10,7), T(14,7) and T(50,4), on capped minimization (

) for T(24,12), and more randomly for any example of T(24,15).
"Full minimization" means I generate all possible candidates up to the known upper bound, to prove the minimal value. I do this by fixing prime powers in each element to satisfy the odd divisors of

, then factorize what has not been fixed to verify whether a solution has been found.
"Capped minimization" means I only generate candidates in which the fixed primes go up to some limit. I do this with the intent of reducing the upper bound, so that a future full minimization can complete faster.
My T(24,15) search is currently doing a capped search (

) with

and fixing powers of 2..13 such that

. A similar approach for T(24,14) found 9721439902882994590514319997146 after about 4 days, so we'll see what happens with this one.
Since the last update I also found improved bounds for T(9,5), T(14,7), T(15,4), T(21,4), T(22,7), T(26,7), T(33,4), T(34,7), T(36,7), T(39,4), T(50,5). Nothing newly proved minimal - I had believed that T(44,7) and T(10,7) were complete, but I found that I was relying on a buggy implementation of sqrtmod(), and have to redo about 3 weeks of calculations.