Those need to look at sets of only three, respectively four integers. For the default n, that’s 200×200, respectively 200 times faster ⇒ I don’t think it makes

Those need to look at sets of only three, respectively four integers. For the default n, that’s 200×200, respectively 200 times faster ⇒ I don’t think it makes much of a difference. Also, the paper talks of “as the smallest instances in which four fifth powers sum to a fifth power” I guess they also looked for sums of 4th powers, but didn’t find 95800⁴ + 217519⁴ + 414560⁴ = 422481⁴ (Found on a connection machine in 1988) or any of the even larger ones.