p.s. sieve.lua is around 12% faster with % instead of math.mod() :)
That's probably mainly because it saves a function call? -- Wim
