Fix bench Mul128 error

[jmjac]

Benchmark for Mul128 was failing when randomFeltElement was bigger than u128. This should fix it. Also there is (&f.Element{}).SetRandom() to create felts between [0,q). I think we could use it instead of current randomFeltElement implementation unless I'm missing something.

[rodrigo-pino]

Benchmark for Mul128 was failing when randomFeltElement was bigger than u128. This should fix it. Also there is (&f.Element{}).SetRandom() to create felts between [0,q). I think we could use it instead of current randomFeltElement implementation unless I'm missing something.

@jmjac the reason behind not using it is that I was looking for certain determinism in benchmark test. There are lot of variables during benchmarks already, just to make sure we are testing the same thing. I've also noticed that randomFeltElement is not entirely correct since it could output invalid felt numbers, something to fix in the near future

[jmjac]

Makes sense. For the randomFeltElement we could use the similar approach as for u128 and limit the last z[3] to 50bits like

v = binary.BigEndian.AppendUint64(nil, rand.Uint64()>>14)
copy(b[0:8], v)

so that at max we would be using 250bits which is lower than 251 bits of felt.

[rodrigo-pino]

so that at max we would be using 250bits which is lower than 251 bits of felt

I agree, would you like to add this fix as part of this PR?

[jmjac]

@rodrigo-pino added the fix and simplified the function