[pizlonator]

Real programs have to do IO and the C and C++ code runs faster while also doing IO.

What do you think they could have done better assuming that the IO is a necessary part of the benchmark?

Also good job to the Rust devs for making the benchmark so much faster in nightly. I wonder what they did.

[igouy]

> Assuming that the IO is a necessary part of the benchmark …

So make it significant, move giga like benchmarks game reverse-complement & fasta & …

https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

[Aurornis]

The file I/O is probably irrelevant, but the startup time is not.

The differences among the really fast languages are probably in different startup times if I had to guess.

[igouy]

What if we try to measure rather than guess :-)

https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

[pizlonator]

> differences among the really fast languages are probably in different startup times

Startup times matter a great deal.