> Current versions of golang are not as fast as GCC C/C++ optimized code by at least a factor of two
This is, in general, not true at all: I've written numeric code as fast as C code compiled by GCC -O3 Il giorno martedì 14 giugno 2016 14:31:18 UTC+2, gordo...@gmail.com ha scritto: > > That's interesting but your time or 0.7 seconds on your specified machine > isn't very informative: you should at least specify how long the original > version of primegen (which is not multithreaded) takes, which we assume as > about 0.7 seconds, and also how long your golang version takes without > multithreading (or set "maxprocs := 1") Also, one should really convert > Bernstein's "eratspeed" to golang for comparison. > > Current versions of golang are not as fast as GCC C/C++ optimized code by > at least a factor of two, and you will find that if one applies maximum > wheel factorization (ie a 2/3/5/7 wheel plus pre-culling for primes of > 11/13/17/19), rather than the limited 2/3/5 wheel factorization Bernstein > used in eratspeed, that the Sieve of Eratosthenes will beat "primegen" for > every range when comparing like for like (multithreading or not for both, > written in the same language), and that golang will currently always be a > factor slower than the C/C++ code due to less efficient compilation. -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
