We're slower than Rubinius here :(. https://github.com/kostya/benchmarks/tree/a2d10ddfeea208309fd53d666a533445a97ae264/base64 A promising attempt would be to replace the translation table with explicit code and to process strings in register sized chunks instead of byte-wise.
dmd* is slow, gdc is faster than every other implementation in the benchmark. http://forum.dlang.org/thread/kdxidecfhivdmonazrra@forum.dlang.org?page=2#post-kgolgxdoptsikvtxfrrp:40forum.dlang.org is this bug still valid?
Sure, the 2 things I named are still valid ideas to speed up the function.
THIS ISSUE HAS BEEN MOVED TO GITHUB https://github.com/dlang/phobos/issues/9657 DO NOT COMMENT HERE ANYMORE, NOBODY WILL SEE IT, THIS ISSUE HAS BEEN MOVED TO GITHUB