Is it weird that for all the talk of C bindings for GMP, an actual C program wasn’t included in the sample?
When you have a single entity that blows the scale of the graph out that much, it would be nice to offer a second graph with everyone else. Or just exclude PHP from the graph entirely and put in a note. But I’m not a visual/designer type.
Click and drag on the graph to zoom in.
As much as I love that PHP’s arbitrary precision arithmetic functions take and return strings, the benchmark code should probably use the gmp_ family of functions and not the bc ones: http://php.net/manual/en/ref.gmp.php
Isn’t it kind of pointless to use OpenJDK7 as a data point, given that the implementation was reworked from the ground up in JDK8?