![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
bolsonBack in the mid nineties I got a C file I got from my dad who got it from someone who ported a FORTRAN program that people had been using to test the math functions of new supercomputers (Cray, etc.). It doesn't do any clever system benchmarking, it just tests how fast a CPU can do math, in Millions of FLOating Point OPerations per Second, or Mega-FLOPS.
I've since ported it to Java, which with a good 1.6+ JVM runs around 95% of the speed of C compiled '-02 -m64' with gcc on x86_64.
Most recently I've ported it to create a MegaFLOPS benchmark in Javascript. This gets 1-4% of the speed of the C program, but that means between browsers on the same machine there's a 1-4x speed difference.
Oh, and javascript on my current desktop at work can run this faster than native code on my home computer of 12-15 years ago, which in turn compared favorably to the supercomputers of a few years before it.
I've since ported it to Java, which with a good 1.6+ JVM runs around 95% of the speed of C compiled '-02 -m64' with gcc on x86_64.
Most recently I've ported it to create a MegaFLOPS benchmark in Javascript. This gets 1-4% of the speed of the C program, but that means between browsers on the same machine there's a 1-4x speed difference.
Oh, and javascript on my current desktop at work can run this faster than native code on my home computer of 12-15 years ago, which in turn compared favorably to the supercomputers of a few years before it.
