On Dec 13, 2005, at 0:27, Joshua Isom wrote:
I made the changes, adding the argv error handling(none of the docs
state any necessity to handle no args so I'm not sure if it's
necessary from their benchmarking point of view), and using ints when
possible.
Thanks.
If you were getting odd results, a difference between the c version
and mine is the c version is a while loop and mine's a do while loop,
off by one. Right now I'm getting around 1.8 seconds.
Well, the C version would loop ~2^32 times for N=0 and random.pasm for
N=0 or N=1. Also having:
set N2, 100.0
inside the loop isn't necessary. OTOH it doesn't influence the timing
at all due to CPU pipelines.
Checked in r10486
Thanks,
leo
