Re: ideal interface for Random Number Generators?
On Jun 11, 10:09 pm, Keith H Duggar <dug...@alum.mit.edu> wrote:
On Jun 11, 4:41 pm, Pete Becker <p...@versatilecoding.com> wrote:
First Alf pulling up all this idiotic "xyz must be stupid" because
he was ignorant of the basic facts of digital division algorithms,
second James (very uncharacteristically) making equally misinformed
statements about division and then refusing to come clean when hard
facts were presented proving him wrong,
Just to be clear: the only *hard* fact that has been presented
is that I did run my benchmark on a Sparc, about a year ago, and
on that machine, with the version of Sun CC I was using,
division took exactly the same time as addition, subtraction and
multiplication. I don't currently have access to a Sparc, so I
can't rerun it, and I don't know exactly what the configuration
was I ran it on, but on at least one Sparc, under Solaris,
division isn't any slower than any of the other four operations.
Out of curiosity, I did rerun the benchmark on the PC's (under
Windows) I currently have access to, and there, division is
slower.
(Another interesting point was that on earlier Sparcs, floating
point division was faster than fixed point.)
Anyway, my statements in this regard were based on concrete
measurements. The code which I used to make the measurements is
available online (whenever my provider decides to make it so).
So you can easily test the situation on your particular
hardware. Expect to find something different than what other
people find, however, because hardware varies enormously.
--
James Kanze