On Nov 20, 2005, at 2:09 PM, Mathieu Bouchard wrote:
On Sun, 20 Nov 2005, Frank Barknecht wrote:
I intend to do a benchmark session on all the list abs, that have equivalents in other libraries like Zexy. I'm curious, how much slower the abstraction versions actually are and which of them maybe should
be included directly into Pd's [list], because the performance hit is too big.I think that the ones which should be put directly in [list] are either the ones who have bad time orders (e.g. the duration of the operation
is proportional to the square of the number of elements...), or just the
ones which are used most often in high-speed situations (even if
indirectly).Better again is to try figuring out what else could be coded in C in a
way that abstractions can use it and become faster and so need not be
recoded in C themselves. It's all about having as little C as possible.
You took the works right out of my mouth! I think that once we start
thinking about Pd as a language in itself, then people will start to
think how to optimize it as well. There is no doubt in my mind that C
code would be faster than Pd for most things, but, like Java, Ruby,
SmallTalk, there is something to be gained by using it to program.
Assembly is even faster than C, so why aren't people writing objects in
assembly?
.hc
http://at.or.at/hans/