I definitely appreciate your working on these optimizations, but I am
wondering whether these optimizations could be applied directly to [*~]
and [line~] so that they would be more broadly applied. Also, I like
the idea of an object like [volctl~], but as part of my pd~conf
realization, I think things should definitely be implemented within Pd
itself as much as possible.
.hc
On Dec 28, 2004, at 10:41 AM, Tim Blechmann wrote:
hi all ...
i just uploaded the new version of the volctl~ external to both the cvs and my website (www.geocities.com/themokabar/software.html)
volctl~ does the same as:
| | | | |pack f f| | | | |line~| | | |*~| |
but it's using handcoded sse instructions to save cpu speed ...
changes: - simd optimized ramps - working runtime dispaching
cheers ... tim
-- mailto:TimBlechmann@gmx.de ICQ: 96771783 http://www.mokabar.tk
After one look at this planet any visitor from outer space would say "I want to see the manager." William S. Burroughs
PD-announce mailing list PD-announce@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-announce
Man has survived hitherto because he was too ignorant to know how to
realize his wishes.
Now that he can realize them, he must either change them, or perish.
-William Carlos Williams
I definitely appreciate your working on these optimizations, but I am wondering whether these optimizations could be applied directly to [*~] and [line~] so that they would be more broadly applied. Also, I like the idea of an object like [volctl~], but as part of my pd~conf realization, I think things should definitely be implemented within Pd itself as much as possible.
well, a part of the optimizing could be done directly on |line~| ... implementing simd optimized slopes ...
for the |*~| part, volctl~ uses the features of both |*~| and |*~ 0|, the first doing a vector-vector multiplication, the second one a vector-scalar multiplication. for the time of constant volume, line~ would output a constant number stream and |*~| would do a vector-vector multiplication for something that could be done as vector-scalar multiplication...
to have the same behaviour with |*~| and |line~| one would have to rewrite the pd dsp tree adding (the flag for constant messages) and adapt all arithmetic tilde functions... could be possible, but i think more a task for weeks than for hours...
i really don't like to write externals for trivial things, but i also don't want to use my cpu to computing something that isn't necessary (the higher my cpu load, the louder the fan of my shitty laptop)...
cheers ... tim
-
Hans-Christoph Steiner -
Tim Blechmann