Thanks Eric,
I don't really mind either way, I was just getting confused trying to
understand the standard. Now I am clear.
Ross.
Eric Scheirer wrote:
>Hi Ross,
>
>>...
>>asig stereoIn[2],stereoOut[2];
>>...
>>stereoOut = flange( stereoIn );
>>...
>
>
>This is not directly legal -- core opcodes always take
>scalar arguments. One thing to do would be to write
>a wrapper function as a user-defined opcode that does
>the vector processing.
>
>Loop-based vector processing can be done as
>
>asig x[2],y[2],i;
>oparray flange[2];
>
>i = 0; while (i < 2) {
> y[i] = flange[i](x[i]);
> i = i + 1;
>}
>
>Note that the semantics are subtly different than you
>requested -- rather than one multi-channel flanger, this
>is an array of single-channel flangers. The effect will
>be the same though.
>
>Vectorized core opcodes are something that really should
>have been in SAOL, and are tops on my list for the first
>time we do improvements to the language. It was a mistake
>not to include this functionality once arrays of singles
>and opcodes were available -- it would make for incredibly
>efficient compilation, too!
>
>Best,
>
> -- Eric
>
>
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 11:46:35 EST