Hi,
As it is, the arm inline assembly is mostly experimental and doesn't provide
that much improvement. If you really care about performance, you should
optimize at the function level (or at least at the loop level). Otherwise,
don't bother.
Jean-Marc
Selon Tomasz Pyra <hellfire@ivo.pl>:
> How big is difference between encoding speeds of libspeex compiled with
> arm-asm, and without it?
> My application is written in EVC++, but there is no inline assembler, so
> I have to compile speex without arm-asm.
>
> But I still trying to find a way to use speex with arm-asm under my
> EVC++ application.
>
> I tried to compile speex with arm-asm under gcc (arm-wince-pe), but
> generated static link library (libspeex.a) is incompatible with EVC++
> linker.
> Maybe somebody know how to compile speex with arm-asm, and then use is
> (as static or shared) under EVC++?
>
> Maybe somebody know ARM assembler good enough to rewrite __asm parts od
> speex using opcodes inserted by __emit - this is only few instructions
> in fixed_arm.h
>
>
>
> _______________________________________________
> Speex-dev mailing list
> Speex-dev@xiph.org
> http://lists.xiph.org/mailman/listinfo/speex-dev
>
>