Chris Rolfe

Does anyone know if there's any advantage to using these routines to, say, block copy a buffer~ float32 samples into a local signal double64 rather than looping sample-by-sample with type casts? Are they optimized? I dread to think what 32 type casts per sample does to the pipeline. I guess I could benchmark both ways myself but I'm lazy ;)

Similarly, in z_dsp.h:
copy_32from64
set_zero64
Does anyone know if there's any advantage to using these routines to, say, block copy a buff

copy_64from32

No benefit... just there for convenience. Code is below.

void copy_32from64(t_float *dst, const t_double *src, long n)
{
    while (n--)
        *dst++ = *src++;
}

void copy_32from64(t_float *dst, const t_double *src, long n)
{
	while (n--)
		*dst++ = *src++;
}