From 30e2f42c82c6c558497aceb5ab0bea24ffeffa3a Mon Sep 17 00:00:00 2001 From: Michael Sevakis Date: Wed, 2 Jun 2010 08:34:10 +0000 Subject: FFT Plugin: Revamp the main code to rid it of 64-bit math. Use 32-bit kiss_fft_scalar because 16-bit integers are generally a poor choice for computation on-target. Simplify display code to speed it up. Add logarithmic frequency display (need keymappings, guessed on some). On dual-core, perform FFT on COP. Add some support function to fixedpoint.c. ... and stuff. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26470 a1c6a512-1295-4272-9138-f99709370657 --- apps/plugins/lib/fixedpoint.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'apps/plugins/lib/fixedpoint.h') diff --git a/apps/plugins/lib/fixedpoint.h b/apps/plugins/lib/fixedpoint.h index ef50dd0085..7cb0098129 100644 --- a/apps/plugins/lib/fixedpoint.h +++ b/apps/plugins/lib/fixedpoint.h @@ -59,6 +59,9 @@ long fp_sqrt(long a, unsigned int fracbits); long fp14_cos(int val); long fp14_sin(int val); long fp16_log(int x); +long fp16_exp(int x); + +unsigned long isqrt(unsigned long x); /* fast unsigned multiplication (16x16bit->32bit or 32x32bit->32bit, * whichever is faster for the architecture) */ -- cgit v1.2.3