From 40c3c117e79112700cf5192ce0c966ad881cfd89 Mon Sep 17 00:00:00 2001 From: Peter Meerwald Date: Tue, 2 Jul 2013 11:43:35 +0200 Subject: [PATCH] sconv: avoid multiply in ARM NEON float->s16 conversion optimization idea taken from libavresample Signed-off-by: Peter Meerwald --- src/pulsecore/sconv_neon.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/src/pulsecore/sconv_neon.c b/src/pulsecore/sconv_neon.c index ddcdcc2d0..6f1a3c9f7 100644 --- a/src/pulsecore/sconv_neon.c +++ b/src/pulsecore/sconv_neon.c @@ -36,12 +36,9 @@ static void pa_sconv_s16le_from_f32ne_neon(unsigned n, const float *src, int16_t "movs %[n], %[n], lsr #2 \n\t" "beq 2f \n\t" - "vdup.f32 q1, %[scale] \n\t" - "1: \n\t" "vld1.32 {q0}, [%[src]]! \n\t" - "vmul.f32 q0, q0, q1 \n\t" /* scale */ - "vcvt.s32.f32 q0, q0, #16 \n\t" /* s32<-f32 as 16:16 fixed-point */ + "vcvt.s32.f32 q0, q0, #31 \n\t" /* s32<-f32 as 16:16 fixed-point, with implicit multiplication by 32768 */ "vqrshrn.s32 d0, q0, #16 \n\t" /* shift, round, narrow */ "subs %[n], %[n], #1 \n\t" "vst1.16 {d0}, [%[dst]]! \n\t" @@ -50,8 +47,8 @@ static void pa_sconv_s16le_from_f32ne_neon(unsigned n, const float *src, int16_t "2: \n\t" : [dst] "+r" (dst), [src] "+r" (src), [n] "+r" (n) /* output operands (or input operands that get modified) */ - : [scale] "r" (32768.0f) /* input operands */ - : "memory", "cc", "q0", "q1" /* clobber list */ + : /* input operands */ + : "memory", "cc", "q0" /* clobber list */ ); /* leftovers */