From 55aa03b9f8f11ebb7535424cc0e5635558590f49 Mon Sep 17 00:00:00 2001
From: "Ronald S. Bultje" <rsbultje@gmail.com>
Date: Sat, 19 Jan 2013 22:26:58 -0800
Subject: floatdsp: move vector_fmul_add from dsputil to avfloatdsp.

---
 libavutil/ppc/float_dsp_altivec.c | 24 ++++++++++++++++++++++++
 libavutil/ppc/float_dsp_altivec.h |  4 ++++
 libavutil/ppc/float_dsp_init.c    |  1 +
 3 files changed, 29 insertions(+)

(limited to 'libavutil/ppc')

diff --git a/libavutil/ppc/float_dsp_altivec.c b/libavutil/ppc/float_dsp_altivec.c
index e5fd9aba33..87379e92f8 100644
--- a/libavutil/ppc/float_dsp_altivec.c
+++ b/libavutil/ppc/float_dsp_altivec.c
@@ -69,3 +69,27 @@ void ff_vector_fmul_window_altivec(float *dst, const float *src0,
         vec_st(t1, j, dst);
     }
 }
+
+void ff_vector_fmul_add_altivec(float *dst, const float *src0,
+                                const float *src1, const float *src2,
+                                int len)
+{
+    int i;
+    vector float d, s0, s1, s2, t0, t1, edges;
+    vector unsigned char align = vec_lvsr(0,dst),
+                         mask = vec_lvsl(0, dst);
+
+    for (i = 0; i < len - 3; i += 4) {
+        t0 = vec_ld(0, dst + i);
+        t1 = vec_ld(15, dst + i);
+        s0 = vec_ld(0, src0 + i);
+        s1 = vec_ld(0, src1 + i);
+        s2 = vec_ld(0, src2 + i);
+        edges = vec_perm(t1, t0, mask);
+        d = vec_madd(s0, s1, s2);
+        t1 = vec_perm(d, edges, align);
+        t0 = vec_perm(edges, d, align);
+        vec_st(t1, 15, dst + i);
+        vec_st(t0, 0, dst + i);
+    }
+}
diff --git a/libavutil/ppc/float_dsp_altivec.h b/libavutil/ppc/float_dsp_altivec.h
index 4d46edf61a..38a9e15434 100644
--- a/libavutil/ppc/float_dsp_altivec.h
+++ b/libavutil/ppc/float_dsp_altivec.h
@@ -28,4 +28,8 @@ extern void ff_vector_fmul_window_altivec(float *dst, const float *src0,
                                           const float *src1, const float *win,
                                           int len);
 
+extern void ff_vector_fmul_add_altivec(float *dst, const float *src0,
+                                       const float *src1, const float *src2,
+                                       int len);
+
 #endif /* AVUTIL_PPC_FLOAT_DSP_ALTIVEC_H */
diff --git a/libavutil/ppc/float_dsp_init.c b/libavutil/ppc/float_dsp_init.c
index 1134b56926..f3d8a42f53 100644
--- a/libavutil/ppc/float_dsp_init.c
+++ b/libavutil/ppc/float_dsp_init.c
@@ -32,6 +32,7 @@ void ff_float_dsp_init_ppc(AVFloatDSPContext *fdsp, int bit_exact)
         return;
 
     fdsp->vector_fmul = ff_vector_fmul_altivec;
+    fdsp->vector_fmul_add = ff_vector_fmul_add_altivec;
 
     if (!bit_exact) {
         fdsp->vector_fmul_window = ff_vector_fmul_window_altivec;
-- 
cgit v1.2.3