x86: sbrdsp: implement SSE qmf_deint_neg

From 133 (unrolled av_intfloat32 C) to 59 cycles on Arrandale/Win64.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec/x86/sbrdsp.asm

@@ -423,3 +423,25 @@ apply_noise_main:
     add    count, mmsize
     jl      .loop
     RET
+
+INIT_XMM sse
+cglobal sbr_qmf_deint_neg, 2,4,4,v,src,vrev,c
+%define COUNT  32*4
+%define OFFSET 32*4
+    mov        cq, -COUNT
+    lea     vrevq, [vq + OFFSET + COUNT]
+    add        vq, OFFSET-mmsize
+    add      srcq, 2*COUNT
+    mova       m3, [ps_neg]
+.loop:
+    mova       m0, [srcq + 2*cq + 0*mmsize]
+    mova       m1, [srcq + 2*cq + 1*mmsize]
+    shufps     m2, m0, m1, q2020
+    shufps     m1, m0, q1313
+    xorps      m2, m3
+    mova     [vq], m1
+    mova  [vrevq + cq], m2
+    sub        vq, mmsize
+    add        cq, mmsize
+    jl      .loop
+    REP_RET

libavcodec/x86/sbrdsp_init.c

@@ -51,6 +51,8 @@ void ff_sbr_hf_apply_noise_3_sse2(float (*Y)[2], const float *s_m,
                                   const float *q_filt, int noise,
                                   int kx, int m_max);
 
+void ff_sbr_qmf_deint_neg_sse(float *v, const float *src);
+
 av_cold void ff_sbrdsp_init_x86(SBRDSPContext *s)
 {
     int cpu_flags = av_get_cpu_flags();
@@ -63,6 +65,7 @@ av_cold void ff_sbrdsp_init_x86(SBRDSPContext *s)
         s->hf_gen     = ff_sbr_hf_gen_sse;
         s->qmf_post_shuffle = ff_sbr_qmf_post_shuffle_sse;
         s->qmf_deint_bfly   = ff_sbr_qmf_deint_bfly_sse;
+        s->qmf_deint_neg    = ff_sbr_qmf_deint_neg_sse;
     }
 
     if (EXTERNAL_SSE2(cpu_flags)) {