lavc/aarch64: add a few SIMD functions for AAC PS

☭ tests/checkasm/checkasm --bench --test=aacpsdsp
checkasm: using random seed 3318985180
MMX implied by specified flags
MMX implied by specified flags
NEON:
 - aacpsdsp.add_squares        [OK]
 - aacpsdsp.mul_pair_single    [OK]
 - aacpsdsp.hybrid_analysis    [OK]
 - aacpsdsp.stereo_interpolate [OK]
checkasm: all 5 tests passed
nop: 10.0
ps_add_squares_c: 63221.2
ps_add_squares_neon: 22311.7
ps_hybrid_analysis_c: 2466.6
ps_hybrid_analysis_neon: 1521.9
ps_mul_pair_single_c: 68592.0
ps_mul_pair_single_neon: 17426.6
ps_stereo_interpolate_c: 72344.3
ps_stereo_interpolate_neon: 72308.8
ps_stereo_interpolate_ipdopd_c: 117415.2
ps_stereo_interpolate_ipdopd_neon: 113386.3

3 files changed, 199 insertions, 0 deletions

diff --git a/libavcodec/aarch64/Makefile b/libavcodec/aarch64/Makefile
index 104bc67802..d440b1b18a 100644
--- a/libavcodec/aarch64/Makefile
+++ b/libavcodec/aarch64/Makefile
@@ -11,6 +11,7 @@ OBJS-$(CONFIG_NEON_CLOBBER_TEST)        += aarch64/neontest.o
 OBJS-$(CONFIG_VIDEODSP)                 += aarch64/videodsp_init.o
 
 # decoders/encoders
+OBJS-$(CONFIG_AAC_DECODER)              += aarch64/aacpsdsp_init_aarch64.o
 OBJS-$(CONFIG_DCA_DECODER)              += aarch64/synth_filter_init.o
 OBJS-$(CONFIG_RV40_DECODER)             += aarch64/rv40dsp_init_aarch64.o
 OBJS-$(CONFIG_VC1DSP)                   += aarch64/vc1dsp_init_aarch64.o
@@ -42,6 +43,7 @@ NEON-OBJS-$(CONFIG_MDCT)                += aarch64/mdct_neon.o
 NEON-OBJS-$(CONFIG_MPEGAUDIODSP)        += aarch64/mpegaudiodsp_neon.o
 
 # decoders/encoders
+NEON-OBJS-$(CONFIG_AAC_DECODER)         += aarch64/aacpsdsp_neon.o
 NEON-OBJS-$(CONFIG_DCA_DECODER)         += aarch64/synth_filter_neon.o
 NEON-OBJS-$(CONFIG_VORBIS_DECODER)      += aarch64/vorbisdsp_neon.o
 NEON-OBJS-$(CONFIG_VP9_DECODER)         += aarch64/vp9itxfm_16bpp_neon.o       \
diff --git a/libavcodec/aarch64/aacpsdsp_init_aarch64.c b/libavcodec/aarch64/aacpsdsp_init_aarch64.c
new file mode 100644
index 0000000000..c15094c45c
--- /dev/null
+++ b/libavcodec/aarch64/aacpsdsp_init_aarch64.c
@@ -0,0 +1,48 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "config.h"
+
+#include "libavutil/aarch64/cpu.h"
+#include "libavcodec/aacpsdsp.h"
+
+void ff_ps_add_squares_neon(float *dst, const float (*src)[2], int n);
+void ff_ps_mul_pair_single_neon(float (*dst)[2], float (*src0)[2],
+                                float *src1, int n);
+void ff_ps_hybrid_analysis_neon(float (*out)[2], float (*in)[2],
+                                const float (*filter)[8][2],
+                                int stride, int n);
+void ff_ps_stereo_interpolate_neon(float (*l)[2], float (*r)[2],
+                                   float h[2][4], float h_step[2][4],
+                                   int len);
+void ff_ps_stereo_interpolate_ipdopd_neon(float (*l)[2], float (*r)[2],
+                                          float h[2][4], float h_step[2][4],
+                                          int len);
+
+av_cold void ff_psdsp_init_aarch64(PSDSPContext *s)
+{
+    int cpu_flags = av_get_cpu_flags();
+
+    if (have_neon(cpu_flags)) {
+        s->add_squares           = ff_ps_add_squares_neon;
+        s->mul_pair_single       = ff_ps_mul_pair_single_neon;
+        s->hybrid_analysis       = ff_ps_hybrid_analysis_neon;
+        s->stereo_interpolate[0] = ff_ps_stereo_interpolate_neon;
+        s->stereo_interpolate[1] = ff_ps_stereo_interpolate_ipdopd_neon;
+    }
+}
diff --git a/libavcodec/aarch64/aacpsdsp_neon.S b/libavcodec/aarch64/aacpsdsp_neon.S
new file mode 100644
index 0000000000..6662971199
--- /dev/null
+++ b/libavcodec/aarch64/aacpsdsp_neon.S
@@ -0,0 +1,149 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/aarch64/asm.S"
+
+function ff_ps_add_squares_neon, export=1
+1:      ld1         {v0.4S,v1.4S}, [x1], #32
+        fmul        v0.4S, v0.4S, v0.4S
+        fmul        v1.4S, v1.4S, v1.4S
+        faddp       v2.4S, v0.4S, v1.4S
+        ld1         {v3.4S}, [x0]
+        fadd        v3.4S, v3.4S, v2.4S
+        st1         {v3.4S}, [x0], #16
+        subs        w2, w2, #4
+        b.gt        1b
+        ret
+endfunc
+
+function ff_ps_mul_pair_single_neon, export=1
+1:      ld1         {v0.4S,v1.4S}, [x1], #32
+        ld1         {v2.4S},       [x2], #16
+        zip1        v3.4S, v2.4S, v2.4S
+        zip2        v4.4S, v2.4S, v2.4S
+        fmul        v0.4S, v0.4S, v3.4S
+        fmul        v1.4S, v1.4S, v4.4S
+        st1         {v0.4S,v1.4S}, [x0], #32
+        subs        w3, w3, #4
+        b.gt        1b
+        ret
+endfunc
+
+function ff_ps_stereo_interpolate_neon, export=1
+        ld1         {v0.4S}, [x2]
+        ld1         {v1.4S}, [x3]
+        zip1        v4.4S, v0.4S, v0.4S
+        zip2        v5.4S, v0.4S, v0.4S
+        zip1        v6.4S, v1.4S, v1.4S
+        zip2        v7.4S, v1.4S, v1.4S
+1:      ld1         {v2.2S}, [x0]
+        ld1         {v3.2S}, [x1]
+        fadd        v4.4S, v4.4S, v6.4S
+        fadd        v5.4S, v5.4S, v7.4S
+        mov         v2.D[1], v2.D[0]
+        mov         v3.D[1], v3.D[0]
+        fmul        v2.4S, v2.4S, v4.4S
+        fmla        v2.4S, v3.4S, v5.4S
+        st1         {v2.D}[0], [x0], #8
+        st1         {v2.D}[1], [x1], #8
+        subs        w4, w4, #1
+        b.gt        1b
+        ret
+endfunc
+
+function ff_ps_stereo_interpolate_ipdopd_neon, export=1
+        ld1         {v0.4S,v1.4S}, [x2]
+        ld1         {v6.4S,v7.4S}, [x3]
+        fneg        v2.4S, v1.4S
+        fneg        v3.4S, v7.4S
+        zip1        v16.4S, v0.4S, v0.4S
+        zip2        v17.4S, v0.4S, v0.4S
+        zip1        v18.4S, v2.4S, v1.4S
+        zip2        v19.4S, v2.4S, v1.4S
+        zip1        v20.4S, v6.4S, v6.4S
+        zip2        v21.4S, v6.4S, v6.4S
+        zip1        v22.4S, v3.4S, v7.4S
+        zip2        v23.4S, v3.4S, v7.4S
+1:      ld1         {v2.2S}, [x0]
+        ld1         {v3.2S}, [x1]
+        fadd        v16.4S, v16.4S, v20.4S
+        fadd        v17.4S, v17.4S, v21.4S
+        mov         v2.D[1], v2.D[0]
+        mov         v3.D[1], v3.D[0]
+        fmul        v4.4S, v2.4S, v16.4S
+        fmla        v4.4S, v3.4S, v17.4S
+        fadd        v18.4S, v18.4S, v22.4S
+        fadd        v19.4S, v19.4S, v23.4S
+        ext         v2.16B, v2.16B, v2.16B, #4
+        ext         v3.16B, v3.16B, v3.16B, #4
+        fmla        v4.4S, v2.4S, v18.4S
+        fmla        v4.4S, v3.4S, v19.4S
+        st1         {v4.D}[0], [x0], #8
+        st1         {v4.D}[1], [x1], #8
+        subs        w4, w4, #1
+        b.gt        1b
+        ret
+endfunc
+
+function ff_ps_hybrid_analysis_neon, export=1
+        sxtw        x3, w3
+        lsl         x3, x3, #3
+        ld2         {v0.4S,v1.4S}, [x1], #32
+        ld2         {v2.2S,v3.2S}, [x1], #16
+        ld1         {v24.2S},      [x1], #8
+        ld2         {v4.2S,v5.2S}, [x1], #16
+        ld2         {v6.4S,v7.4S}, [x1]
+        rev64       v6.4S, v6.4S
+        rev64       v7.4S, v7.4S
+        ext         v6.16B, v6.16B, v6.16B, #8
+        ext         v7.16B, v7.16B, v7.16B, #8
+        rev64       v4.2S, v4.2S
+        rev64       v5.2S, v5.2S
+        mov         v2.D[1], v3.D[0]
+        mov         v4.D[1], v5.D[0]
+        mov         v5.D[1], v2.D[0]
+        mov         v3.D[1], v4.D[0]
+        fadd        v16.4S, v0.4S, v6.4S
+        fadd        v17.4S, v1.4S, v7.4S
+        fsub        v18.4S, v1.4S, v7.4S
+        fsub        v19.4S, v0.4S, v6.4S
+        fadd        v22.4S, v2.4S, v4.4S
+        fsub        v23.4S, v5.4S, v3.4S
+        trn1        v20.2D, v22.2D, v23.2D      // {re4+re8, re5+re7, im8-im4, im7-im5}
+        trn2        v21.2D, v22.2D, v23.2D      // {im4+im8, im5+im7, re4-re8, re5-re7}
+1:      ld2         {v2.4S,v3.4S}, [x2], #32
+        ld2         {v4.2S,v5.2S}, [x2], #16
+        ld1         {v6.2S},       [x2], #8
+        add         x2, x2, #8
+        mov         v4.D[1], v5.D[0]
+        mov         v6.S[1], v6.S[0]
+        fmul        v6.2S, v6.2S, v24.2S
+        fmul        v0.4S, v2.4S, v16.4S
+        fmul        v1.4S, v2.4S, v17.4S
+        fmls        v0.4S, v3.4S, v18.4S
+        fmla        v1.4S, v3.4S, v19.4S
+        fmla        v0.4S, v4.4S, v20.4S
+        fmla        v1.4S, v4.4S, v21.4S
+        faddp       v0.4S, v0.4S, v1.4S
+        faddp       v0.4S, v0.4S, v0.4S
+        fadd        v0.2S, v0.2S, v6.2S
+        st1         {v0.2S}, [x0], x3
+        subs        w4, w4, #1
+        b.gt        1b
+        ret
+endfunc