[FFmpeg-devel] [PATCH v2 3/4] avcodec/aarch64/hevcdsp: add idct_dc NEON

[Josh Dekker]

Signed-off-by: Josh Dekker <josh at itanimul.li>
---
 libavcodec/aarch64/hevcdsp_idct_neon.S    | 54 +++++++++++++++++++++++
 libavcodec/aarch64/hevcdsp_init_aarch64.c | 16 +++++++
 2 files changed, 70 insertions(+)

diff --git a/libavcodec/aarch64/hevcdsp_idct_neon.S b/libavcodec/aarch64/hevcdsp_idct_neon.S
index 329038a958..d3902a9e0f 100644
--- a/libavcodec/aarch64/hevcdsp_idct_neon.S
+++ b/libavcodec/aarch64/hevcdsp_idct_neon.S
@@ -5,6 +5,7 @@
  *
  * Ported from arm/hevcdsp_idct_neon.S by
  * Copyright (c) 2020 Reimar Döffinger
+ * Copyright (c) 2020 Josh Dekker
  *
  * This file is part of FFmpeg.
  *
@@ -568,3 +569,56 @@ tr_16x4 secondpass_10, 20 - 10, 512, 1
 
 idct_16x16 8
 idct_16x16 10
+
+// void ff_hevc_idct_NxN_dc_DEPTH_neon(int16_t *coeffs)
+.macro idct_dc size bitdepth
+function ff_hevc_idct_\size\()x\size\()_dc_\bitdepth\()_neon, export=1
+        ldrsh   w1, [x0]
+        mov     w2,  #(1 << (13 - \bitdepth))
+        add     w1,  w1, #1
+        asr     w1,  w1, #1
+        add     w1,  w1, w2
+        asr     w1,  w1, #(14 - \bitdepth)
+        dup  v0.8h,  w1
+        dup  v1.8h,  w1
+.if \size > 4
+        dup  v2.8h,  w1
+        dup  v3.8h,  w1
+.if \size > 16 /* dc 32x32 */
+        mov     x2,  #4
+1:
+        subs    x2,  x2, #1
+.endif
+        add    x12,  x0,  #64
+        mov    x13,  #128
+.if \size > 8 /* dc 16x16 */
+        st1   {v0.8h-v3.8h}, [ x0], x13
+        st1   {v0.8h-v3.8h}, [x12], x13
+        st1   {v0.8h-v3.8h}, [ x0], x13
+        st1   {v0.8h-v3.8h}, [x12], x13
+        st1   {v0.8h-v3.8h}, [ x0], x13
+        st1   {v0.8h-v3.8h}, [x12], x13
+.endif /* dc 8x8 */
+        st1   {v0.8h-v3.8h}, [ x0], x13
+        st1   {v0.8h-v3.8h}, [x12], x13
+.if \size > 16 /* dc 32x32 */
+        bne 1b
+.endif
+.else /* dc 4x4 */
+        st1   {v0.8h-v1.8h}, [x0]
+.endif
+        ret
+endfunc
+.endm
+
+idct_dc 4 8
+idct_dc 4 10
+
+idct_dc 8 8
+idct_dc 8 10
+
+idct_dc 16 8
+idct_dc 16 10
+
+idct_dc 32 8
+idct_dc 32 10
diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c
index 4c29daa6d5..fe111bd1ac 100644
--- a/libavcodec/aarch64/hevcdsp_init_aarch64.c
+++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c
@@ -45,6 +45,14 @@ void ff_hevc_idct_8x8_8_neon(int16_t *coeffs, int col_limit);
 void ff_hevc_idct_8x8_10_neon(int16_t *coeffs, int col_limit);
 void ff_hevc_idct_16x16_8_neon(int16_t *coeffs, int col_limit);
 void ff_hevc_idct_16x16_10_neon(int16_t *coeffs, int col_limit);
+void ff_hevc_idct_4x4_dc_8_neon(int16_t *coeffs);
+void ff_hevc_idct_8x8_dc_8_neon(int16_t *coeffs);
+void ff_hevc_idct_16x16_dc_8_neon(int16_t *coeffs);
+void ff_hevc_idct_32x32_dc_8_neon(int16_t *coeffs);
+void ff_hevc_idct_4x4_dc_10_neon(int16_t *coeffs);
+void ff_hevc_idct_8x8_dc_10_neon(int16_t *coeffs);
+void ff_hevc_idct_16x16_dc_10_neon(int16_t *coeffs);
+void ff_hevc_idct_32x32_dc_10_neon(int16_t *coeffs);
 
 av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
 {
@@ -57,6 +65,10 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
         c->add_residual[3]             = ff_hevc_add_residual_32x32_8_neon;
         c->idct[1]                     = ff_hevc_idct_8x8_8_neon;
         c->idct[2]                     = ff_hevc_idct_16x16_8_neon;
+        c->idct_dc[0]                  = ff_hevc_idct_4x4_dc_8_neon;
+        c->idct_dc[1]                  = ff_hevc_idct_8x8_dc_8_neon;
+        c->idct_dc[2]                  = ff_hevc_idct_16x16_dc_8_neon;
+        c->idct_dc[3]                  = ff_hevc_idct_32x32_dc_8_neon;
     }
     if (bit_depth == 10) {
         c->add_residual[0]             = ff_hevc_add_residual_4x4_10_neon;
@@ -65,5 +77,9 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
         c->add_residual[3]             = ff_hevc_add_residual_32x32_10_neon;
         c->idct[1]                     = ff_hevc_idct_8x8_10_neon;
         c->idct[2]                     = ff_hevc_idct_16x16_10_neon;
+        c->idct_dc[0]                  = ff_hevc_idct_4x4_dc_10_neon;
+        c->idct_dc[1]                  = ff_hevc_idct_8x8_dc_10_neon;
+        c->idct_dc[2]                  = ff_hevc_idct_16x16_dc_10_neon;
+        c->idct_dc[3]                  = ff_hevc_idct_32x32_dc_10_neon;
     }
 }
-- 
2.24.3 (Apple Git-128)