avcodec/mips: [loongson] optimize put_hevc_qpel_hv_8 with mmi.

Optimize put_hevc_qpel_hv_8 with mmi in the case width=4/8/12/16/24/32/48/64.
This optimization improved HEVC decoding performance 11%(1.81x to 2.01x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/mips/hevcdsp_init_mips.c

@@ -25,6 +25,15 @@ static av_cold void hevc_dsp_init_mmi(HEVCDSPContext *c,
                                       const int bit_depth)
 {
     if (8 == bit_depth) {
+        c->put_hevc_qpel[1][1][1] = ff_hevc_put_hevc_qpel_hv4_8_mmi;
+        c->put_hevc_qpel[3][1][1] = ff_hevc_put_hevc_qpel_hv8_8_mmi;
+        c->put_hevc_qpel[4][1][1] = ff_hevc_put_hevc_qpel_hv12_8_mmi;
+        c->put_hevc_qpel[5][1][1] = ff_hevc_put_hevc_qpel_hv16_8_mmi;
+        c->put_hevc_qpel[6][1][1] = ff_hevc_put_hevc_qpel_hv24_8_mmi;
+        c->put_hevc_qpel[7][1][1] = ff_hevc_put_hevc_qpel_hv32_8_mmi;
+        c->put_hevc_qpel[8][1][1] = ff_hevc_put_hevc_qpel_hv48_8_mmi;
+        c->put_hevc_qpel[9][1][1] = ff_hevc_put_hevc_qpel_hv64_8_mmi;
+
         c->put_hevc_qpel_bi[3][0][0] = ff_hevc_put_hevc_pel_bi_pixels8_8_mmi;
         c->put_hevc_qpel_bi[5][0][0] = ff_hevc_put_hevc_pel_bi_pixels16_8_mmi;
         c->put_hevc_qpel_bi[6][0][0] = ff_hevc_put_hevc_pel_bi_pixels24_8_mmi;

libavcodec/mips/hevcdsp_mips.h

@@ -480,16 +480,33 @@ void ff_hevc_addblk_32x32_msa(uint8_t *dst, int16_t *pi16Coeffs,
 void ff_hevc_idct_luma_4x4_msa(int16_t *pi16Coeffs);
 
 /* Loongson optimization */
-#define L_BI_MC(PEL, DIR, WIDTH, TYPE)                                         \
-void ff_hevc_put_hevc_##PEL##_bi_##DIR##WIDTH##_8_##TYPE(uint8_t *dst,       \
-                                                        ptrdiff_t dst_stride,  \
-                                                        uint8_t *src,          \
-                                                        ptrdiff_t src_stride,  \
-                                                        int16_t *src_16bit,    \
-                                                        int height,            \
-                                                        intptr_t mx,           \
-                                                        intptr_t my,           \
-                                                        int width)
+#define L_MC(PEL, DIR, WIDTH, TYPE)                                          \
+void ff_hevc_put_hevc_##PEL##_##DIR##WIDTH##_8_##TYPE(int16_t *dst,          \
+                                                      uint8_t *src,          \
+                                                      ptrdiff_t src_stride,  \
+                                                      int height,            \
+                                                      intptr_t mx,           \
+                                                      intptr_t my,           \
+                                                      int width)
+L_MC(qpel, hv, 4, mmi);
+L_MC(qpel, hv, 8, mmi);
+L_MC(qpel, hv, 12, mmi);
+L_MC(qpel, hv, 16, mmi);
+L_MC(qpel, hv, 24, mmi);
+L_MC(qpel, hv, 32, mmi);
+L_MC(qpel, hv, 48, mmi);
+L_MC(qpel, hv, 64, mmi);
+
+#define L_BI_MC(PEL, DIR, WIDTH, TYPE)                                          \
+void ff_hevc_put_hevc_##PEL##_bi_##DIR##WIDTH##_8_##TYPE(uint8_t *dst,          \
+                                                         ptrdiff_t dst_stride,  \
+                                                         uint8_t *src,          \
+                                                         ptrdiff_t src_stride,  \
+                                                         int16_t *src_16bit,    \
+                                                         int height,            \
+                                                         intptr_t mx,           \
+                                                         intptr_t my,           \
+                                                         int width)
 
 L_BI_MC(pel, pixels, 8, mmi);
 L_BI_MC(pel, pixels, 16, mmi);

libavutil/mips/mmiutils.h

@@ -250,6 +250,15 @@
       : "memory"                                                \
     );
 
+/**
+ * brief: Transpose 2X2 word packaged data.
+ * fr_i0, fr_i1: src
+ * fr_o0, fr_o1: dst
+ */
+#define TRANSPOSE_2W(fr_i0, fr_i1, fr_o0, fr_o1)                          \
+        "punpcklwd  "#fr_o0",   "#fr_i0",   "#fr_i1"                \n\t" \
+        "punpckhwd  "#fr_o1",   "#fr_i0",   "#fr_i1"                \n\t"
+
 /**
  * brief: Transpose 4X4 half word packaged data.
  * fr_i0, fr_i1, fr_i2, fr_i3: src & dst