avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for AVC intra prediction functions

This patch adds MSA (MIPS-SIMD-Arch) optimizations for AVC intra prediction functions in new file h264pred_msa.c
Adds new generic macros (needed for this patch) in libavutil/mips/generic_macros_msa.h
Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec/h264pred.c

@@ -594,4 +594,5 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
 
     if (ARCH_ARM) ff_h264_pred_init_arm(h, codec_id, bit_depth, chroma_format_idc);
     if (ARCH_X86) ff_h264_pred_init_x86(h, codec_id, bit_depth, chroma_format_idc);
+    if (ARCH_MIPS) ff_h264_pred_init_mips(h, codec_id, bit_depth, chroma_format_idc);
 }

libavcodec/h264pred.h

@@ -117,5 +117,7 @@ void ff_h264_pred_init_arm(H264PredContext *h, int codec_id,
                            const int bit_depth, const int chroma_format_idc);
 void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
                            const int bit_depth, const int chroma_format_idc);
+void ff_h264_pred_init_mips(H264PredContext *h, int codec_id,
+                            const int bit_depth, const int chroma_format_idc);
 
 #endif /* AVCODEC_H264PRED_H */

libavcodec/mips/Makefile

@@ -22,6 +22,7 @@ OBJS-$(CONFIG_HEVC_DECODER)               += mips/hevcdsp_init_mips.o      \
                                              mips/hevcpred_init_mips.o
 OBJS-$(CONFIG_H264DSP)                    += mips/h264dsp_init_mips.o
 OBJS-$(CONFIG_H264CHROMA)                 += mips/h264chroma_init_mips.o
+OBJS-$(CONFIG_H264PRED)                   += mips/h264pred_init_mips.o
 MSA-OBJS-$(CONFIG_HEVC_DECODER)           += mips/hevcdsp_msa.o            \
                                              mips/hevc_mc_uni_msa.o        \
                                              mips/hevc_mc_uniw_msa.o       \
@@ -32,5 +33,6 @@ MSA-OBJS-$(CONFIG_HEVC_DECODER)           += mips/hevcdsp_msa.o            \
                                              mips/hevcpred_msa.o
 MSA-OBJS-$(CONFIG_H264DSP)                += mips/h264dsp_msa.o
 MSA-OBJS-$(CONFIG_H264CHROMA)             += mips/h264chroma_msa.o
+MSA-OBJS-$(CONFIG_H264PRED)               += mips/h264pred_msa.o
 LOONGSON3-OBJS-$(CONFIG_H264DSP)          += mips/h264dsp_mmi.o
 LOONGSON3-OBJS-$(CONFIG_H264CHROMA)       += mips/h264chroma_mmi.o

libavcodec/mips/h264dsp_mips.h

@@ -68,6 +68,33 @@ void ff_weight_h264_pixels8_8_msa(uint8_t *src, int stride, int height,
 void ff_weight_h264_pixels4_8_msa(uint8_t *src, int stride, int height,
                                   int log2_denom, int weight, int offset);
 
+void ff_h264_intra_predict_plane_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_predict_dc_4blk_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_predict_hor_dc_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_predict_vert_dc_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_predict_mad_cow_dc_l0t_8x8_msa(uint8_t *src,
+                                                  ptrdiff_t stride);
+void ff_h264_intra_predict_mad_cow_dc_0lt_8x8_msa(uint8_t *src,
+                                                  ptrdiff_t stride);
+void ff_h264_intra_predict_mad_cow_dc_l00_8x8_msa(uint8_t *src,
+                                                  ptrdiff_t stride);
+void ff_h264_intra_predict_mad_cow_dc_0l0_8x8_msa(uint8_t *src,
+                                                  ptrdiff_t stride);
+void ff_h264_intra_predict_plane_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_vert_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_horiz_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_dc_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_vert_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_horiz_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_dc_left_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_dc_top_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_dc_128_8x8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_h264_intra_pred_dc_128_16x16_msa(uint8_t *src, ptrdiff_t stride);
+void ff_vp8_pred8x8_127_dc_8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_vp8_pred8x8_129_dc_8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_vp8_pred16x16_127_dc_8_msa(uint8_t *src, ptrdiff_t stride);
+void ff_vp8_pred16x16_129_dc_8_msa(uint8_t *src, ptrdiff_t stride);
+
 void ff_h264_weight_pixels16_8_mmi(uint8_t *block, int stride, int height,
         int log2_denom, int weight, int offset);
 void ff_h264_biweight_pixels16_8_mmi(uint8_t *dst, uint8_t *src,

libavcodec/mips/h264pred_init_mips.c

+/*
+ * Copyright (c) 2015 Shivraj Patil (Shivraj.Patil@imgtec.com)
+ *
+ * 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 "h264dsp_mips.h"
+
+#if HAVE_MSA
+static av_cold void h264_pred_init_msa(H264PredContext *h, int codec_id,
+                                       const int bit_depth,
+                                       const int chroma_format_idc)
+{
+    if (8 == bit_depth) {
+        if (chroma_format_idc == 1) {
+            h->pred8x8[VERT_PRED8x8] = ff_h264_intra_pred_vert_8x8_msa;
+            h->pred8x8[HOR_PRED8x8] = ff_h264_intra_pred_horiz_8x8_msa;
+        }
+
+        if (codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8) {
+            if (chroma_format_idc == 1) {
+                h->pred8x8[PLANE_PRED8x8] = ff_h264_intra_predict_plane_8x8_msa;
+            }
+        }
+        if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7
+            && codec_id != AV_CODEC_ID_VP8) {
+            if (chroma_format_idc == 1) {
+                h->pred8x8[DC_PRED8x8] = ff_h264_intra_predict_dc_4blk_8x8_msa;
+                h->pred8x8[LEFT_DC_PRED8x8] =
+                    ff_h264_intra_predict_hor_dc_8x8_msa;
+                h->pred8x8[TOP_DC_PRED8x8] =
+                    ff_h264_intra_predict_vert_dc_8x8_msa;
+                h->pred8x8[ALZHEIMER_DC_L0T_PRED8x8] =
+                    ff_h264_intra_predict_mad_cow_dc_l0t_8x8_msa;
+                h->pred8x8[ALZHEIMER_DC_0LT_PRED8x8] =
+                    ff_h264_intra_predict_mad_cow_dc_0lt_8x8_msa;
+                h->pred8x8[ALZHEIMER_DC_L00_PRED8x8] =
+                    ff_h264_intra_predict_mad_cow_dc_l00_8x8_msa;
+                h->pred8x8[ALZHEIMER_DC_0L0_PRED8x8] =
+                    ff_h264_intra_predict_mad_cow_dc_0l0_8x8_msa;
+            }
+        } else {
+            if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {
+                h->pred8x8[7] = ff_vp8_pred8x8_127_dc_8_msa;
+                h->pred8x8[8] = ff_vp8_pred8x8_129_dc_8_msa;
+            }
+        }
+
+        if (chroma_format_idc == 1) {
+            h->pred8x8[DC_128_PRED8x8] = ff_h264_intra_pred_dc_128_8x8_msa;
+        }
+
+        h->pred16x16[DC_PRED8x8] = ff_h264_intra_pred_dc_16x16_msa;
+        h->pred16x16[VERT_PRED8x8] = ff_h264_intra_pred_vert_16x16_msa;
+        h->pred16x16[HOR_PRED8x8] = ff_h264_intra_pred_horiz_16x16_msa;
+
+        switch (codec_id) {
+        case AV_CODEC_ID_SVQ3:
+            ;
+            break;
+        case AV_CODEC_ID_RV40:
+            ;
+            break;
+        case AV_CODEC_ID_VP7:
+        case AV_CODEC_ID_VP8:
+            h->pred16x16[7] = ff_vp8_pred16x16_127_dc_8_msa;
+            h->pred16x16[8] = ff_vp8_pred16x16_129_dc_8_msa;
+            break;
+        default:
+            h->pred16x16[PLANE_PRED8x8] =
+                ff_h264_intra_predict_plane_16x16_msa;
+            break;
+        }
+
+        h->pred16x16[LEFT_DC_PRED8x8] = ff_h264_intra_pred_dc_left_16x16_msa;
+        h->pred16x16[TOP_DC_PRED8x8] = ff_h264_intra_pred_dc_top_16x16_msa;
+        h->pred16x16[DC_128_PRED8x8] = ff_h264_intra_pred_dc_128_16x16_msa;
+    }
+}
+#endif  // #if HAVE_MSA
+
+av_cold void ff_h264_pred_init_mips(H264PredContext *h, int codec_id,
+                                    int bit_depth,
+                                    const int chroma_format_idc)
+{
+#if HAVE_MSA
+    h264_pred_init_msa(h, codec_id, bit_depth, chroma_format_idc);
+#endif  // #if HAVE_MSA
+}

libavutil/mips/generic_macros_msa.h

@@ -2307,4 +2307,15 @@
                                                           \
     SW4(out0_m, out1_m, out2_m, out3_m, pdst, stride);    \
 }
+
+/* Description : Pack even byte elements and store byte vector in destination
+                 memory
+   Arguments   : Inputs  - in0, in1, pdst
+*/
+#define PCKEV_ST_SB(in0, in1, pdst)                   \
+{                                                     \
+    v16i8 tmp_m;                                      \
+    tmp_m = __msa_pckev_b((v16i8) in1, (v16i8) in0);  \
+    ST_SB(tmp_m, (pdst));                             \
+}
 #endif  /* AVUTIL_MIPS_GENERIC_MACROS_MSA_H */