Browse code

avcodec/nvdec: Implement vp8 hwaccel

Philip Langdale authored on 2017/11/20 01:42:39
Showing 8 changed files
... ...
@@ -13,7 +13,7 @@ version <next>:
13 13
 - PCE support for extended channel layouts in the AAC encoder
14 14
 - native aptX encoder and decoder
15 15
 - Raw aptX muxer and demuxer
16
-- NVIDIA NVDEC-accelerated H.264, HEVC, MPEG-1/2/4, VC1 and VP9 hwaccel decoding
16
+- NVIDIA NVDEC-accelerated H.264, HEVC, MPEG-1/2/4, VC1, VP8/9 hwaccel decoding
17 17
 - Intel QSV-accelerated overlay filter
18 18
 - mcompand audio filter
19 19
 - acontrast audio filter
... ...
@@ -2746,6 +2746,8 @@ vc1_vaapi_hwaccel_deps="vaapi"
2746 2746
 vc1_vaapi_hwaccel_select="vc1_decoder"
2747 2747
 vc1_vdpau_hwaccel_deps="vdpau"
2748 2748
 vc1_vdpau_hwaccel_select="vc1_decoder"
2749
+vp8_nvdec_hwaccel_deps="nvdec"
2750
+vp8_nvdec_hwaccel_select="vp8_decoder"
2749 2751
 vp8_vaapi_hwaccel_deps="vaapi VAPictureParameterBufferVP8"
2750 2752
 vp8_vaapi_hwaccel_select="vp8_decoder"
2751 2753
 vp9_d3d11va_hwaccel_deps="d3d11va DXVA_PicParams_VP9"
... ...
@@ -871,6 +871,7 @@ OBJS-$(CONFIG_VC1_NVDEC_HWACCEL)          += nvdec_vc1.o
871 871
 OBJS-$(CONFIG_VC1_QSV_HWACCEL)            += qsvdec_other.o
872 872
 OBJS-$(CONFIG_VC1_VAAPI_HWACCEL)          += vaapi_vc1.o
873 873
 OBJS-$(CONFIG_VC1_VDPAU_HWACCEL)          += vdpau_vc1.o
874
+OBJS-$(CONFIG_VP8_NVDEC_HWACCEL)          += nvdec_vp8.o
874 875
 OBJS-$(CONFIG_VP8_VAAPI_HWACCEL)          += vaapi_vp8.o
875 876
 OBJS-$(CONFIG_VP9_D3D11VA_HWACCEL)        += dxva2_vp9.o
876 877
 OBJS-$(CONFIG_VP9_DXVA2_HWACCEL)          += dxva2_vp9.o
... ...
@@ -59,6 +59,7 @@ extern const AVHWAccel ff_vc1_dxva2_hwaccel;
59 59
 extern const AVHWAccel ff_vc1_nvdec_hwaccel;
60 60
 extern const AVHWAccel ff_vc1_vaapi_hwaccel;
61 61
 extern const AVHWAccel ff_vc1_vdpau_hwaccel;
62
+extern const AVHWAccel ff_vp8_nvdec_hwaccel;
62 63
 extern const AVHWAccel ff_vp8_vaapi_hwaccel;
63 64
 extern const AVHWAccel ff_vp9_d3d11va_hwaccel;
64 65
 extern const AVHWAccel ff_vp9_d3d11va2_hwaccel;
... ...
@@ -58,6 +58,7 @@ static int map_avcodec_id(enum AVCodecID id)
58 58
     case AV_CODEC_ID_MPEG2VIDEO: return cudaVideoCodec_MPEG2;
59 59
     case AV_CODEC_ID_MPEG4:      return cudaVideoCodec_MPEG4;
60 60
     case AV_CODEC_ID_VC1:        return cudaVideoCodec_VC1;
61
+    case AV_CODEC_ID_VP8:        return cudaVideoCodec_VP8;
61 62
     case AV_CODEC_ID_VP9:        return cudaVideoCodec_VP9;
62 63
     case AV_CODEC_ID_WMV3:       return cudaVideoCodec_VC1;
63 64
     }
64 65
new file mode 100644
... ...
@@ -0,0 +1,97 @@
0
+/*
1
+ * VP8 HW decode acceleration through NVDEC
2
+ *
3
+ * Copyright (c) 2017 Philip Langdale
4
+ *
5
+ * This file is part of FFmpeg.
6
+ *
7
+ * FFmpeg is free software; you can redistribute it and/or
8
+ * modify it under the terms of the GNU Lesser General Public
9
+ * License as published by the Free Software Foundation; either
10
+ * version 2.1 of the License, or (at your option) any later version.
11
+ *
12
+ * FFmpeg is distributed in the hope that it will be useful,
13
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
14
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15
+ * Lesser General Public License for more details.
16
+ *
17
+ * You should have received a copy of the GNU Lesser General Public
18
+ * License along with FFmpeg; if not, write to the Free Software
19
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20
+ */
21
+
22
+#include "avcodec.h"
23
+#include "nvdec.h"
24
+#include "decode.h"
25
+#include "internal.h"
26
+#include "vp8.h"
27
+
28
+static unsigned char safe_get_ref_idx(VP8Frame *frame)
29
+{
30
+    return frame ? ff_nvdec_get_ref_idx(frame->tf.f) : 255;
31
+}
32
+
33
+static int nvdec_vp8_start_frame(AVCodecContext *avctx, const uint8_t *buffer, uint32_t size)
34
+{
35
+    VP8Context *h = avctx->priv_data;
36
+
37
+    NVDECContext      *ctx = avctx->internal->hwaccel_priv_data;
38
+    CUVIDPICPARAMS     *pp = &ctx->pic_params;
39
+    FrameDecodeData *fdd;
40
+    NVDECFrame *cf;
41
+    AVFrame *cur_frame = h->framep[VP56_FRAME_CURRENT]->tf.f;
42
+
43
+    int ret;
44
+
45
+    ret = ff_nvdec_start_frame(avctx, cur_frame);
46
+    if (ret < 0)
47
+        return ret;
48
+
49
+    fdd = (FrameDecodeData*)cur_frame->private_ref->data;
50
+    cf  = (NVDECFrame*)fdd->hwaccel_priv;
51
+
52
+    *pp = (CUVIDPICPARAMS) {
53
+        .PicWidthInMbs     = (cur_frame->width  + 15) / 16,
54
+        .FrameHeightInMbs  = (cur_frame->height + 15) / 16,
55
+        .CurrPicIdx        = cf->idx,
56
+
57
+        .CodecSpecific.vp8 = {
58
+            .width                       = cur_frame->width,
59
+            .height                      = cur_frame->height,
60
+
61
+            .first_partition_size        = h->header_partition_size,
62
+
63
+            .LastRefIdx                  = safe_get_ref_idx(h->framep[VP56_FRAME_PREVIOUS]),
64
+            .GoldenRefIdx                = safe_get_ref_idx(h->framep[VP56_FRAME_GOLDEN]),
65
+            .AltRefIdx                   = safe_get_ref_idx(h->framep[VP56_FRAME_GOLDEN2]),
66
+
67
+            .frame_type                  = !h->keyframe,
68
+            .version                     = h->profile,
69
+            .show_frame                  = !h->invisible,
70
+            .update_mb_segmentation_data = h->segmentation.enabled ? h->segmentation.update_feature_data : 0,
71
+       }
72
+    };
73
+
74
+    return 0;
75
+}
76
+
77
+static int nvdec_vp8_frame_params(AVCodecContext *avctx,
78
+                                  AVBufferRef *hw_frames_ctx)
79
+{
80
+    // VP8 uses a fixed size pool of 3 possible reference frames
81
+    return ff_nvdec_frame_params(avctx, hw_frames_ctx, 3);
82
+}
83
+
84
+AVHWAccel ff_vp8_nvdec_hwaccel = {
85
+    .name                 = "vp8_nvdec",
86
+    .type                 = AVMEDIA_TYPE_VIDEO,
87
+    .id                   = AV_CODEC_ID_VP8,
88
+    .pix_fmt              = AV_PIX_FMT_CUDA,
89
+    .start_frame          = nvdec_vp8_start_frame,
90
+    .end_frame            = ff_nvdec_simple_end_frame,
91
+    .decode_slice         = ff_nvdec_simple_decode_slice,
92
+    .frame_params         = nvdec_vp8_frame_params,
93
+    .init                 = ff_nvdec_decode_init,
94
+    .uninit               = ff_nvdec_decode_uninit,
95
+    .priv_data_size       = sizeof(NVDECContext),
96
+};
... ...
@@ -29,7 +29,7 @@
29 29
 
30 30
 #define LIBAVCODEC_VERSION_MAJOR  58
31 31
 #define LIBAVCODEC_VERSION_MINOR   6
32
-#define LIBAVCODEC_VERSION_MICRO 100
32
+#define LIBAVCODEC_VERSION_MICRO 101
33 33
 
34 34
 #define LIBAVCODEC_VERSION_INT  AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \
35 35
                                                LIBAVCODEC_VERSION_MINOR, \
... ...
@@ -2602,6 +2602,9 @@ int vp78_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
2602 2602
 #if CONFIG_VP8_VAAPI_HWACCEL
2603 2603
             AV_PIX_FMT_VAAPI,
2604 2604
 #endif
2605
+#if CONFIG_VP8_NVDEC_HWACCEL
2606
+            AV_PIX_FMT_CUDA,
2607
+#endif
2605 2608
             AV_PIX_FMT_YUV420P,
2606 2609
             AV_PIX_FMT_NONE,
2607 2610
         };
... ...
@@ -2950,6 +2953,9 @@ AVCodec ff_vp8_decoder = {
2950 2950
 #if CONFIG_VP8_VAAPI_HWACCEL
2951 2951
                                HWACCEL_VAAPI(vp8),
2952 2952
 #endif
2953
+#if CONFIG_VP8_NVDEC_HWACCEL
2954
+                               HWACCEL_NVDEC(vp8),
2955
+#endif
2953 2956
                                NULL
2954 2957
                            },
2955 2958
 };