[libav-devel] [PATCH 1/2] lavc/qsvenc_jpeg: add async_depth support

[Zhong Li]

Currently qsv (m)jpeg encoding is broken.
Regression introducing by the commit(id: c1bcd3): fix async support,
which requires the minimum async_depth to be 1, instead previous zero.
But the default async_depth of qsv (m)jpeg encoding is still initialized
(mostly) as zero.

This patch also abviously improves qsv (m)jpeg encoding performance
due to the default async_depth is changed to 4.

Signed-off-by: Zhong Li <***@intel.com>
---
libavcodec/qsvenc_jpeg.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/libavcodec/qsvenc_jpeg.c b/libavcodec/qsvenc_jpeg.c
index e11b74d..7465bb9 100644
--- a/libavcodec/qsvenc_jpeg.c
+++ b/libavcodec/qsvenc_jpeg.c
@@ -64,6 +64,7 @@ static av_cold int qsv_enc_close(AVCodecContext *avctx)
#define OFFSET(x) offsetof(QSVMJPEGEncContext, x)
#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
static const AVOption options[] = {
+ { "async_depth", "Maximum processing parallelism", OFFSET(qsv.async_depth), AV_OPT_TYPE_INT, { .i64 = ASYNC_DEPTH_DEFAULT }, 1, INT_MAX, VE },
{ NULL },
};

[Zhong Li]

Keep alignment with vaapi mjpeg encoder.

Signed-off-by: Zhong Li <***@intel.com>
---
libavcodec/qsvenc_jpeg.c | 6 ++++++
1 file changed, 6 insertions(+)

diff --git a/libavcodec/qsvenc_jpeg.c b/libavcodec/qsvenc_jpeg.c
index 7465bb9..bbfd009 100644
--- a/libavcodec/qsvenc_jpeg.c
+++ b/libavcodec/qsvenc_jpeg.c
@@ -75,6 +75,11 @@ static const AVClass class = {
.version = LIBAVUTIL_VERSION_INT,
};

+static const AVCodecDefault qsv_enc_defaults[] = {
+ { "global_quality", "80" },
+ { NULL },
+};
+
AVCodec ff_mjpeg_qsv_encoder = {
.name = "mjpeg_qsv",
.long_name = NULL_IF_CONFIG_SMALL("MJPEG (Intel Quick Sync Video acceleration)"),
@@ -89,5 +94,6 @@ AVCodec ff_mjpeg_qsv_encoder = {
AV_PIX_FMT_QSV,
AV_PIX_FMT_NONE },
.priv_class = &class,
+ .defaults = qsv_enc_defaults,
.wrapper_name = "qsv",
};

[Luca Barbato]

Post by Zhong Li
Currently qsv (m)jpeg encoding is broken.
Regression introducing by the commit(id: c1bcd3): fix async support,
which requires the minimum async_depth to be 1, instead previous zero.
But the default async_depth of qsv (m)jpeg encoding is still initialized
(mostly) as zero.
This patch also abviously improves qsv (m)jpeg encoding performance
due to the default async_depth is changed to 4.
---
libavcodec/qsvenc_jpeg.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavcodec/qsvenc_jpeg.c b/libavcodec/qsvenc_jpeg.c
index e11b74d..7465bb9 100644
--- a/libavcodec/qsvenc_jpeg.c
+++ b/libavcodec/qsvenc_jpeg.c
@@ -64,6 +64,7 @@ static av_cold int qsv_enc_close(AVCodecContext *avctx)
#define OFFSET(x) offsetof(QSVMJPEGEncContext, x)
#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
static const AVOption options[] = {
+ { "async_depth", "Maximum processing parallelism", OFFSET(qsv.async_depth), AV_OPT_TYPE_INT, { .i64 = ASYNC_DEPTH_DEFAULT }, 1, INT_MAX, VE },
{ NULL },
};

both look fine. I'd merge them this evening.

lu

[Li, Zhong]

Barbato
Sent: Thursday, October 11, 2018 5:34 PM
Subject: Re: [libav-devel] [PATCH 1/2] lavc/qsvenc_jpeg: add async_depth
support

Post by Zhong Li
Currently qsv (m)jpeg encoding is broken.
Regression introducing by the commit(id: c1bcd3): fix async support,
which requires the minimum async_depth to be 1, instead previous zero.
But the default async_depth of qsv (m)jpeg encoding is still
initialized
(mostly) as zero.
This patch also abviously improves qsv (m)jpeg encoding performance
due to the default async_depth is changed to 4.
---
libavcodec/qsvenc_jpeg.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavcodec/qsvenc_jpeg.c b/libavcodec/qsvenc_jpeg.c index
e11b74d..7465bb9 100644
--- a/libavcodec/qsvenc_jpeg.c
+++ b/libavcodec/qsvenc_jpeg.c
@@ -64,6 +64,7 @@ static av_cold int qsv_enc_close(AVCodecContext
*avctx) #define OFFSET(x) offsetof(QSVMJPEGEncContext, x) #define

VE

Post by Zhong Li
AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM

static const

Post by Zhong Li
AVOption options[] = {
+ { "async_depth", "Maximum processing parallelism",
+ OFFSET(qsv.async_depth), AV_OPT_TYPE_INT, { .i64 =
+ ASYNC_DEPTH_DEFAULT }, 1, INT_MAX, VE },
{ NULL },
};

both look fine. I'd merge them this evening.
lu

Thanks for quick reply, : )