Message ID | 20201210060510.6594-1-wenbin.chen@intel.com |
---|---|
State | Superseded |
Headers | show |
Series | [FFmpeg-devel] libavcodec/qsvenc: add low latency P-pyramid support for qsv | expand |
Context | Check | Description |
---|---|---|
andriy/x86_make | success | Make finished |
andriy/x86_make_fate | success | Make fate finished |
andriy/PPC64_make | success | Make finished |
andriy/PPC64_make_fate | success | Make fate finished |
On Thu, Dec 10, 2020 at 2:07 PM <wenbin.chen@intel.com> wrote: > > From: Wenbinc-Bin <wenbin.chen@intel.com> > > Add low latency P-pyramid support for qsv, and it relates to a new > command line parameter "-p_strategy". To enable this flag, user also > need to set "-bf" to -1 or 0. IMO correcting the related options and prompting a warning would be more user-friendly. > Signed-off-by Wenbin Chen <wenbin.chen@intel.com> > --- > libavcodec/qsvenc.c | 22 ++++++++++++++++++++++ > libavcodec/qsvenc.h | 2 ++ > 2 files changed, 24 insertions(+) > > diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c > index 2bd2a56227..c3b41374e4 100644 > --- a/libavcodec/qsvenc.c > +++ b/libavcodec/qsvenc.c > @@ -268,6 +268,13 @@ static void dump_video_param(AVCodecContext *avctx, QSVEncContext *q, > case MFX_B_REF_PYRAMID: av_log(avctx, AV_LOG_VERBOSE, "pyramid"); break; > default: av_log(avctx, AV_LOG_VERBOSE, "auto"); break; > } > + av_log(avctx, AV_LOG_VERBOSE, "; PRefType: "); > + switch(co3->PRefType){ > + case MFX_P_REF_DEFAULT: av_log(avctx, AV_LOG_VERBOSE, "default"); break; > + case MFX_P_REF_SIMPLE: av_log(avctx, AV_LOG_VERBOSE, "simple"); break; > + case MFX_P_REF_PYRAMID: av_log(avctx, AV_LOG_VERBOSE, "pyramid"); break; > + default: break; > + } > av_log(avctx, AV_LOG_VERBOSE, "\n"); > #endif > > @@ -777,6 +784,21 @@ FF_ENABLE_DEPRECATION_WARNINGS > #if QSV_HAVE_CO3 > q->extco3.Header.BufferId = MFX_EXTBUFF_CODING_OPTION3; > q->extco3.Header.BufferSz = sizeof(q->extco3); > + switch(q->p_strategy){ > + case 0: > + q->extco3.PRefType = MFX_P_REF_DEFAULT; > + break; > + case 1: > + q->extco3.PRefType = MFX_P_REF_SIMPLE; > + break; > + case 2: > + q->extco3.PRefType = MFX_P_REF_PYRAMID; > + break; > + default: > + q->extco3.PRefType = MFX_P_REF_DEFAULT; > + av_log(avctx, AV_LOG_VERBOSE, "invalid p_strategy, set to default\n"); > + break; > + } It'll be great if one could help to provide some links/background information like [1] in commit messages or docs, to elaborate more about "default" and "simple", and make a new option easier to be understood: [1] https://github.com/Intel-Media-SDK/MediaSDK/blob/master/doc/mediasdk-man.md#preftype - linjie
> -----Original Message----- > From: Linjie Fu <linjie.justin.fu@gmail.com> > Sent: Thursday, December 10, 2020 3:49 PM > To: FFmpeg development discussions and patches <ffmpeg- > devel@ffmpeg.org> > Cc: Chen, Wenbin <wenbin.chen@intel.com> > Subject: Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: add low latency P- > pyramid support for qsv > > On Thu, Dec 10, 2020 at 2:07 PM <wenbin.chen@intel.com> wrote: > > > > From: Wenbinc-Bin <wenbin.chen@intel.com> > > > > Add low latency P-pyramid support for qsv, and it relates to a new > > command line parameter "-p_strategy". To enable this flag, user also > > need to set "-bf" to -1 or 0. > > IMO correcting the related options and prompting a warning would be more > user-friendly. > > > Signed-off-by Wenbin Chen <wenbin.chen@intel.com> > > --- > > libavcodec/qsvenc.c | 22 ++++++++++++++++++++++ libavcodec/qsvenc.h > > | 2 ++ > > 2 files changed, 24 insertions(+) > > > > diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index > > 2bd2a56227..c3b41374e4 100644 > > --- a/libavcodec/qsvenc.c > > +++ b/libavcodec/qsvenc.c > > @@ -268,6 +268,13 @@ static void dump_video_param(AVCodecContext > *avctx, QSVEncContext *q, > > case MFX_B_REF_PYRAMID: av_log(avctx, AV_LOG_VERBOSE, > "pyramid"); break; > > default: av_log(avctx, AV_LOG_VERBOSE, "auto"); break; > > } > > + av_log(avctx, AV_LOG_VERBOSE, "; PRefType: "); > > + switch(co3->PRefType){ > > + case MFX_P_REF_DEFAULT: av_log(avctx, AV_LOG_VERBOSE, > "default"); break; > > + case MFX_P_REF_SIMPLE: av_log(avctx, AV_LOG_VERBOSE, "simple"); > break; > > + case MFX_P_REF_PYRAMID: av_log(avctx, AV_LOG_VERBOSE, > "pyramid"); break; > > + default: break; > > + } > > av_log(avctx, AV_LOG_VERBOSE, "\n"); #endif > > > > @@ -777,6 +784,21 @@ FF_ENABLE_DEPRECATION_WARNINGS #if > QSV_HAVE_CO3 > > q->extco3.Header.BufferId = MFX_EXTBUFF_CODING_OPTION3; > > q->extco3.Header.BufferSz = sizeof(q->extco3); > > + switch(q->p_strategy){ > > + case 0: > > + q->extco3.PRefType = MFX_P_REF_DEFAULT; > > + break; > > + case 1: > > + q->extco3.PRefType = MFX_P_REF_SIMPLE; > > + break; > > + case 2: > > + q->extco3.PRefType = MFX_P_REF_PYRAMID; > > + break; > > + default: > > + q->extco3.PRefType = MFX_P_REF_DEFAULT; > > + av_log(avctx, AV_LOG_VERBOSE, "invalid p_strategy, set to > default\n"); > > + break; > > + } > > It'll be great if one could help to provide some links/background information > like [1] in commit messages or docs, to elaborate more about "default" and > "simple", and make a new option easier to be understood: > > [1] https://github.com/Intel-Media- > SDK/MediaSDK/blob/master/doc/mediasdk-man.md#preftype > > - linjie Sorry, I missed your replay. Thanks for your advice and I will resubmit the patch. Wenbin
diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index 2bd2a56227..c3b41374e4 100644 --- a/libavcodec/qsvenc.c +++ b/libavcodec/qsvenc.c @@ -268,6 +268,13 @@ static void dump_video_param(AVCodecContext *avctx, QSVEncContext *q, case MFX_B_REF_PYRAMID: av_log(avctx, AV_LOG_VERBOSE, "pyramid"); break; default: av_log(avctx, AV_LOG_VERBOSE, "auto"); break; } + av_log(avctx, AV_LOG_VERBOSE, "; PRefType: "); + switch(co3->PRefType){ + case MFX_P_REF_DEFAULT: av_log(avctx, AV_LOG_VERBOSE, "default"); break; + case MFX_P_REF_SIMPLE: av_log(avctx, AV_LOG_VERBOSE, "simple"); break; + case MFX_P_REF_PYRAMID: av_log(avctx, AV_LOG_VERBOSE, "pyramid"); break; + default: break; + } av_log(avctx, AV_LOG_VERBOSE, "\n"); #endif @@ -777,6 +784,21 @@ FF_ENABLE_DEPRECATION_WARNINGS #if QSV_HAVE_CO3 q->extco3.Header.BufferId = MFX_EXTBUFF_CODING_OPTION3; q->extco3.Header.BufferSz = sizeof(q->extco3); + switch(q->p_strategy){ + case 0: + q->extco3.PRefType = MFX_P_REF_DEFAULT; + break; + case 1: + q->extco3.PRefType = MFX_P_REF_SIMPLE; + break; + case 2: + q->extco3.PRefType = MFX_P_REF_PYRAMID; + break; + default: + q->extco3.PRefType = MFX_P_REF_DEFAULT; + av_log(avctx, AV_LOG_VERBOSE, "invalid p_strategy, set to default\n"); + break; + } #if QSV_HAVE_GPB if (avctx->codec_id == AV_CODEC_ID_HEVC) q->extco3.GPB = q->gpb ? MFX_CODINGOPTION_ON : MFX_CODINGOPTION_OFF; diff --git a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index 6d305f87dd..30aaf72ebd 100644 --- a/libavcodec/qsvenc.h +++ b/libavcodec/qsvenc.h @@ -95,6 +95,7 @@ { "adaptive_i", "Adaptive I-frame placement", OFFSET(qsv.adaptive_i), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, \ { "adaptive_b", "Adaptive B-frame placement", OFFSET(qsv.adaptive_b), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, \ { "b_strategy", "Strategy to choose between I/P/B-frames", OFFSET(qsv.b_strategy), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, \ +{ "p_strategy", "enable P-pyramid 0-default 1-simple 2-pyramis", OFFSET(qsv.p_strategy), AV_OPT_TYPE_INT, { .i64 = 0}, 0, 2, VE }, \ { "forced_idr", "Forcing I frames as IDR frames", OFFSET(qsv.forced_idr), AV_OPT_TYPE_BOOL,{ .i64 = 0 }, 0, 1, VE }, \ { "low_power", "enable low power mode(experimental: many limitations by mfx version, BRC modes, etc.)", OFFSET(qsv.low_power), AV_OPT_TYPE_BOOL, { .i64 = 0}, 0, 1, VE},\ @@ -182,6 +183,7 @@ typedef struct QSVEncContext { int adaptive_i; int adaptive_b; int b_strategy; + int p_strategy; int cavlc; int int_ref_type;
From: Wenbinc-Bin <wenbin.chen@intel.com> Add low latency P-pyramid support for qsv, and it relates to a new command line parameter "-p_strategy". To enable this flag, user also need to set "-bf" to -1 or 0. Signed-off-by Wenbin Chen <wenbin.chen@intel.com> --- libavcodec/qsvenc.c | 22 ++++++++++++++++++++++ libavcodec/qsvenc.h | 2 ++ 2 files changed, 24 insertions(+)