MSS R6 test result

[Marvin_C_]

Hi Sir:

We use Artesyn PCIE-7207 card with MSS R6 to do the transcoding, OS is Centos7.

The attached file is the test result, we only use one cpu to do the transcoding.

As you can see, if transcode 1.h264 1080p file to 720p, test result is not linear.

Ex: Transcode 1 1080p to 720p channels takes 44.16 sec, but if transcode 2 channels, only takes 63.5 sec,transcode 5 channels takes 143.66sec.

It seems if transcode more channels, average transcoding time will decrease.

Would you mind check the attached file and let me know it is reasonable or not.

Thanks.

Marvin

[Leo_Z_]

Hi Marvin,

I'm using Artesyn PCIE-7207 card for transcoding too, may i know the code sample you used for testing,  and the video source info such as bitrate/framerate, trandcode fps? 

[Marvin_C_]

Hi Leo:

I'm using R6 PXE bootable image with MSS R6 sample code.

The video info is as below lists.

Thanks.

Marvin

Format	MPEG-4
Format_Profile	Base Media
CodecID	isom
FileSize/String	109 MiB
Duration/String	3mn 29s
OverallBitRate/String	4 344 Kbps
Encoded_Application	Lavf55.37.101
Video
ID/String	1
Format	AVC
Format/Info	Advanced Video Codec
Format_Profile	High@L4.0
Format_Settings_CABAC/String	Yes
Format_Settings_RefFrames/String	3 frames
CodecID	avc1
CodecID/Info	Advanced Video Coding
Duration/String	3mn 29s
BitRate/String	4 083 Kbps
Width/String	1 920 pixels
Height/String	1 080 pixels
DisplayAspectRatio/String	16:09
FrameRate_Mode/String	Constant
FrameRate/String	29.970 fps
ColorSpace	YUV
ChromaSubsampling	4:02:00
BitDepth/String	8 bits
ScanType/String	Progressive
Bits-(Pixel*Frame)	0.066
StreamSize/String	102 MiB (94%)

[Marvin_C_]

Hi All:

Any idea about this issue, would you mind check it and advise it?

Thanks.

Marvin

[Jeffrey_M_Intel1]

The results you've recorded seem very reasonable.  Unlike CPU-only limitations we don't expect linear behavior.  The 5th generation processors in the Artesyn card contain multiple hardware units which can execute simultaneously.  It becomes easier to schedule work efficiently as more streams are added.  For more information on the architecture you can read 

https://01.org/linuxgraphics/documentation/hardware-specification-prms/2014-2015-intel-processors-based-broadwell-platform

[Marvin_C_]

Hi Jeffery:

Noted, thanks for your support.

Marvin