Poor QSV performance on Intel NUC DN2820FYKH

Archive of historical bug reports.
Please use the GitHub link above to report issues.
Forum rules
*******************************
Please be aware we are now using GitHub for issue tracking and feature requests.
- This section of the forum is now closed to new topics.

*******************************
Post Reply
oliverwale
Posts: 4
Joined: Tue Mar 04, 2014 9:26 pm

Poor QSV performance on Intel NUC DN2820FYKH

Post by oliverwale » Tue Mar 04, 2014 10:09 pm

Source file: http://jell.yfish.us/media/Jellyfish-20-Mbps.mkv
Preset: Normal

Using x264 codec:
Average FPS: ~8.0
Average CPU load: ~80%

Using Intel QSV codec:
Average FPS: ~3.0
Average CPU load: ~80%

Device: Intel NUC DN2820FYKH
OS: Windows 8.1
Handbrake build: svn6098 (Nightly Build)

Activity log:

Code: Select all

HandBrake svn6098 (Nightly Build) - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0
Ram: 3990 MB, Screen: 1440x900
Temp Dir: C:\Users\oliver\AppData\Local\Temp\
Install Dir: C:\Program Files\Handbrake
Data Dir: C:\Users\oliver\AppData\Roaming\HandBrake\HandBrake\0.0.0.6098

-------------------------------------------
CLI Query:  -i "C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv" -t 1 --angle 1 -c 1 -o "C:\Users\oliver\Downloads\test.mp4"  -f mp4  -w 1920 --crop 132:132:0:0 --loose-anamorphic  --modulus 2 -e qsv_h264 -q 20 --cfr -a none  --audio-fallback ac3 --markers="C:\Users\oliver\AppData\Local\Temp\test-1-chapters.csv" --qsv-preset=balanced --h264-level="4.0"  --h264-profile=main  --verbose=1

hb_qsv_info_init: mfxExtCodingOption2 check failed (0x20435641, 0x104, -17)
[21:37:37] hb_init: starting libhb thread
HandBrake svn6098 (2014030401) - MinGW x86_64 - http://handbrake.fr
2 CPUs detected
Opening C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv...
[21:37:37] CPU: Intel(R) Celeron(R) CPU  N2820  @ 2.13GHz
[21:37:37]  - Intel microarchitecture Silvermont
[21:37:37]  - logical processor count: 2
[21:37:37] Intel Quick Sync Video support: yes
[21:37:37]  - Intel Media SDK hardware: API 1.7 (minimum: 1.3)
[21:37:37]  - Intel Media SDK software: API 1.8 (minimum: 1.3)
[21:37:37]  - H.264 encoder: yes
[21:37:37]     - preferred implementation: hardware (any)
[21:37:37]  - H.265 encoder: no
[21:37:37] hb_scan: path=C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv, title_index=1
libbluray/bdnav/index_parse.c:162: indx_parse(): error opening C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv/BDMV/index.bdmv
libbluray/bdnav/index_parse.c:162: indx_parse(): error opening C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv/BDMV/BACKUP/index.bdmv
libbluray/bluray.c:2182: nav_get_title_list(C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv) failed
[21:37:37] bd: not a bd - trying as a stream/file instead
libdvdnav: Using dvdnav version 5.0.0
libdvdread: Encrypted DVD support unavailable.
libdvdread:DVDOpenFileUDF:UDFFindFile /VIDEO_TS/VIDEO_TS.IFO failed
libdvdread:DVDOpenFileUDF:UDFFindFile /VIDEO_TS/VIDEO_TS.BUP failed
libdvdread: Can't open file VIDEO_TS.IFO.
libdvdnav: vm: failed to read VIDEO_TS.IFO
[21:37:37] dvd: not a dvd - trying as a stream/file instead
Input #0, matroska,webm, from 'C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv':
  Duration: 00:00:59.97, start: 0.000000, bitrate: N/A
    Stream #0.0(eng): Video: h264 (High), yuv420p, 1920x1080 [PAR 1:1 DAR 16:9], 23.98 fps, 1k tbn, 47.95 tbc (default)
[21:37:37] scan: decoding previews for title 1
Scanning title 1 of 1, preview 1, 10.00 %
Scanning title 1 of 1, preview 2, 20.00 %
Scanning title 1 of 1, preview 3, 30.00 %
Scanning title 1 of 1, preview 5, 50.00 %
Scanning title 1 of 1, preview 6, 60.00 %
Scanning title 1 of 1, preview 7, 70.00 %
Scanning title 1 of 1, preview 8, 80.00 %
Scanning title 1 of 1, preview 9, 90.00 %
Scanning title 1 of 1, preview 10, 100.00 %[21:37:39] scan: 10 previews, 1920x1080, 23.976 fps, autocrop = 132/132/0/0, aspect 16:9, PAR 1:1
[21:37:39] libhb: scan thread found 1 valid title(s)
+ title 1:
  + stream: C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv
  + duration: 00:00:59
  + size: 1920x1080, pixel aspect: 1/1, display aspect: 1.78, 23.976 fps
  + autocrop: 132/132/0/0
  + support opencl: no
  + support hwd: yes
  + chapters:
    + 1: cells 0->0, 0 blocks, duration 00:00:59
  + audio tracks:
  + subtitle tracks:
Reading chapter markers from file C:\Users\oliver\AppData\Local\Temp\test-1-chapters.csv
[21:37:39] 1 job(s) to process
[21:37:39] starting job
[21:37:39] do_job: QSV: full path, removing unsupported filter 'Framerate Shaper'
[21:37:39] sync: expecting 1438 video frames
[21:37:39] work: only 1 chapter, disabling chapter markers
[21:37:39] job configuration:
[21:37:39]  * source
[21:37:39]    + C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv
[21:37:39]    + title 1, chapter(s) 1 to 1
[21:37:39]    + container: matroska,webm
[21:37:39]  * destination
[21:37:39]    + C:\Users\oliver\Downloads\test.mp4
[21:37:39]    + container: MPEG-4 (libavformat)
[21:37:39]  * video track
[21:37:39]    + decoder: h264_qsv
[21:37:39]    + filter
[21:37:39]      + Quick Sync Video VPP (1920:816:132:132:0:0_dei:0)
[21:37:39]        + source: 1920 * 1080, crop (132/132/0/0): 1920 * 816, scale: 1920 * 816
[21:37:39]    + loose anamorphic
[21:37:39]      + storage dimensions: 1920 * 816, mod 2
[21:37:39]      + pixel aspect ratio: 1 / 1
[21:37:39]      + display dimensions: 1920 * 816
[21:37:39]    + encoder: H.264 (Intel Media SDK)
[21:37:39]      + QSV preset: balanced
[21:37:39]      + H.264 profile: main
[21:37:39]      + H.264 level: 4.0
[21:37:39]      + quality: 20.00 (QP)
[21:37:39] reader: first SCR 0 id 0x0 DTS 0
[21:37:40] encqsvInit: using full QSV path
[21:37:40] encqsvInit: TargetUsage 2 AsyncDepth 4
[21:37:40] encqsvInit: GopRefDist 4 GopPicSize 32 NumRefFrame 3
[21:37:40] encqsvInit: BFrames on
[21:37:40] encqsvInit: RateControlMethod CQP with QPI 20 QPP 22 QPB 24
[21:37:40] encqsvInit: CAVLC off
[21:37:40] encqsvInit: H.264 profile Main @ level 4.0
[21:37:41] qsv_enc_init: using 'hardware (1)' implementation, API: 1.7
[21:37:42] encqsvWork: BPyramid off not respected (delay: 1)
[21:37:43] encqsvWork: PTS 3780 < DTS 7470 for frame 4 with type 'B' (bfrm_workaround: 1)
[21:37:44] encqsvWork: BPyramid off not respected (delay: 1)
[21:38:50] encqsvWork: BPyramid off not respected (delay: 1)
[21:38:50] encqsvWork: PTS 514260 < DTS 518040 for frame 140 with type 'B' (bfrm_workaround: 1)
[21:38:52] encqsvWork: BPyramid off not respected (delay: 1)
[21:38:53] encqsvWork: PTS 529290 < DTS 533070 for frame 144 with type 'B' (bfrm_workaround: 1)
[21:38:54] encqsvWork: BPyramid off not respected (delay: 1)
[21:38:55] encqsvWork: PTS 544320 < DTS 548010 for frame 148 with type 'B' (bfrm_workaround: 1)
[21:38:57] encqsvWork: BPyramid off not respected (delay: 1)
[21:38:57] encqsvWork: PTS 559350 < DTS 563040 for frame 152 with type 'B' (bfrm_workaround: 1)
[21:38:59] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:00] encqsvWork: PTS 574290 < DTS 578070 for frame 156 with type 'B' (bfrm_workaround: 1)
[21:39:01] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:02] encqsvWork: PTS 589320 < DTS 593100 for frame 160 with type 'B' (bfrm_workaround: 1)
[21:39:05] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:05] encqsvWork: PTS 615600 < DTS 619380 for frame 167 with type 'B' (bfrm_workaround: 1)
[21:39:07] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:07] encqsvWork: PTS 630630 < DTS 634410 for frame 171 with type 'B' (bfrm_workaround: 1)
[21:39:09] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:09] encqsvWork: PTS 645660 < DTS 649440 for frame 175 with type 'B' (bfrm_workaround: 1)
[21:39:11] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:11] encqsvWork: PTS 660690 < DTS 664380 for frame 179 with type 'B' (bfrm_workaround: 1)
[21:39:13] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:14] encqsvWork: PTS 675720 < DTS 679410 for frame 183 with type 'B' (bfrm_workaround: 1)
[21:39:15] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:16] encqsvWork: PTS 690660 < DTS 694440 for frame 187 with type 'B' (bfrm_workaround: 1)
[21:39:17] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:18] encqsvWork: PTS 705690 < DTS 709470 for frame 191 with type 'B' (bfrm_workaround: 1)
[21:39:21] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:22] encqsvWork: PTS 739530 < DTS 743220 for frame 200 with type 'B' (bfrm_workaround: 1)
[21:39:23] encqsvWork: BPyramid off not respected (delay: 1)
[21:39:24] encqsvWork: PTS 754470 < DTS 758250 for frame 204 with type 'B' (bfrm_workaround: 1)
[21:39:33] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:24] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:24] encqsvWork: PTS 1159920 < DTS 1163700 for frame 312 with type 'B' (bfrm_workaround: 1)
[21:40:26] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:27] encqsvWork: PTS 1174950 < DTS 1178640 for frame 316 with type 'B' (bfrm_workaround: 1)
[21:40:29] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:29] encqsvWork: PTS 1189980 < DTS 1193670 for frame 320 with type 'B' (bfrm_workaround: 1)
[21:40:31] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:31] encqsvWork: PTS 1204920 < DTS 1208700 for frame 324 with type 'B' (bfrm_workaround: 1)
Last edited by Rodeo on Tue Mar 04, 2014 11:20 pm, edited 1 time in total.
Reason: Logs must be placed between [code] [/code] blocks

User avatar
Rodeo
HandBrake Team
Posts: 12203
Joined: Tue Mar 03, 2009 8:55 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by Rodeo » Tue Mar 04, 2014 11:33 pm

oliverwale wrote:

Code: Select all

[21:40:31] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:31] encqsvWork: PTS 1204920 < DTS 1208700 for frame 324 with type 'B' (bfrm_workaround: 1)
Probably unrelated but looks like Silvermont isn't exactly as Ivy Bridge and does have B-pyramid capability.

Do you know which driver you're using, and are you sure it's up-to-date? Since your driver doesn't have OpenCL support, the log doesn't tell us that.

oliverwale
Posts: 4
Joined: Tue Mar 04, 2014 9:26 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by oliverwale » Wed Mar 05, 2014 9:14 am

Yes, I have the latest graphics driver installed (version 10.18.10.3408 - https://downloadcenter.intel.com/Detail ... ldID=23636).

maxim_d33
Bright Spark User
Posts: 173
Joined: Mon Mar 21, 2011 12:02 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by maxim_d33 » Thu Mar 06, 2014 6:15 pm

it might take some time to investigate but will look into it

User avatar
Rodeo
HandBrake Team
Posts: 12203
Joined: Tue Mar 03, 2009 8:55 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by Rodeo » Thu Mar 06, 2014 6:21 pm

maxim_d33 wrote:it might take some time to investigate but will look into it
Thanks :-)

kvaruni
Posts: 56
Joined: Tue Mar 24, 2009 1:12 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by kvaruni » Thu Mar 13, 2014 11:09 am

After launching the thread over here and after ordering the given PC, I can now confirm that the J1800 is not offering the normal QSV performance :) . I would not say it is as bad as the experience from overwale, but QSV is definitely slower than normal encoding. Still, ARK confirms that this particular CPU indeed offers QSV.

If you need any specific results or if I can assist in any way, just give me a shout.

kvaruni
Posts: 56
Joined: Tue Mar 24, 2009 1:12 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by kvaruni » Sun Mar 16, 2014 8:05 am

Seems like HandBrake or any other tool for QSV is not at fault, it's just Intel holding back on the drivers. It took me quite some effort but hidden in the corners of the internet is this little page: http://www.touslesdrivers.com/index.php ... code=40504. For those that don't speak French, it lists a newer driver for Bay Trail graphics (made in January, released in February) that is currently unavailable from the Intel website. Sure enough, installing these drivers suddenly unlocks QSV to its full potential. Converting Wreck-It Ralph in medium at 1080p was going at a nice 4fps, but enable QSV and it jumps up to an impressive 70fps.

So, in summary: go grab the drivers from http://fichiers3.touslesdrivers.com/405 ... 0.3408.zip (don't get them from the Intel mirror; the download is missing) and enjoy QSV.

User avatar
Rodeo
HandBrake Team
Posts: 12203
Joined: Tue Mar 03, 2009 8:55 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by Rodeo » Sun Mar 16, 2014 5:22 pm

Rodeo wrote:
oliverwale wrote:

Code: Select all

[21:40:31] encqsvWork: BPyramid off not respected (delay: 1)
[21:40:31] encqsvWork: PTS 1204920 < DTS 1208700 for frame 324 with type 'B' (bfrm_workaround: 1)
Can you please check whether you see any of these warnings in your log with a working driver?

Edit: nevermind, saw viewtopic.php?f=9&t=29724

Still, anyone seeing BPyramid-related messages in your logs on Silvermont (Bay Trail, more or less) hardware, please report it.

oliverwale
Posts: 4
Joined: Tue Mar 04, 2014 9:26 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by oliverwale » Sat Mar 22, 2014 1:52 pm

kvaruni wrote: So, in summary: go grab the drivers from http://fichiers3.touslesdrivers.com/405 ... 0.3408.zip (don't get them from the Intel mirror; the download is missing) and enjoy QSV.
These appear to be the same drivers as the ones I linked to a few posts up (https://downloadcenter.intel.com/Detail ... ldID=23636).

Unfortunately, I'm still seeing low frame rates & high CPU usage with nightly build svn6117.

Activity log:

Code: Select all

HandBrake svn6117 (Nightly Build) - 64bit Version
OS: Microsoft Windows NT 6.2.9200.0
CPU: Intel(R) Celeron(R) CPU  N2820  @ 2.13GHz
Ram: 3990 MB, 
GPU Information:
  Intel(R) HD Graphics - 10.18.10.3408
Screen: 1920x1080
Temp Dir: C:\Users\oliver\AppData\Local\Temp\
Install Dir: C:\Program Files\Handbrake
Data Dir: C:\Users\oliver\AppData\Roaming\HandBrake\HandBrake\0.0.0.6117

-------------------------------------------
CLI Query:  -i "C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv" -t 1 --angle 1 -c 1 -o "C:\Users\oliver\Downloads\test.mp4"  -f mp4  -w 1920 --crop 132:132:0:0 --loose-anamorphic  --modulus 2 -e qsv_h264 -q 20 --cfr -a none  --audio-fallback ac3 --markers="C:\Users\oliver\AppData\Local\Temp\test-1-chapters.csv" --qsv-preset=balanced --h264-level="4.0"  --h264-profile=main  --verbose=1

hb_qsv_info_init: mfxExtCodingOption2 check failed (0x20435641, 0x104, -17)
[13:36:09] hb_init: starting libhb thread
HandBrake svn6117 (2014031701) - MinGW x86_64 - http://handbrake.fr
2 CPUs detected
Opening C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv...
[13:36:09] CPU: Intel(R) Celeron(R) CPU  N2820  @ 2.13GHz
[13:36:09]  - Intel microarchitecture Silvermont
[13:36:09]  - logical processor count: 2
[13:36:09] OpenCL device #1: Intel(R) Corporation Intel(R) HD Graphics
[13:36:09]  - OpenCL version: 1.2 
[13:36:09]  - driver version: 10.18.10.3408
[13:36:09]  - device type:    GPU
[13:36:09]  - supported:      YES
[13:36:09] Intel Quick Sync Video support: yes
[13:36:09]  - Intel Media SDK hardware: API 1.7 (minimum: 1.3)
[13:36:09]  - Intel Media SDK software: API 1.8 (minimum: 1.3)
[13:36:09]  - H.264 encoder: yes
[13:36:09]     - preferred implementation: hardware (any)
[13:36:09]  - H.265 encoder: no
[13:36:09] hb_scan: path=C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv, title_index=1
libbluray/bdnav/index_parse.c:162: indx_parse(): error opening C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv/BDMV/index.bdmv
libbluray/bdnav/index_parse.c:162: indx_parse(): error opening C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv/BDMV/BACKUP/index.bdmv
libbluray/bluray.c:2182: nav_get_title_list(C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv) failed
[13:36:09] bd: not a bd - trying as a stream/file instead
libdvdnav: Using dvdnav version 5.0.0
libdvdread: Encrypted DVD support unavailable.
libdvdread:DVDOpenFileUDF:UDFFindFile /VIDEO_TS/VIDEO_TS.IFO failed
libdvdread:DVDOpenFileUDF:UDFFindFile /VIDEO_TS/VIDEO_TS.BUP failed
libdvdread: Can't open file VIDEO_TS.IFO.
libdvdnav: vm: failed to read VIDEO_TS.IFO
[13:36:09] dvd: not a dvd - trying as a stream/file instead
Input #0, matroska,webm, from 'C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv':
  Duration: 00:00:59.97, start: 0.000000, bitrate: N/A
    Stream #0.0(eng): Video: h264 (High), yuv420p, 1920x1080 [PAR 1:1 DAR 16:9], 23.98 fps, 1k tbn, 47.95 tbc (default)
[13:36:09] scan: decoding previews for title 1
Scanning title 1 of 1, preview 2, 20.00 %
Scanning title 1 of 1, preview 3, 30.00 %
Scanning title 1 of 1, preview 4, 40.00 %
Scanning title 1 of 1, preview 6, 60.00 %
Scanning title 1 of 1, preview 7, 70.00 %
Scanning title 1 of 1, preview 9, 90.00 %
Scanning title 1 of 1, preview 10, 100.00 %[13:36:11] scan: 10 previews, 1920x1080, 23.976 fps, autocrop = 132/132/0/0, aspect 16:9, PAR 1:1
[13:36:11] libhb: scan thread found 1 valid title(s)
+ title 1:
  + stream: C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv
  + duration: 00:00:59
  + size: 1920x1080, pixel aspect: 1/1, display aspect: 1.78, 23.976 fps
  + autocrop: 132/132/0/0
  + support opencl: yes
  + support hwd: yes
  + chapters:
    + 1: cells 0->0, 0 blocks, duration 00:00:59
  + audio tracks:
  + subtitle tracks:
Reading chapter markers from file C:\Users\oliver\AppData\Local\Temp\test-1-chapters.csv
[13:36:11] 1 job(s) to process
[13:36:11] starting job
[13:36:11] do_job: QSV: full path, removing unsupported filter 'Framerate Shaper'
[13:36:11] sync: expecting 1438 video frames
[13:36:11] work: only 1 chapter, disabling chapter markers
[13:36:11] job configuration:
[13:36:11]  * source
[13:36:11]    + C:\Users\oliver\Downloads\Jellyfish-20-Mbps.mkv
[13:36:11]    + title 1, chapter(s) 1 to 1
[13:36:11]    + container: matroska,webm
[13:36:11]  * destination
[13:36:11]    + C:\Users\oliver\Downloads\test.mp4
[13:36:11]    + container: MPEG-4 (libavformat)
[13:36:11]  * video track
[13:36:11]    + decoder: h264_qsv
[13:36:11]    + filter
[13:36:11]      + Quick Sync Video VPP (1920:816:132:132:0:0_dei:0)
[13:36:11]        + source: 1920 * 1080, crop (132/132/0/0): 1920 * 816, scale: 1920 * 816
[13:36:11]    + loose anamorphic
[13:36:11]      + storage dimensions: 1920 * 816, mod 2
[13:36:11]      + pixel aspect ratio: 1 / 1
[13:36:11]      + display dimensions: 1920 * 816
[13:36:11]    + encoder: H.264 (Intel Media SDK)
[13:36:11]      + QSV preset: balanced
[13:36:11]      + H.264 profile: main
[13:36:11]      + H.264 level: 4.0
[13:36:11]      + quality: 20.00 (QP)
[13:36:11] reader: first SCR 0 id 0x0 DTS 0
[13:36:11] encqsvInit: using full QSV path
[13:36:11] encqsvInit: TargetUsage 2 AsyncDepth 4
[13:36:11] encqsvInit: GopRefDist 4 GopPicSize 32 NumRefFrame 3
[13:36:11] encqsvInit: BFrames on
[13:36:11] encqsvInit: RateControlMethod CQP with QPI 20 QPP 22 QPB 24
[13:36:11] encqsvInit: CAVLC off
[13:36:11] encqsvInit: H.264 profile Main @ level 4.0
[13:36:12] qsv_enc_init: using 'hardware (1)' implementation, API: 1.7
[13:36:13] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:13] encqsvWork: PTS 3780 < DTS 7470 for frame 4 with type 'B' (bfrm_workaround: 1)
[13:36:13] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:53] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:53] encqsvWork: PTS 514260 < DTS 518040 for frame 140 with type 'B' (bfrm_workaround: 1)
[13:36:54] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:55] encqsvWork: PTS 529290 < DTS 533070 for frame 144 with type 'B' (bfrm_workaround: 1)
[13:36:56] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:56] encqsvWork: PTS 544320 < DTS 548010 for frame 148 with type 'B' (bfrm_workaround: 1)
[13:36:57] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:57] encqsvWork: PTS 559350 < DTS 563040 for frame 152 with type 'B' (bfrm_workaround: 1)
[13:36:58] encqsvWork: BPyramid off not respected (delay: 1)
[13:36:59] encqsvWork: PTS 574290 < DTS 578070 for frame 156 with type 'B' (bfrm_workaround: 1)
[13:37:00] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:00] encqsvWork: PTS 589320 < DTS 593100 for frame 160 with type 'B' (bfrm_workaround: 1)
[13:37:02] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:02] encqsvWork: PTS 615600 < DTS 619380 for frame 167 with type 'B' (bfrm_workaround: 1)
[13:37:03] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:03] encqsvWork: PTS 630630 < DTS 634410 for frame 171 with type 'B' (bfrm_workaround: 1)
[13:37:04] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:05] encqsvWork: PTS 645660 < DTS 649440 for frame 175 with type 'B' (bfrm_workaround: 1)
[13:37:06] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:06] encqsvWork: PTS 660690 < DTS 664380 for frame 179 with type 'B' (bfrm_workaround: 1)
[13:37:07] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:07] encqsvWork: PTS 675720 < DTS 679410 for frame 183 with type 'B' (bfrm_workaround: 1)
[13:37:08] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:08] encqsvWork: PTS 690660 < DTS 694440 for frame 187 with type 'B' (bfrm_workaround: 1)
[13:37:09] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:10] encqsvWork: PTS 705690 < DTS 709470 for frame 191 with type 'B' (bfrm_workaround: 1)
[13:37:12] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:12] encqsvWork: PTS 739530 < DTS 743220 for frame 200 with type 'B' (bfrm_workaround: 1)
[13:37:13] encqsvWork: BPyramid off not respected (delay: 1)
[13:37:14] encqsvWork: PTS 754470 < DTS 758250 for frame 204 with type 'B' (bfrm_workaround: 1)
[13:37:20] encqsvWork: BPyramid off not respected (delay: 1)
78uzxas

User avatar
Rodeo
HandBrake Team
Posts: 12203
Joined: Tue Mar 03, 2009 8:55 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by Rodeo » Sat Mar 22, 2014 5:01 pm

Hmm, I have a guess as to what's going on.

ARK | Intel® Celeron® Processor J1800
ARK | Intel® Celeron® Processor N2820

Unlike the J1800 used by kvaruni, the Intel ARK website indicates that the N2820 processor does not support Quick Sync Video. However, since the encoder was detected as available, my original assumption was that either of the following applied:

- the Intel ARK website was incorrect for some reason, and the processor does in fact support QSV
- the graphics driver was buggy and used a code path meant for another processor/GPU combo

Now, looking at your first post (80% CPU usage and B-pyramid used even though the implementation supposedly doesn't support it), I have another theory: Intel Quick Sync Video is indeed NOT supported on your processor, and the graphics driver is using the hardware implementation…'s fallback software implementation.

If that theory is correct, you'll need a new processor is you want hardware-accelerated QSV encoding.

oliverwale
Posts: 4
Joined: Tue Mar 04, 2014 9:26 pm

Re: Poor QSV performance on Intel NUC DN2820FYKH

Post by oliverwale » Sat Mar 22, 2014 5:25 pm

Thanks Rodeo. The results of running 'C:\Program Files\Intel\Media SDK 2014 for Clients\tools\mediasdk_sys_analyzer\mediasdk_system_analyzer_64.exe' seem to support your theory:
Image

Post Reply