Another test appimage, hopefully this time with onevpl/vulkan/libplacebo/zimg
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM! Try when you can and tell me what works? I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
Short test with the AppImage as-is that QSV rendering worked on Alder Lake/DG2 gen12 build plugin index for: /tmp/.mount_cin-12mlfklM/usr/bin/plugins PluginFFilter::new_ffilter(blend_vulkan) err: Input/output error MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers PluginFFilter::new_ffilter(overlay_vulkan) err: Input/output error PluginFFilter::new_ffilter(xfade_vulkan) err: Input/output error build lv2 index for: $CIN_PATH/lv2 MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers build ladspa plugin index for: /tmp/.mount_cin-12mlfklM/usr/bin/ladspa libva info: VA-API version 1.22.0 libva info: Trying to open /usr/lib64/dri-nonfree/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri-freeworld/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri/iHD_drv_video.so libva info: Found init function __vaDriverInit_1_22 libva info: va_openDriver() returns 0 libva info: VA-API version 1.22.0 libva info: Trying to open /usr/lib64/dri-nonfree/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri-freeworld/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri/iHD_drv_video.so libva info: Found init function __vaDriverInit_1_22 libva info: va_openDriver() returns 0 Render::render_single: Session finished. ** rendered 5972 frames in 24.670 secs, 242.075 fps audio0 pad 64 0 (64)
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to squashfs-root/usr/bin/ffmpeg/video But unhappily no run success with Vulkan rendering: ----- First run without export first; squashfs-root/ ./usr/bin/cin MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers [h264_vulkan @ 0x7f209cf718c0] Device does not support the VK_KHR_video_encode_queue extension! FFMPEG::open_encoder err: Function not implemented int FFMPEG::open_encoder(const char*, const char*): open failed h264_vulkan:/Videoklipp/VULKAN/hdv09-04_cin_h264_vulkan_8b420.mp4 Render::render_single: Session finished. -------------- Second run with export ANV_DEBUG="video-decode,video-encode" /squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 ..... MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers FFStream::encode_frame: vaapi encode failed. file: /Videoklipp/VULKAN/hdv09-04_cin_h264_vulkan_8b420.mp4 err: Generic error in an external library FFMPEG::mux_video err: Operation not permitted ** segv at 0x180792d in pid 84240, tid 86839 writing debug data to /tmp/cinelerra_84240.dmp lock_items: 22 lock_frees: 4 ** dump complete Segmentation fault (core dumped) ./usr/bin/cin
On 2/12/26 12:13 PM, Terje J. Hanssen wrote:
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to squashfs-root/usr/bin/ffmpeg/video
Second run with
export ANV_DEBUG="video-decode,video-encode"
/squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 .....
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620). hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8. Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
пт, 13 февр. 2026 г., 01:04 Terje J. Hanssen <[email protected]>:
On 2/12/26 12:13 PM, Terje J. Hanssen wrote:
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to squashfs-root/usr/bin/ffmpeg/video
Second run with
export ANV_DEBUG="video-decode,video-encode"
/squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 .....
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620).
hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps
This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8.
\0/ I think this is our first! Does *quality* look ok?
Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
On 2/12/26 11:18 PM, Andrew Randrianasulu wrote:
пт, 13 февр. 2026 г., 01:04 Terje J. Hanssen <[email protected]>:
On 2/12/26 12:13 PM, Terje J. Hanssen wrote: > > > On 2/12/26 1:15 AM, Andrew Randrianasulu wrote: >> https://disk.yandex.ru/d/Eqvrpylgwvc7Rw >> >> I fighted libplacebo and not sure who won - there is no Vulkan device >> in qemu VM! >> >> Try when you can and tell me what works? >> >> I did not include 16bpc crash fixes there, but without specifical >> env. variable those old new formats should not show up > > Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to > squashfs-root/usr/bin/ffmpeg/video > > > > Second run with > > export ANV_DEBUG="video-decode,video-encode" > > /squashfs-root> ./usr/bin/cin > Cinelerra Infinity - built: Feb 12 2026 02:45:04 > ..... >
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620).
hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps
This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8.
\0/
I think this is our first!
Does *quality* look ok?
There was a catch or two ...... 1. The vulkan output.mp4 was 47% larger than the hdv.m2t input file 2. Lack of red color and greenish output video So I had to do more render tests for comparison: Correspondg vaapi and qsv rendering have good colors and compression. The hdv.m2t input is interlaced but all hwaccel outputs are progressive. At last I used native ffmpeg for the same vulkan transcoding, and got the same larger size and color failure. First a summary output from the renderings: du -sh hdv09*.m2t hdv09*h264*.mp4 * 745M hdv09_04.m2t * 219M hdv09_04_cin_h264_qsv_8b420.mp4 * 470M hdv09_04_cin_h264_vaapi_8b420.mp4 * 673M hdv09_04_cin_h264_vulkan_8b420-gq22.mp4 * 1.1G hdv09_04_cin_h264_vulkan_8b420.mp4 * 1.1G hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4 ------------- So output from the Cingg renderings export ANV_DEBUG="video-decode,video-encode" ~/Application/squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 ..... MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers ---------- CIN VULKAN File> hdv09-04.m2t /home/terje/Videos/hdv09-04_cin_h264_vulkan_8b420.mp4 Compression: h264_vulkan_8b420.mp4 cin_hw_dev=vulkan cin_pix_fmt=nv12 profile=high # global_quality=22 Render::render_single: Session finished. ** rendered 5972 frames in 35.940 secs, 166.166 fps audio0 pad 64 0 (64) --------------- CIN VAAPI /home/terje/Videos/hdv09_04_cin_h264_vaapi_8b420.mp4 h264_vaapi_8b420.mp4 # Requires specific GPU hardware encoding capability # as well as required software implementation. cin_hw_dev=vaapi cin_pix_fmt=nv12 profile=high # global_quality=22 Render::render_single: Session finished. ** rendered 5972 frames in 44.521 secs, 134.139 fps audio0 pad 64 0 (64) ----------- CIN QSV /home/terje/Videos/hdv09-04_cin_h264_qsv_8b420.mp4 h264_qsv_8b420.mp4 # Requires specific GPU hardware encoding capability # as well as required software implementation. cin_pix_fmt=nv12 profile=high # global_quality=25 libva info: VA-API version 1.22.0 libva info: Trying to open /usr/lib64/dri-nonfree/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri-freeworld/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri/iHD_drv_video.so libva info: Found init function __vaDriverInit_1_22 libva info: va_openDriver() returns 0 Render::render_single: Session finished. ** rendered 5972 frames in 38.771 secs, 154.033 fps audio0 pad 64 0 (64) ======================== FFMPEG VULKAN: ffmpeg -hide_banner -init_hw_device "vulkan=vk:0" -hwaccel vulkan -hwaccel_output_format vulkan -i hdv09_04.m2t -vf "format=nv12,hwupload" -c:v h264_vulkan -y hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4 [mpeg2video @ 0x56219358fac0] Invalid frame dimensions 0x0. Last message repeated 3 times [mpegts @ 0x5621935150c0] PES packet size mismatch [mpegts @ 0x5621935150c0] Packet corrupt (stream = 1, dts = 258142320). [mpegts @ 0x5621935150c0] Could not find codec parameters for stream 2 (Unknown: none ([160][0][0][0] / 0x00A0)): unknown codec Consider increasing the value for the 'analyzeduration' (0) and 'probesize' (5000000) options [mpegts @ 0x5621935150c0] Could not find codec parameters for stream 3 (Unknown: none ([161][0][0][0] / 0x00A1)): unknown codec Consider increasing the value for the 'analyzeduration' (0) and 'probesize' (5000000) options Input #0, mpegts, from 'hdv09_04.m2t': Duration: 00:03:59.06, start: 2629.496000, bitrate: 26110 kb/s Program 100 Stream #0:0[0x810]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv, bt709, top first), 1440x1080 [SAR 4:3 DAR 16:9], 25000 kb/s, 25 fps, 25 tbr, 90k tbn, start 2629.680000 Side data: cpb: bitrate max/min/avg: 25000000/0/0 buffer size: 7340032 vbv_delay: N/A Stream #0:1[0x814]: Audio: mp2 (mp3float) ([3][0][0][0] / 0x0003), 48000 Hz, stereo, fltp, 384 kb/s, start 2629.496000 Stream #0:2[0x815]: Unknown: none ([160][0][0][0] / 0x00A0), start 2629.496000 Stream #0:3[0x811]: Unknown: none ([161][0][0][0] / 0x00A1), start 2629.496000 Stream mapping: Stream #0:0 -> #0:0 (mpeg2video (native) -> h264 (h264_vulkan)) Stream #0:1 -> #0:1 (mp2 (native) -> aac (native)) Press [q] to stop, [?] for help [h264_vulkan @ 0x56219360c300] No rate control settings specified, using fixed QP = 18 Output #0, mp4, to 'hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4': Metadata: encoder : Lavf62.3.100 Stream #0:0: Video: h264 (High) (avc1 / 0x31637661), vulkan(tv, bt709, top coded first (swapped)), 1440x1080 [SAR 4:3 DAR 16:9], q=2-31, 25 fps, 12800 tbn Metadata: encoder : Lavc62.11.100 h264_vulkan Stream #0:1: Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 128 kb/s Metadata: encoder : Lavc62.11.100 aac [mpegts @ 0x5621935150c0] PES packet size mismatch0:03:51.24 bitrate=37219.8kbits/s speed=15.9x elapsed=0:00:14.50 [mpegts @ 0x5621935150c0] Packet corrupt (stream = 1, dts = 258142320). [mpeg2video @ 0x56219360cb80] ac-tex damaged at 10 61 [mpeg2video @ 0x56219360cb80] Warning MVs not available [mpeg2video @ 0x56219360cb80] concealing 630 DC, 630 AC, 630 MV errors in P frame [vist#0:0/mpeg2video @ 0x5621935aad80] [dec:mpeg2video @ 0x56219357cf80] corrupt decoded frame [out#0/mp4 @ 0x562193579cc0] video:1077398KiB audio:3774KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.012263% frame= 5963 fps=398 q=-0.0 Lsize= 1081304KiB time=00:03:58.77 bitrate=37097.7kbits/s speed=15.9x elapsed=0:00:14.99 [aac @ 0x5621935a8040] Qavg: 488.382 ------------
Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
On Fri, Feb 13, 2026 at 6:20 PM Terje J. Hanssen <[email protected]> wrote:
On 2/12/26 11:18 PM, Andrew Randrianasulu wrote:
пт, 13 февр. 2026 г., 01:04 Terje J. Hanssen <[email protected]>:
On 2/12/26 12:13 PM, Terje J. Hanssen wrote:
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to squashfs-root/usr/bin/ffmpeg/video
Second run with
export ANV_DEBUG="video-decode,video-encode"
/squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 .....
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620).
hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps
This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8.
\0/
I think this is our first!
Does *quality* look ok?
There was a catch or two ......
1. The vulkan output.mp4 was 47% larger than the hdv.m2t input file 2. Lack of red color and greenish output video
So I had to do more render tests for comparison: Correspondg vaapi and qsv rendering have good colors and compression. The hdv.m2t input is interlaced but all hwaccel outputs are progressive. At last I used native ffmpeg for the same vulkan transcoding, and got the same larger size and color failure.
So it technically works but in fact unusable? Does progressive input fare better?
First a summary output from the renderings:
du -sh hdv09*.m2t hdv09*h264*.mp4
745M hdv09_04.m2t 219M hdv09_04_cin_h264_qsv_8b420.mp4 470M hdv09_04_cin_h264_vaapi_8b420.mp4 673M hdv09_04_cin_h264_vulkan_8b420-gq22.mp4 1.1G hdv09_04_cin_h264_vulkan_8b420.mp4 1.1G hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4
-------------
So output from the Cingg renderings
export ANV_DEBUG="video-decode,video-encode"
~/Application/squashfs-root>
./usr/bin/cin
Cinelerra Infinity - built: Feb 12 2026 02:45:04 ..... MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers
----------
CIN VULKAN
File> hdv09-04.m2t /home/terje/Videos/hdv09-04_cin_h264_vulkan_8b420.mp4
Compression: h264_vulkan_8b420.mp4 cin_hw_dev=vulkan cin_pix_fmt=nv12 profile=high # global_quality=22
Render::render_single: Session finished. ** rendered 5972 frames in 35.940 secs, 166.166 fps audio0 pad 64 0 (64)
---------------
CIN VAAPI
/home/terje/Videos/hdv09_04_cin_h264_vaapi_8b420.mp4 h264_vaapi_8b420.mp4 # Requires specific GPU hardware encoding capability # as well as required software implementation. cin_hw_dev=vaapi cin_pix_fmt=nv12 profile=high # global_quality=22
Render::render_single: Session finished. ** rendered 5972 frames in 44.521 secs, 134.139 fps audio0 pad 64 0 (64)
-----------
CIN QSV
/home/terje/Videos/hdv09-04_cin_h264_qsv_8b420.mp4 h264_qsv_8b420.mp4 # Requires specific GPU hardware encoding capability # as well as required software implementation. cin_pix_fmt=nv12 profile=high # global_quality=25
libva info: VA-API version 1.22.0 libva info: Trying to open /usr/lib64/dri-nonfree/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri-freeworld/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri/iHD_drv_video.so libva info: Found init function __vaDriverInit_1_22 libva info: va_openDriver() returns 0 Render::render_single: Session finished. ** rendered 5972 frames in 38.771 secs, 154.033 fps audio0 pad 64 0 (64)
========================
FFMPEG VULKAN:
ffmpeg -hide_banner -init_hw_device "vulkan=vk:0" -hwaccel vulkan -hwaccel_output_format vulkan -i hdv09_04.m2t -vf "format=nv12,hwupload" -c:v h264_vulkan -y hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4
[mpeg2video @ 0x56219358fac0] Invalid frame dimensions 0x0. Last message repeated 3 times [mpegts @ 0x5621935150c0] PES packet size mismatch [mpegts @ 0x5621935150c0] Packet corrupt (stream = 1, dts = 258142320). [mpegts @ 0x5621935150c0] Could not find codec parameters for stream 2 (Unknown: none ([160][0][0][0] / 0x00A0)): unknown codec Consider increasing the value for the 'analyzeduration' (0) and 'probesize' (5000000) options [mpegts @ 0x5621935150c0] Could not find codec parameters for stream 3 (Unknown: none ([161][0][0][0] / 0x00A1)): unknown codec Consider increasing the value for the 'analyzeduration' (0) and 'probesize' (5000000) options
Input #0, mpegts, from 'hdv09_04.m2t': Duration: 00:03:59.06, start: 2629.496000, bitrate: 26110 kb/s Program 100 Stream #0:0[0x810]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv, bt709, top first), 1440x1080 [SAR 4:3 DAR 16:9], 25000 kb/s, 25 fps, 25 tbr, 90k tbn, start 2629.680000 Side data: cpb: bitrate max/min/avg: 25000000/0/0 buffer size: 7340032 vbv_delay: N/A Stream #0:1[0x814]: Audio: mp2 (mp3float) ([3][0][0][0] / 0x0003), 48000 Hz, stereo, fltp, 384 kb/s, start 2629.496000 Stream #0:2[0x815]: Unknown: none ([160][0][0][0] / 0x00A0), start 2629.496000 Stream #0:3[0x811]: Unknown: none ([161][0][0][0] / 0x00A1), start 2629.496000 Stream mapping: Stream #0:0 -> #0:0 (mpeg2video (native) -> h264 (h264_vulkan)) Stream #0:1 -> #0:1 (mp2 (native) -> aac (native)) Press [q] to stop, [?] for help [h264_vulkan @ 0x56219360c300] No rate control settings specified, using fixed QP = 18
Output #0, mp4, to 'hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4': Metadata: encoder : Lavf62.3.100 Stream #0:0: Video: h264 (High) (avc1 / 0x31637661), vulkan(tv, bt709, top coded first (swapped)), 1440x1080 [SAR 4:3 DAR 16:9], q=2-31, 25 fps, 12800 tbn Metadata: encoder : Lavc62.11.100 h264_vulkan Stream #0:1: Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 128 kb/s Metadata: encoder : Lavc62.11.100 aac [mpegts @ 0x5621935150c0] PES packet size mismatch0:03:51.24 bitrate=37219.8kbits/s speed=15.9x elapsed=0:00:14.50 [mpegts @ 0x5621935150c0] Packet corrupt (stream = 1, dts = 258142320). [mpeg2video @ 0x56219360cb80] ac-tex damaged at 10 61 [mpeg2video @ 0x56219360cb80] Warning MVs not available [mpeg2video @ 0x56219360cb80] concealing 630 DC, 630 AC, 630 MV errors in P frame [vist#0:0/mpeg2video @ 0x5621935aad80] [dec:mpeg2video @ 0x56219357cf80] corrupt decoded frame [out#0/mp4 @ 0x562193579cc0] video:1077398KiB audio:3774KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.012263%
frame= 5963 fps=398 q=-0.0 Lsize= 1081304KiB time=00:03:58.77 bitrate=37097.7kbits/s speed=15.9x elapsed=0:00:14.99 [aac @ 0x5621935a8040] Qavg: 488.382
------------
Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
On 2/13/26 5:22 PM, Andrew Randrianasulu wrote:
On Fri, Feb 13, 2026 at 6:20 PM Terje J. Hanssen <[email protected]> wrote:
On 2/12/26 11:18 PM, Andrew Randrianasulu wrote:
пт, 13 февр. 2026 г., 01:04 Terje J. Hanssen <[email protected]>:
On 2/12/26 12:13 PM, Terje J. Hanssen wrote:
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to squashfs-root/usr/bin/ffmpeg/video
Second run with
export ANV_DEBUG="video-decode,video-encode"
/squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 .....
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620).
hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps
This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8.
\0/
I think this is our first!
Does *quality* look ok?
There was a catch or two ......
1. The vulkan output.mp4 was 47% larger than the hdv.m2t input file 2. Lack of red color and greenish output video
So I had to do more render tests for comparison: Correspondg vaapi and qsv rendering have good colors and compression. The hdv.m2t input is interlaced but all hwaccel outputs are progressive. At last I used native ffmpeg for the same vulkan transcoding, and got the same larger size and color failure.
So it technically works but in fact unusable?
Does progressive input fare better?
The easy part to fix is the file size. I tested with ffmpeg using -global_quality 28 and got file size 282M. Obviously vulkan encoding default prioritize is speed and high high bitrate. The loss of red or color_shift might also be a driver bug (libvulkan_intel). Vulkan. Mesa and libvulkan_intel are on v. 25.3.4 and libvulkan1 on v. 1.4.341 are often updated on Slowroll. So I'll try new tests tests now and then accordingly.
First a summary output from the renderings:
du -sh hdv09*.m2t hdv09*h264*.mp4
745M hdv09_04.m2t 219M hdv09_04_cin_h264_qsv_8b420.mp4 470M hdv09_04_cin_h264_vaapi_8b420.mp4 673M hdv09_04_cin_h264_vulkan_8b420-gq22.mp4 1.1G hdv09_04_cin_h264_vulkan_8b420.mp4 1.1G hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4
-------------
So output from the Cingg renderings
export ANV_DEBUG="video-decode,video-encode"
~/Application/squashfs-root>
./usr/bin/cin
Cinelerra Infinity - built: Feb 12 2026 02:45:04 ..... MESA-INTEL: warning: ../src/intel/vulkan/anv_formats.c:993: FINISHME: support more multi-planar formats with DRM modifiers
----------
CIN VULKAN
File> hdv09-04.m2t /home/terje/Videos/hdv09-04_cin_h264_vulkan_8b420.mp4
Compression: h264_vulkan_8b420.mp4 cin_hw_dev=vulkan cin_pix_fmt=nv12 profile=high # global_quality=22
Render::render_single: Session finished. ** rendered 5972 frames in 35.940 secs, 166.166 fps audio0 pad 64 0 (64)
---------------
CIN VAAPI
/home/terje/Videos/hdv09_04_cin_h264_vaapi_8b420.mp4 h264_vaapi_8b420.mp4 # Requires specific GPU hardware encoding capability # as well as required software implementation. cin_hw_dev=vaapi cin_pix_fmt=nv12 profile=high # global_quality=22
Render::render_single: Session finished. ** rendered 5972 frames in 44.521 secs, 134.139 fps audio0 pad 64 0 (64)
-----------
CIN QSV
/home/terje/Videos/hdv09-04_cin_h264_qsv_8b420.mp4 h264_qsv_8b420.mp4 # Requires specific GPU hardware encoding capability # as well as required software implementation. cin_pix_fmt=nv12 profile=high # global_quality=25
libva info: VA-API version 1.22.0 libva info: Trying to open /usr/lib64/dri-nonfree/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri-freeworld/iHD_drv_video.so libva info: Trying to open /usr/lib64/dri/iHD_drv_video.so libva info: Found init function __vaDriverInit_1_22 libva info: va_openDriver() returns 0 Render::render_single: Session finished. ** rendered 5972 frames in 38.771 secs, 154.033 fps audio0 pad 64 0 (64)
========================
FFMPEG VULKAN:
ffmpeg -hide_banner -init_hw_device "vulkan=vk:0" -hwaccel vulkan -hwaccel_output_format vulkan -i hdv09_04.m2t -vf "format=nv12,hwupload" -c:v h264_vulkan -y hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4
[mpeg2video @ 0x56219358fac0] Invalid frame dimensions 0x0. Last message repeated 3 times [mpegts @ 0x5621935150c0] PES packet size mismatch [mpegts @ 0x5621935150c0] Packet corrupt (stream = 1, dts = 258142320). [mpegts @ 0x5621935150c0] Could not find codec parameters for stream 2 (Unknown: none ([160][0][0][0] / 0x00A0)): unknown codec Consider increasing the value for the 'analyzeduration' (0) and 'probesize' (5000000) options [mpegts @ 0x5621935150c0] Could not find codec parameters for stream 3 (Unknown: none ([161][0][0][0] / 0x00A1)): unknown codec Consider increasing the value for the 'analyzeduration' (0) and 'probesize' (5000000) options
Input #0, mpegts, from 'hdv09_04.m2t': Duration: 00:03:59.06, start: 2629.496000, bitrate: 26110 kb/s Program 100 Stream #0:0[0x810]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv, bt709, top first), 1440x1080 [SAR 4:3 DAR 16:9], 25000 kb/s, 25 fps, 25 tbr, 90k tbn, start 2629.680000 Side data: cpb: bitrate max/min/avg: 25000000/0/0 buffer size: 7340032 vbv_delay: N/A Stream #0:1[0x814]: Audio: mp2 (mp3float) ([3][0][0][0] / 0x0003), 48000 Hz, stereo, fltp, 384 kb/s, start 2629.496000 Stream #0:2[0x815]: Unknown: none ([160][0][0][0] / 0x00A0), start 2629.496000 Stream #0:3[0x811]: Unknown: none ([161][0][0][0] / 0x00A1), start 2629.496000 Stream mapping: Stream #0:0 -> #0:0 (mpeg2video (native) -> h264 (h264_vulkan)) Stream #0:1 -> #0:1 (mp2 (native) -> aac (native)) Press [q] to stop, [?] for help [h264_vulkan @ 0x56219360c300] No rate control settings specified, using fixed QP = 18
Output #0, mp4, to 'hdv09_04_ffmpeg_h264_vulkan_nv12_vk0.mp4': Metadata: encoder : Lavf62.3.100 Stream #0:0: Video: h264 (High) (avc1 / 0x31637661), vulkan(tv, bt709, top coded first (swapped)), 1440x1080 [SAR 4:3 DAR 16:9], q=2-31, 25 fps, 12800 tbn Metadata: encoder : Lavc62.11.100 h264_vulkan Stream #0:1: Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 128 kb/s Metadata: encoder : Lavc62.11.100 aac [mpegts @ 0x5621935150c0] PES packet size mismatch0:03:51.24 bitrate=37219.8kbits/s speed=15.9x elapsed=0:00:14.50 [mpegts @ 0x5621935150c0] Packet corrupt (stream = 1, dts = 258142320). [mpeg2video @ 0x56219360cb80] ac-tex damaged at 10 61 [mpeg2video @ 0x56219360cb80] Warning MVs not available [mpeg2video @ 0x56219360cb80] concealing 630 DC, 630 AC, 630 MV errors in P frame [vist#0:0/mpeg2video @ 0x5621935aad80] [dec:mpeg2video @ 0x56219357cf80] corrupt decoded frame [out#0/mp4 @ 0x562193579cc0] video:1077398KiB audio:3774KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.012263%
frame= 5963 fps=398 q=-0.0 Lsize= 1081304KiB time=00:03:58.77 bitrate=37097.7kbits/s speed=15.9x elapsed=0:00:14.99 [aac @ 0x5621935a8040] Qavg: 488.382
------------
Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
пт, 13 февр. 2026 г., 01:04 Terje J. Hanssen <[email protected]>:
On 2/12/26 12:13 PM, Terje J. Hanssen wrote:
On 2/12/26 1:15 AM, Andrew Randrianasulu wrote:
https://disk.yandex.ru/d/Eqvrpylgwvc7Rw
I fighted libplacebo and not sure who won - there is no Vulkan device in qemu VM!
Try when you can and tell me what works?
I did not include 16bpc crash fixes there, but without specifical env. variable those old new formats should not show up
Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to squashfs-root/usr/bin/ffmpeg/video
Second run with
export ANV_DEBUG="video-decode,video-encode"
/squashfs-root> ./usr/bin/cin Cinelerra Infinity - built: Feb 12 2026 02:45:04 .....
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620).
hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps
This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8.
Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
Huh, "it does not work - let disable it". and close bug with such resolution .... It sort of sad that RICH Corporation like Intel can't arrange gpu hardware for its open source team! I guess Vulkan encoding is sort of extra for them, with focus on vaapi/qsv, but not like you have much of gpu selection with open at least at CPU side drivers on Linux with Vulkan encode .... As it turned out in this case both old and new hardware is punished, and there was no dedicated Intel GPU before Arc? Let's hope Andrea's AMD gpu from "good" generation and will produce some results…
On 12/02/2026 23:29, Andrew Randrianasulu wrote:
пт, 13 февр. 2026 г., 01:04 Terje J. Hanssen <[email protected]>:
On 2/12/26 12:13 PM, Terje J. Hanssen wrote: > > > On 2/12/26 1:15 AM, Andrew Randrianasulu wrote: >> https://disk.yandex.ru/d/Eqvrpylgwvc7Rw >> >> I fighted libplacebo and not sure who won - there is no Vulkan device >> in qemu VM! >> >> Try when you can and tell me what works? >> >> I did not include 16bpc crash fixes there, but without specifical >> env. variable those old new formats should not show up > > Extracted the AppImage and copied the h264_vulkan_8b420.mp4 preset to > squashfs-root/usr/bin/ffmpeg/video > > > > Second run with > > export ANV_DEBUG="video-decode,video-encode" > > /squashfs-root> ./usr/bin/cin > Cinelerra Infinity - built: Feb 12 2026 02:45:04 > ..... >
Good news. Vulkan rendering works now on my Dell XPX 13 (9370) ultrabook (Intel gen8 Coffee Lake w/ i7/UHD 620).
hdv.m2t -> preset h264_vulkan.mp4, rendering at ~140 fps
This is max capability of this 2017/18 hardware (not hevc_vulkan) and confirms what I got with native ffmpeg-8.
Therefore I think the reason it didn't work above on the newer gen12 Alder Lake w/DG2 (Arc A750), is due to the bug I have reported to Mesa regarding this hardware. https://gitlab.freedesktop.org/mesa/mesa/-/issues/14449#note_3238156
Huh, "it does not work - let disable it". and close bug with such resolution ....
Yeah, I found it confirmed in Phoronix https://www.phoronix.com/news/Intel-Vulkan-Video-Disable-New
It sort of sad that RICH Corporation like Intel can't arrange gpu hardware for its open source team!
I guess Vulkan encoding is sort of extra for them, with focus on vaapi/qsv, but not like you have much of gpu selection with open at least at CPU side drivers on Linux with Vulkan encode ....
As it turned out in this case both old and new hardware is punished, and there was no dedicated Intel GPU before Arc?
Let's hope Andrea's AMD gpu from "good" generation and will produce some results…
participants (2)
-
Andrew Randrianasulu -
Terje J. Hanssen