В сообщении от Wednesday 04 September 2019 16:25:04 Phyllis Smith написал(а):
Olivier:
So I did. I backported nasm 2.14.02 and tried to build cin from there. It's working for the 64 bits build, but not for the 32 bits build. The error is :
... zip ...
[ 98%] Building CXX object CMakeFiles/test_aom_encoder.dir/test/reconinter_test.cc.o mv -f .deps/ImfDeepImageStateAttribute.Tpo .=CXX --mode=compile g++ -DHAVE_CONFIG_H -I. -I../config -I/build/cin-5.1+20190831/thirdparty/../thirdparty/ilmbase-2.2.1/usr/include/OpenEXR
-
... zip ...
If that rings a bell, please let me know.
For your information, I've disabled the dav1d option for the 32 bits build for now.
Yes, correct permanent solution. Some of the assembler instructions that libdav1d uses are not available for 32-bit. GG says he will put in a check for that (someday!).
Strange, I tought I had dav1d 0.4.0 build as part of Cinelerra-GG, but apparently build was picking up older 0.3.1 I installed via slackbuild! Recompiling libdav1d 0.4.0 on 32-bit slackware still worked, at least Mplayer + my patch played webm stream: mplayer /dev/shm/Stream3_AV3_HD_6.3mbps.webm MPlayer SVN-r38151-5.5.0 (C) 2000-2019 MPlayer Team 224 audio & 465 video codecs Playing /dev/shm/Stream3_AV3_HD_6.3mbps.webm. libavformat version 58.32.104 (internal) libavformat file format detected. [libdav1d @ 0x57696b60]libdav1d 0.4.0 [libdav1d @ 0x57696b60]Frame size limit reduced from 2147483647 to 67108864. [lavf] stream 0: video (libdav1d), -vid 0 VIDEO: [AV01] 1920x1080 0bpp 25.000 fps 0.0 kbps ( 0.0 kbyte/s) ========================================================================== Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family libavcodec version 58.56.101 (internal) [libdav1d @ 0x57696b60]libdav1d 0.4.0 [libdav1d @ 0x57696b60]Frame size limit reduced from 2147483647 to 67108864. Selected video codec: [av1] vfm: ffmpeg (AV1) ========================================================================== Clip info: encoder: libwebm-0.2.1.0 Load subtitles in /dev/shm/ Audio: no sound Starting playback... Movie-Aspect is 1.78:1 - prescaling to correct movie aspect. VO: [vdpau] 1920x1080 => 1920x1080 Planar YV12 V: 0.2 0/ 0 ??% ??% ??,?% 0 0 [libdav1d @ 0x57696b60]The deprecated avcodec_decode_* API cannot return all the frames for this decoder. Some frames will be dropped. Update your code to the new decoding API to fix this. V: 16.2 0/ 0 32% 21% 0.0% 0 0 So libdav1d as compiled by its own meson-based build system works on at least some 32-bit x86 CPUs. I use Meson because mesa git (19.3.0) requires it. Actually libdav1d's Changelog for 0.4.0 says: Changes for 0.4.0 'Cheetah': ---------------------------- - Fix playback with unknown OBUs - Add an option to limit the maximum frame size - SSE2 and ARM64 optimizations for MSAC - Improve speed on 32bits systems <<<<<<<<<<<<<<<<- this - Optimization in obmc blend - Reduce RAM usage significantly - The initial PPC SIMD code, cdef_filter - NEON optimizations for blend functions on ARM - NEON optimizations for w_mask functions on ARM - NEON optimizations for inverse transforms on ARM64 - Improve handling of malloc failures - Simple Player example in tools https://code.videolan.org/videolan/dav1d/blob/master/NEWS Just in case you want mplayer patch I attach it (added to MPlayer's own tarcker too, but can't recall exact issue #) What IS not working on 32-bit - https://github.com/OpenVisualCloud/SVT-AV1 I only tried 0.5.0 on 64-bit Slackware.... They have 0.6.0 already: [0.6.0] - 2019-06-28 Inital decoder implementation Static library support Alt-ref pictures - temporal filtering Adaptive Transform Block for INTRA Adaptive QP scaling Decoder - Support for Tiles and 10bit API - add option to calculate / report PSNR values Support for segmentation SIMD Optimizations Downsampling 2x2 filtering Handle incomplete SBs MACROS / trailing / tabs-spaces cleanup http://ffmpeg.org/pipermail/ffmpeg-devel/2019-July/246488.html [FFmpeg-devel] [PATCH] lavc: Add libsvt-av1 encoder wrapper It was not yet merged, as far as I can see .... THIS one mostly likely will be 64-bit ONLY, because Intel not very interested in 32-bit mode even on new(ish) CPUs, it seems .... ? But who knows, may be after 1.0.0 someone will fix it up .....