[Ffmpeg-devel-irc] ffmpeg-devel.log.20181027

[burek]

[00:00:53 CEST] <BBB> oh, and his nick is sometimes inverted (taripe)
[00:00:56 CEST] <BBB> that one is online
[00:01:02 CEST] <BBB> anyway, good to see it works now
[00:01:09 CEST] <BBB> Im gonna say hi to the kids, bbl
[00:02:03 CEST] <Taripe> atomnuker hi
[00:02:45 CEST] <Taripe> sorry for the nickname confusion
[05:16:16 CEST] <cone-962> ffmpeg 03James Almer 07master:59a35fe1f63d: tests/api-h264-slice-test: use the correct function to free the AVHashContext
[05:28:00 CEST] <philipl> BtbN: https://github.com/philipl/nv-codec-headers/commit/94c72c5a74a4fae39d60964c79186bf8eb40bb57
[05:28:17 CEST] <philipl> Allowed me to build and run the filter without SDK deps (beyond nvcc, of course)
[09:50:14 CEST] <BtbN> philipl, now manage to build the kernels with clang/llvm, and you freed it.
[09:51:26 CEST] <BtbN> No idea how practical it is, but llvm has an nvptx backend
[13:38:02 CEST] <cone-384> ffmpeg 03Paul B Mahol 07master:40ac62246028: avfilter/window_func: add bohman window
[13:54:19 CEST] <durandal_1707> >> The attacker's Bitcoin address contained the equivalent of only $40, with the last transfer being received way back in April, and suggesting that the colourama package failed to make any money. <<
[14:34:20 CEST] <atomnuker> BBB: no x86util, vpbroadcastw is an invalid instruction, movd m2, r4d doesn't work unlike in lavc
[14:35:15 CEST] <BBB> movd xm2, r4d
[14:35:35 CEST] <BBB> and yes, no x86util, not bsd, sorry
[14:35:39 CEST] <BBB> Id like to use it but we can't
[14:35:48 CEST] <BBB> its not that bad
[14:35:53 CEST] <BBB> we dont care about mmx/sse2 support anyway
[14:35:56 CEST] <BBB> wihch is half of x86util
[14:36:27 CEST] <BBB> vpbroadcastw should work, I use that a lot, just not vpbroadcastw m0, r1d, thats invalid
[14:36:39 CEST] <BBB> you need movd followed by vpbroadcastw m, m
[14:36:52 CEST] <atomnuker> vpbroadcastw m2, r4w compiles but crashes
[14:37:15 CEST] <durandal_1707> you cant use r*
[14:37:34 CEST] <atomnuker> vpbroadcastw isn't mm, mm, its mm, gpr
[14:59:42 CEST] <iive> i see xmm,xmm and ymm,xmm forms
[15:01:23 CEST] <iive> are you writing avx512 ?
[15:05:12 CEST] <iive> avx512VL/BL have gpr form.
[15:08:02 CEST] <atomnuker> no
[17:37:33 CEST] <philipl> BtbN: There's a working compiler but the language feature support is uneven (Doesn't look like texture objects are supported) and when they are supported (texture references), you need the cuda SDK around to use them.
[18:37:23 CEST] <durandal_1707> i was thinking about writing filter which would count queued frames in filtergraph per each filter
[18:37:43 CEST] <durandal_1707> anybody against it?
[18:38:33 CEST] <JEEB> sounds like a general debugging tool
[18:55:50 CEST] <durandal_1707> JEEB: yes, it is supposed to debug filtergraph
[19:04:06 CEST] <durandal_1707> how should name it ? graphmonitor?
[20:41:40 CEST] <cone-511> ffmpeg 03Mark Thompson 07master:2923ed247ee2: vaapi_encode: Support configurable slices
[20:41:41 CEST] <cone-511> ffmpeg 03Mark Thompson 07master:29816e278f4f: vaapi_encode_mpeg2: Use common slice sizing code
[20:41:42 CEST] <cone-511> ffmpeg 03Mark Thompson 07master:a769e72c750e: vaapi_encode_h264: Enable multiple-slice support
[20:41:43 CEST] <cone-511> ffmpeg 03Mark Thompson 07master:a7eda762dce6: vaapi_encode_h265: Enable multiple-slice support
[20:41:44 CEST] <cone-511> ffmpeg 03Mark Thompson 07master:fef2162b6ea7: vaapi_encode: Add flag to mark encoders supporting only constant-quality
[21:11:03 CEST] <cone-511> ffmpeg 03James Almer 07master:99ef8b8afd99: avcodec/cbs_av1: fix parsing frame_size_with_refs
[21:11:04 CEST] <cone-511> ffmpeg 03James Almer 07master:a5d98da4d6f3: avcodec/cbs_vp9: fix parsing sRGB samples
[21:12:18 CEST] <pkv> @atomnuker reading page 10 of iso/iec 14496-3:2009/AMD.4:2013 , chan_config = 5 and 6 have been changed to 5.0(side) and 5.1(side) , and chan_config = 7 to 7.1(wide-side), chan_config =12 is 7.1. The interesting bit is chan_config =11 which is 6.1
[21:12:30 CEST] <pkv> http://www.doc88.com/p-7754944656492.html
[21:13:23 CEST] <pkv> there are also two chan_configs with top channels: 13 = 22.2 and 14 = 7.1 top
[22:48:39 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07master:718044dc1987: avutil/pixdesc: Add av_write_image_line2(), av_read_image_line2()
[22:48:40 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07master:cd34c6a57ee6: avfilter/vf_pixdesctest: Use 32bit read/write
[00:00:00 CEST] --- Sun Oct 28 2018