Den 03.12.2024 01:16, skrev Andrew Randrianasulu:
вт, 3 дек. 2024 г., 02:42 Phyllis Smith <[email protected]>:
Well, yes, BUT the point is people have procedures set up to build this automatically for years and now all of a sudden they will have to install more software. Probably not a good idea.
Note, we do not talk about changing default globally, just about machinery behind
Possibly a test-build will clarify(?)
but at 3am (GMT+3) I think we surely can stop holding our breath for same-minute answer ;)
Good night, sorry I did not looked at labels issue yet
On Mon, Dec 2, 2024 at 4:40 PM Andrew Randrianasulu <[email protected]> wrote:
вт, 3 дек. 2024 г., 02:36 Phyllis Smith <[email protected]>:
Unfortunately, it won't even build when I add: --with-onevpl on the configure line:
configure: error: requires onevpl support. make: *** No targets specified and no makefile found. Stop.
This is my 6 year old AMD laptop.
you need to install
https://github.com/intel/libvpl
I used 2.13 for my dummy tests, but something down to 2.6 (I think) should work, at least as build requirement.
On Mon, Dec 2, 2024 at 4:01 PM Terje J. Hanssen <[email protected]> wrote:
Den 02.12.2024 23:32, skrev Phyllis Smith:
Adding OneVPL by default, which only works on the latest Intel computers and not on older ones or AMD, would not be a good idea; especially not for the RPM packaging. That is my opinion based on our attempt to keep CinGG generically usable on older computers and even 32-bit systems.
I didn't think "onevpl on" would prohibit other than Intel hardware to work as else, and that hwaccell work or not, dependent of the platform support. Actually my thought was to test what legacy Intel hardware are good for. Is possibly a Cingg user-switch possible instead?
I have yet to go through the last 5 attachment patches that Andrew graciously put all in one place for viability on all systems.
On Mon, Dec 2, 2024 at 2:01 PM Terje J. Hanssen <[email protected]> wrote:
Den 30.11.2024 01:03, skrev Andrew Randrianasulu:
сб, 30 нояб. 2024 г., 01:07 Phyllis Smith via Cin <[email protected]>:
Andrew, should these and the QSV presets be checked into GIT now that Terje has tested them on his specific hardware?
yes, but 10bit ones require patch for themselves and recent one removing addition of ildct flag for some qsv encoders ....
"Exp: do not set ildct flag for qsv_av1/h264/vp9" 0001-yet-another-10bit-attempt.patch 0002-Hopefully-wire-up-various-cin_pix_fmt-for-vaapi-enco.patch
re-attached so you do not need to hunt them in mailthreads ....
Is the onevpl patch already included in the buildsystem? It was said it was "default to off", so I wonder if it will be available or "on" in the upcoming rpm?
I also have mediacodec/termux related profiles, and one hackfix for mjpegtools failing to build here on clang.
P.S. There will not be new Appimages created at the end of this month, but will wait until December.
On Tue, Nov 26, 2024 at 3:52 PM Terje J. Hanssen <[email protected]> wrote:
Attached tested vaapi hwaccel presets as gzip'd tar file
av1_vaapi_10b420.mp4 av1_vaapi_10b420.webm av1_vaapi_8b420.mp4 av1_vaapi_8b420.webm h264_vaapi_8b420.mp4 hevc_vaapi_10b420.mp4 hevc_vaapi_10b422.mp4 hevc_vaapi_8b420.mp4 vp9_vaapi_10b420.mp4 vp9_vaapi_10b420.webm vp9_vaapi_8b420.mp4 vp9_vaapi_8b420.webm
-- Cin mailing list [email protected] https://lists.cinelerra-gg.org/mailman/listinfo/cin