https://bugs.documentfoundation.org/show_bug.cgi?id=166507
V Stuart Foote <vsfo...@libreoffice.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |vsfo...@libreoffice.org
See Also| |https://bugs.documentfounda
| |tion.org/show_bug.cgi?id=16
| |6122
--- Comment #3 from V Stuart Foote <vsfo...@libreoffice.org> ---
(In reply to launchpad from comment #2)
> LibreOffice urgently needs reliable detection to determine whether hardware
> acceleration is actually working. If it's not, LibreOffice should
> automatically disable it—without forcing end users to troubleshoot or
> navigate unintuitive settings.
>
Actually already had very effective testing around support of Vulkan (and
before that for OpenGL), now being done with Google chromium skia libs. With
fall back to skia based raster framing. Or manual de-selection of the skia
libs, for fallback to 'default' GDI+ based rendering libs--CPU only or with CPU
controlled HA.
Those skia lib releases, external to project, were updated from m116 to m130
for the 25.2.0 builds. And some as yet undetermined fact of the testing
framework has slipped. It is a routine issue, and once fully identified would
likely make it into the 25.2 branch in an incremental patch release.
> Until that detection is fully reliable, hardware acceleration should be
> disabled by default. Please make this change in the very next release.
Well that won't happen. More to the point to identify what's changed in the
skia skwindow call when Vulkan test for a non-denylisted device is run and
fails.
Also, the denylist for Virtual devices should be more robust, specifically
VMWare VMSVGA or MS Hypervisor VM Bus Video Device should be deny listed.
And there is the possibility the skia based software rendering is also degraded
on marginally supported GPUs and vendor drivers. Too soon to be certain.
--
You are receiving this mail because:
You are the assignee for the bug.
