https://bugs.documentfoundation.org/show_bug.cgi?id=161879
--- Comment #14 from V Stuart Foote <[email protected]> --- (In reply to ady from comment #13) > If I may... > > I would still get the impression that there is some issue, somewhere. > > Even if there is nothing that LO's devs could do (on LO's code side), there > seems to be something to improve, maybe about Skia, or maybe about OpenCL > (or maybe about OpenGL, or...). > > IDK whether LO devs report such potential issues upstream, at least with the > intention of potential future improvements in performance for LO/Calc. To what end? We have adopted a denylist approach for known bad/suboptimal GPU/driver parings to prevent use of skia/Vulkan rendering--thereby forcing those systems to use skia based raster graphics framing. While for macOS we don't provide Vulkan vector graphics as Apple does not support it (using a skia/Metal implementation instead but with no denylist mechanism). We're treading water, and the GPU vendors don't help much by not releasing debug symbols (e.g. as PDB) for their driver source, meaning we can't really pursue anything "upstream". We code to published API, but specifics of Vulkan, or "legacy" OpenGL rendering (which has been stripped out of the LO code base) accelerated graphics is a quagmire with no active dev interest. -- You are receiving this mail because: You are the assignee for the bug.
