Should I push the ones you mark as r-b or wait for a separate go-ahead based on CI results? <div class="gmail_extra"> <div class="gmail_quote">On Nov 14, 2016 8:45 PM, "Cherniak, Bruce" <<a href="mailto:bruce.cherniak@intel.com">bruce.cherniak@intel.com</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Reviewing/commenting on each patch individually. We’re also testing more thoroughly in our CI. In general, your help and interest in OpenSWR is very much appreciated! > On Nov 12, 2016, at 5:00 PM, Ilia Mirkin <<a href="mailto:imirkin@alum.mit.edu">imirkin@alum.mit.edu</a>> wrote: > > This is a bit of a hodge-podge, but largely related to improving texture > resource manipulation - mapping, texturing, rendering, etc. There's a WIP > commit at the end which was required to make swr not crash on piglit exit > half the time, but I think it needs to be redone in a proper way. > > The meat is in the "swr: rework resource layout and surface setup" commit. > This redoes surface layout for swr to match with the rasterizer's > expectations. This way we can tell it which array index or LOD to render to. > This will also be important for layered rendering when GS support comes > along. It also paves the way to supporting the swr backend's tiling to > improve cache performance. > > Unfortunately it does mean more overhead for all textures, not just the > renderable ones. I thought this was acceptable, as supporting multiple > different layout mechanisms seems a little confusing. > > This series overall fixes ~600 piglit tests. You can see this series (+ one > more commit that's not really ready) at > > <a href="http://github.com/imirkin/mesa/commits/swr" rel="noreferrer" target="_blank">http://github.com/imirkin/mesa/commits/swr</a> > > I highly recommend testing this in your higher-core/higher-perf environments > than my desktop SKL (or, *gasp*, SDE on a Core i7-920). > > Ilia Mirkin (14): > swr: [rasterizer memory] add support for R32_FLOAT_X8X24 formats > swr: [rasterizer memory] round up when dividing by block sizes > swr: [rasterizer memory] fix store tile for 128-bit ymajor tiling > swr: [rasterizer jitter] don't bother quantizing unused channels > swr: add archrast generated files to gitignore > swr: fix texture layout for compressed formats > swr: mark both frag and vert textures read, don't forget about cbs > swr: no support for shader stencil export > swr: mark rgb9_e5 as unrenderable > swr: disable blending for integer formats > swr: rework resource layout and surface setup > swr: remove unnecessary -1 entries in format mapping table > swr: remove formats from mapping table that don't have StoreTile impls > WIP swr: make sure that all rendering is finished on shader destroy > > src/gallium/drivers/swr/.gitignore | 4 + > .../drivers/swr/rasterizer/jitter/blend_jit.cpp | 2 +- > .../drivers/swr/rasterizer/memory/LoadTile.h | 1 + > .../drivers/swr/rasterizer/memory/StoreTile.h | 3 +- > .../swr/rasterizer/memory/TilingFunctions.h | 15 +- > src/gallium/drivers/swr/swr_context.cpp | 108 ++++-- > src/gallium/drivers/swr/swr_draw.cpp | 4 +- > src/gallium/drivers/swr/swr_resource.h | 8 +- > src/gallium/drivers/swr/swr_screen.cpp | 397 +++++++++------------ > src/gallium/drivers/swr/swr_shader.cpp | 28 +- > src/gallium/drivers/swr/swr_state.cpp | 197 ++++++---- > 11 files changed, 433 insertions(+), 334 deletions(-) > > -- > 2.7.3 > > _______________________________________________ > mesa-dev mailing list > <a href="mailto:mesa-dev@lists.freedesktop.org">mesa-dev@lists.freedesktop.org</a> > <a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev" rel="noreferrer" target="_blank">https://lists.freedesktop.org/mailman/listinfo/mesa-dev</a> </blockquote></div></div>