Commit 660e4a7c0d4f919b58a7ef22903aad2efabefa1d
2023-01-20T15:42:29
Don't override D16_UNORM to D32_FLOAT on Apple GPUs. Local testing of: dEQP-GLES3.functional.polygon_offset.fixed16_render_with_units on M1 indicates that depth bias works correctly with MTLPixelFormatDepth16Unorm on this GPU type, indicating that previous workarounds only need to be applied to non-Apple GPUs. Reimplement d24s8_fallbacks_mac as depth_fallbacks_mac, add this workaround there, and move the conditions to the JSON file per review feedback from Alexey Knyazev. Avoid emitting duplicate depth formats in gen_mtl_format_table.py. Leave the iOS Simulator alone - the workaround is likely still needed there, and difficult to test. Tested via: ./out/Release/angle_deqp_gles3_tests \ --gtest_filter=\ dEQP-GLES3.functional.polygon_offset.fixed16_render_with_units \ --use-angle=metal Bug: angleproject:6597 Change-Id: I52a99b3f710bb88e1f43ed17862f6e6f2e6471ac Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4184943 Auto-Submit: Kenneth Russell <kbr@chromium.org> Reviewed-by: Kyle Piddington <kpiddington@apple.com> Reviewed-by: Quyen Le <lehoangquyen@chromium.org> Reviewed-by: Gregg Tavares <gman@chromium.org> Commit-Queue: Kenneth Russell <kbr@chromium.org>