Blender Git Loki
Git Commits -> Revision da81747
Revision da81747 by Sergey Sharybin (cycles-x) March 3, 2021, 11:15 (GMT) |
Cycles: Initial implementation of tiled rendering on CUDA Schedules more than one pixel to CUDA device. The tile size calculation is implemented as a dedicated testable function, which should simplify tweaking the "division" algorithm in the future. The actual splitting logic is commented out and tiles are set to 256 by 256. There seems to be some memory allocation or access issue which makes proper tile size to fail. |
Commit Details:
Full Hash: da817472095264e0396b259a4e0f49fd82489ba0
Parent Commit: 3e4a448
Lines Changed: +217, -24
3 Added Paths:
/intern/cycles/integrator/tile.cpp (+57, -0) (View)
/intern/cycles/integrator/tile.h (+31, -0) (View)
/intern/cycles/test/integrator_tile_test.cpp (+47, -0) (View)
/intern/cycles/integrator/tile.h (+31, -0) (View)
/intern/cycles/test/integrator_tile_test.cpp (+47, -0) (View)
7 Modified Paths:
/intern/cycles/device/cuda/queue.cpp (+1, -2) (Diff)
/intern/cycles/integrator/CMakeLists.txt (+2, -0) (Diff)
/intern/cycles/integrator/path_trace.cpp (+3, -0) (Diff)
/intern/cycles/integrator/path_trace.h (+5, -1) (Diff)
/intern/cycles/integrator/work_scheduler.cpp (+35, -16) (Diff)
/intern/cycles/integrator/work_scheduler.h (+34, -5) (Diff)
/intern/cycles/test/CMakeLists.txt (+2, -0) (Diff)
/intern/cycles/integrator/CMakeLists.txt (+2, -0) (Diff)
/intern/cycles/integrator/path_trace.cpp (+3, -0) (Diff)
/intern/cycles/integrator/path_trace.h (+5, -1) (Diff)
/intern/cycles/integrator/work_scheduler.cpp (+35, -16) (Diff)
/intern/cycles/integrator/work_scheduler.h (+34, -5) (Diff)
/intern/cycles/test/CMakeLists.txt (+2, -0) (Diff)