DescriptionThis is further setup for the 3on/3off optimization.
Without "optimization"
drawpointsdash_5_5_aa 8888: msecs = 24.43 GPU: msecs = 36.71
drawpointsdash_5_5_bw 8888: msecs = 11.05 GPU: msecs = 14.67
drawpointsdash_3_1_aa 8888: msecs = 15.96 GPU: msecs = 31.00
drawpointsdash_3_1_bw 8888: msecs = 12.85 GPU: msecs = 20.44
drawpointsdash_1_1_aa 8888: msecs = 38.58 GPU: msecs = 61.46
drawpointsdash_1_1_bw 8888: msecs = 34.65 GPU: msecs = 44.79
With "optimization"
drawpointsdash_5_5_aa 8888: msecs = 24.86 GPU: msecs = 36.28
drawpointsdash_5_5_bw 8888: msecs = 11.31 GPU: msecs = 14.84
drawpointsdash_3_1_aa 8888: msecs = 16.57 GPU: msecs = 31.26
drawpointsdash_3_1_bw 8888: msecs = 13.32 GPU: msecs = 20.47
drawpointsdash_1_1_aa 8888: msecs = 9.16 GPU: msecs = 161.06
drawpointsdash_1_1_bw 8888: msecs = 3.71 GPU: msecs = 48.80
With "optimization" and memory allocation changes
drawpointsdash_5_5_aa 8888: msecs = 25.49 GPU: msecs = 37.12
drawpointsdash_5_5_bw 8888: msecs = 11.73 GPU: msecs = 15.68
drawpointsdash_3_1_aa 8888: msecs = 16.36 GPU: msecs = 30.89
drawpointsdash_3_1_bw 8888: msecs = 13.87 GPU: msecs = 21.37
drawpointsdash_1_1_aa 8888: msecs = 8.70 GPU: msecs = 158.68
drawpointsdash_1_1_bw 8888: msecs = 3.56 GPU: msecs = 50.64
The 5_5 && 3_1 cases don't have a fast path yet so they remain the same - as expected. The 1_1 cases show the expected speedup for raster but a massive slow down for GPU (particularly for AA rects). The use of new[] instead of an SkTDArray for the allocation of the results provides a marginal improvement (at least on Windows) for the raster case.
Patch Set 1 #Patch Set 2 : Add minor optimizations #Patch Set 3 : Fixed debug-only error #Patch Set 4 : Minor fix for returned point count #
Total comments: 1
MessagesTotal messages: 2
|