【Auto-tuning】The meaning of generated 'rk3399.resnet-18.log'


#1

Who can help me explain the following .log file:

{“r”: [[0.0006094604], 0, 10.233074188232422, 1555837488.239555], “v”: 0.1, “i”: [“llvm -device=arm_cpu -target=aarch64-linux-gnu”, “topi_nn_conv2d”, [[“TENSOR”, [1, 256, 14, 14], “float32”], [“TENSOR”, [512, 256, 1, 1], “float32”], [2, 2], [0, 0], [1, 1], “NCHW”, “float32”], {}, [“conv2d”, [1, 256, 14, 14, “float32”], [512, 256, 1, 1, “float32”], [2, 2], [0, 0], [1, 1], “NCHW”, “float32”], {“c”: null, “e”: [[“tile_co”, “sp”, [64, 8]], [“tile_oh”, “sp”, [7, 1]], [“tile_ow”, “sp”, [1, 7]], [“reorder_0”, “re”, [0, 1, 2, 3, 4, 5, 6, 9, 7, 8]], [“ann_reduce”, “an”, [“unroll”, “none”]], [“ann_spatial”, “an”, [“none”, “none”, “vec”]]], “t”: “direct”, “i”: 863}]}


#2

These are the parameters (e.g., tile_oh, reorder_0) tuned for a particular workload (in this case conv2d with shape 1, 256, 14, 14, …). You can inspect the corresponding schedule template in topi https://github.com/dmlc/tvm/blob/master/topi/python/topi/arm_cpu/conv2d.py to see what the parameters meme.