错误堆栈： File “/code/ailice_14/tvm/release/python/tvm/topi/cuda/conv2d.py”, line 46, in schedule_conv2d_nchw traverse_inline(s, outs[0].op, _callback) File “/code/ailice_14/tvm/release/python/tvm/topi/utils.py”, line 81, in traverse_inline _traverse(final_op) File “/code/ailice_14/tvm/release/python/tvm/topi/utils.py”, line 78, in _traverse _traverse(tensor.op) File “/code/ailice_14/tvm/release/python/tvm/topi/utils.py”, line 78, in _traverse _traverse(tensor.op) File “/code/ailice_14/tvm/release/python/tvm/topi/utils.py”, line 79, in _traverse callback(op) File “/code/ailice_14/tvm/release/python/tvm/topi/cuda/conv2d.py”, line 44, in _callback schedule_direct_cuda(cfg, s, op.output(0)) File “/code/ailice_14/tvm/release/python/tvm/topi/cuda/conv2d_direct.py”, line 50, in schedule_direct_cuda cfg.fallback_with_reference_log(ref_log) File “/code/ailice_14/tvm/release/python/tvm/autotvm/task/space.py”, line 1409, in fallback_with_reference_log factors = get_factors(int(np.prod(inp.config[knob_name].size))) File “/code/ailice_14/tvm/release/python/tvm/autotvm/task/space.py”, line 170, in get_factors ([i, n // i] for i in range(1, int(math.sqrt(n)) + 1, step) if n % i == 0), ValueError: math domain error
通过打日志发现，n为-5。继续向上查找，打印日志，我的target为：cuda -keys=cuda,gpu -arch=sm_70 -max_num_threads=1024 -model=v100 -thread_warp_size=32 ； workload为 (‘conv2d_nchw.cuda’, (‘TENSOR’, (1, 10, 70835, 1), ‘float32’), (‘TENSOR’, (48, 10, 1, 1), ‘float32’), (1, 1), (0, 0, 0, 0), (1, 1), ‘float32’)。 有什么办法可以解决这个报错吗？