Cannot run autotuning against aarch64 android device

DEBUG:autotvm:No: 9     GFLOPS: 0.00/0.00       result: MeasureResult(costs=(InstantiationError(['Too large factor for unrolling', 'Too large factor for unrolli
ng']),), error_no=MeasureErrorNo.INSTANTIATION_ERROR, all_cost=0.006788015365600586, timestamp=1637888568.2369335)      [('tile_co', [-1, 2]), ('tile_oh', [-1,
3]), ('tile_ow', [-1, 160]), ('reorder_conv', [0, 1, 2, 3, 7, 4, 5, 6, 8, 9]), ('ann_reduce', ['unroll', 'none']), ('ann_spatial', ['unroll', 'unroll', 'vec']),
('compat', 2)],None,1976413
DEBUG:autotvm:No: 10    GFLOPS: 0.00/0.00       result: MeasureResult(costs=(InstantiationError(['Too large factor for unrolling', 'Too large factor for unrolli
ng']),), error_no=MeasureErrorNo.INSTANTIATION_ERROR, all_cost=0.006908416748046875, timestamp=1637888568.2370837)      [('tile_co', [-1, 12]), ('tile_oh', [-1,
72]), ('tile_ow', [-1, 160]), ('reorder_conv', [0, 1, 2, 3, 7, 4, 5, 8, 6, 9]), ('ann_reduce', ['unroll', 'none']), ('ann_spatial', ['unroll', 'vec', 'none']),
('compat', 2)],None,2015687
WARNING:root:Could not find any valid schedule for task Task(func_name=conv2d_nhwc_spatial_pack.arm_cpu, args=(('TENSOR', (1, 360, 640, 1), 'float32'), ('TENSOR
', (3, 3, 1, 12), 'float32'), (1, 1), (1, 1, 1, 1), (1, 1), 'float32'), kwargs={}, workload=('conv2d_nhwc_spatial_pack.arm_cpu', ('TENSOR', (1, 360, 640, 1), 'f
loat32'), ('TENSOR', (3, 3, 1, 12), 'float32'), (1, 1), (1, 1, 1, 1), (1, 1), 'float32')). A file containing the errors has been written to /tmp/tvm_tuning_erro
rs_9sl9h1b4.log.

in log files I can see:

Require environment variable TVM_NDK_CC to be the NDK standalone compiler
['Too large factor for unrolling', 'Too large factor for unrolling', 'Too large factor for unrolling', 'Too large factor for unrolling']
['Too large factor for unrolling', 'Too large factor for unrolling', 'Too large factor for unrolling', 'Too large factor for unrolling']
Require environment variable TVM_NDK_CC to be the NDK standalone compiler
['Too large factor for unrolling', 'Too large factor for unrolling']
['Too large factor for unrolling', 'Too large factor for unrolling']
Require environment variable TVM_NDK_CC to be the NDK standalone compiler
Require environment variable TVM_NDK_CC to be the NDK standalone compiler
['Too large factor for unrolling', 'Too large factor for unrolling']
['Too large factor for unrolling', 'Too large factor for unrolling']

Can I solve the problem using runtime with opencl?

After little digging in codebase, I can see that max_unroll can be set in tvm/python/tvm/topi/arm_cpu/conv2d_spatial_pack.py in lines 430…

after setting TVM_NDK_CC= Android/Sdk/ndk/21.4.7075529/toolchains/llvm/prebuilt/linux-x86_64/bin/aarch64-linux-android30-clang++, I think that I solved the problem