trevor-m edited a comment on issue #5195: [RELAY] Fixes to MergeCompilerRegions URL: https://github.com/apache/incubator-tvm/pull/5195#issuecomment-607376003 Hi @mbaret, thanks again for fixing these issues! I noticed the call to MergeCompilerRegions hangs indefinitely for Densenet121 when all nodes are annotated. Here is a script to reproduce the issue. You will also need this commit https://github.com/trevor-m/tvm/commit/2b7c3b1d040abf1f88fca3857c65f13c4f012b2e to make sure that Tuple + concatenate is annotated properly also. There might a some infinite loop in the pass code or the runtime is scaling very poorly with respect to number of nodes. ``` from tvm import relay import mxnet as mx from mxnet.gluon.model_zoo.vision import get_model from tvm.relay import op as reg def _register_external_op_helper(op_name, supported=True): @reg.register(op_name, "target.test") def _func_wrapper(attrs, args): return supported return _func_wrapper _register_external_op_helper("nn.conv2d") _register_external_op_helper("nn.dense") _register_external_op_helper("nn.relu") _register_external_op_helper("add") _register_external_op_helper("multiply") _register_external_op_helper("nn.bias_add") _register_external_op_helper("nn.batch_flatten") _register_external_op_helper("nn.max_pool2d") _register_external_op_helper("nn.dropout") _register_external_op_helper("nn.batch_norm") _register_external_op_helper("nn.global_avg_pool2d") _register_external_op_helper("clip") _register_external_op_helper("concatenate") _register_external_op_helper("nn.avg_pool2d") def test_densenet(): block = get_model('densenet121', pretrained=True) mod, params = relay.frontend.from_mxnet(block, shape={'data': (1, 3, 224, 224)}, dtype='float32') mod = relay.transform.AnnotateTarget("test")(mod) print('Running MergeCompilerRegions...') mod = relay.transform.MergeCompilerRegions()(mod) mod = relay.transform.PartitionGraph()(mod) print(mod) test_densenet() ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
