manupa-arm commented on a change in pull request #10763: URL: https://github.com/apache/tvm/pull/10763#discussion_r836315102
########## File path: python/tvm/relay/backend/contrib/ethosu/te/common.py ########## @@ -0,0 +1,53 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +"""Common methods for the NPU tensor expressions""" + +from typing import Tuple, List + + +def get_layout_transform_matrices(ofm_channels: int) -> Tuple[List[List[float]], List[List[float]]]: Review comment: docs : Lets link to this page : https://developer.arm.com/documentation/102420/0200/Functional-description/Control-and-data-flow/Supported-memory-formats-for-feature-maps here. ########## File path: python/tvm/relay/backend/contrib/ethosu/te/common.py ########## @@ -0,0 +1,53 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +"""Common methods for the NPU tensor expressions""" + +from typing import Tuple, List + + +def get_layout_transform_matrices(ofm_channels: int) -> Tuple[List[List[float]], List[List[float]]]: + """Get the NHWC->NHCWB16 and NHCWB16->NHWC layout transform matrices. + + Parameters + ---------- + ofm_channels : int + The number of output channels in a NHWC layout + + Returns + ------- + nhwc_to_nhcwb16, nhcwb16_to_nhwc : Tuple[List[List[float]], List[List[float]]] + The layout transformation matrices + """ + + nhwc_to_nhcwb16 = [ + [1, 0, 0, 0, 0], + [0, 1, 0, 0, 0], + [0, 0, 0, 1 / 16, 0], + [0, 0, 1, 0, 0], + [0, 0, 0, 0, 16], Review comment: docs : might worth putting a comment to indicate that b16 axis is always going to be of fixed size of 16. ########## File path: python/tvm/relay/backend/contrib/ethosu/te/common.py ########## @@ -0,0 +1,53 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +"""Common methods for the NPU tensor expressions""" + +from typing import Tuple, List + + +def get_layout_transform_matrices(ofm_channels: int) -> Tuple[List[List[float]], List[List[float]]]: + """Get the NHWC->NHCWB16 and NHCWB16->NHWC layout transform matrices. + + Parameters + ---------- + ofm_channels : int + The number of output channels in a NHWC layout + + Returns + ------- + nhwc_to_nhcwb16, nhcwb16_to_nhwc : Tuple[List[List[float]], List[List[float]]] + The layout transformation matrices + """ + + nhwc_to_nhcwb16 = [ + [1, 0, 0, 0, 0], + [0, 1, 0, 0, 0], + [0, 0, 0, 1 / 16, 0], + [0, 0, 1, 0, 0], + [0, 0, 0, 0, 16], + [0, 0, 0, 0, 1], + ] + + nhcwb16_to_nhwc = [ + [1, 0, 0, 0, 0, 0], + [0, 1, 0, 0, 0, 0], + [0, 0, 0, 1, 0, 0], + [0, 0, 0, 0, 0, ofm_channels], Review comment: docs : lets state that the conversion of nhwc_to_nhcwb16 is lossy (because b16 axis is fixed to 16). Therefore, to recover the original "c" of "nhwc", we need to use the original channels in the transform matrix. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
