ashutosh-arm commented on code in PR #13092: URL: https://github.com/apache/tvm/pull/13092#discussion_r999319824
########## src/relay/backend/contrib/ethosn/inline_partitions.cc: ########## @@ -0,0 +1,120 @@ +/* + * 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. + */ + +/*! + * \file src/relay/backend/contrib/ethosn/inline_partitions.cc + * \brief A pass to inline NPU partitions that are not considered compute + * intensive. + */ + +#include <tvm/relay/expr.h> +#include <tvm/relay/expr_functor.h> + +#include "../../../transforms/compiler_function_utils.h" + +namespace tvm { +namespace relay { +namespace contrib { +namespace ethosn { + +class IsComputeIntensivePartition : MixedModeVisitor { + public: + /*! + * \brief Check if the partitioned function is compute + * intensive. If it has not multiply-accumulate operations + * it is not considered compute intensive. + * + * \param expr The partitioned function to check. + */ + bool CheckSubgraph(const Expr& expr) { + is_compute_intensive = false; + VisitExpr(expr); + return is_compute_intensive; + } + + /*! + * \brief Visit the call nodes of a partitioned function + * and check if operators or composite functions make the + * partitioned function compute intensive. + * + * \param op The call node to check. + */ + void VisitExpr_(const CallNode* op) override { + Call call = GetRef<Call>(op); + std::string op_name = ""; + if (const auto* op = call->op.as<OpNode>()) { + op_name = op->name; + } else if (const auto* func = call->op.as<FunctionNode>()) { + op_name = func->GetAttr<String>(attr::kComposite, "").value(); + } + + if (op_name != "") { + if (compute_intensive_operators.find(op_name) != compute_intensive_operators.end()) { + is_compute_intensive = true; + } + } + } + + private: + /*! \brief Whether or not the partitioned function is consdiered compute intensive. */ + bool is_compute_intensive; + /*! \brief A set of operators considered compute intensive. */ + const std::unordered_set<std::string> compute_intensive_operators{ + "ethos-n.qnn_add", "ethos-n.qnn_conv2d", "ethos-n.qnn_conv2d_transpose", Review Comment: Question: By def compute intensive ops are that involve macs, why is qnn.add included in this list? Is it because of scalar multiplies and zp additions per pixel still happen? Also, why depth_to_space? -- 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: commits-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
