leandron commented on a change in pull request #7326: URL: https://github.com/apache/tvm/pull/7326#discussion_r562527228
########## File path: tutorials/auto_scheduler/tune_network_arm.py ########## @@ -0,0 +1,420 @@ +# 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. +""" +Auto-scheduling a Neural Network for ARM CPU +============================================= +**Author**: `Thierry Moreau <https://github.com/tmoreau89, Lianmin Zheng <https://github.com/merrymercy>>`_ + +Auto-tuning for specific devices and workloads is critical for getting the +best performance. This is a tutorial on how to tune a whole neural +network for ARM CPU with the auto-scheduler via RPC. + +To auto-tune a neural network, we partition the network into small subgraphs and +tune them independently. Each subgraph is treated as one search task. +A task scheduler slices the time and dynamically allocates time resources to +these tasks. The task scheduler predicts the impact of each task on the end-to-end +execution time and prioritizes the one that can reduce the execution time the most. + +For each subgraph, we use the compute declaration in :code:`tvm/python/topi` to +get the computational DAG in the tensor expression form. +We then use the auto-scheduler to construct a search space of this DAG and search +for good schedules (low-level optimizations). + +Different from the template-based :ref:`autotvm <tutorials-autotvm-sec>` which relies on +manual templates to define the search space, the auto-scheduler does not require any +schedule templates. In other words, the auto-scheduler only uses the compute declarations +in :code:`tvm/python/topi` and does not use existing schedule templates. + +Note that this tutorial will not run on Windows or recent versions of macOS. To +get it to run, you will need to wrap the body of this tutorial in a :code:`if +__name__ == "__main__":` block. +""" + +import numpy as np + +import tvm +from tvm import relay, auto_scheduler, autotvm +import tvm.relay.testing +from tvm.contrib import graph_runtime +from tvm.contrib.utils import tempdir + +################################################################# +# Define a Network +# ---------------- +# First, we need to define the network with relay frontend API. +# We can load some pre-defined network from :code:`tvm.relay.testing`. +# We can also load models from MXNet, ONNX, PyTorch, and TensorFlow +# (see :ref:`front end tutorials<tutorial-frontend>`). +# +# For convolutional neural networks, although auto-scheduler can work correctly +# with any layout, we found the best performance is typically achieved with NHWC layout. +# We also implemented more optimizations for NHWC layout with the auto-scheduler. +# So it is recommended to convert your models to NHWC layout to use the auto-scheduler. +# You can use :ref:`ConvertLayout <convert-layout-usage>` pass to do the layout conversion in TVM. + + +def get_network(name, batch_size, layout="NHWC", dtype="float32"): + """Get the symbol definition and random weight of a network""" + + # auto-scheduler prefers NHWC layout + if layout == "NHWC": + image_shape = (224, 224, 3) + elif layout == "NCHW": + image_shape = (3, 224, 224) + else: + raise ValueError("Invalid layout: " + layout) + + input_shape = (batch_size,) + image_shape + output_shape = (batch_size, 1000) + + if name.startswith("resnet-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name.startswith("resnet3d-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "mobilenet": + mod, params = relay.testing.mobilenet.get_workload( + batch_size=batch_size, layout=layout, dtype=dtype, image_shape=image_shape + ) + elif name == "squeezenet_v1.1": + assert layout == "NCHW", "squeezenet_v1.1 only supports NCHW layout" + mod, params = relay.testing.squeezenet.get_workload( + version="1.1", + batch_size=batch_size, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "inception_v3": + input_shape = (batch_size, 3, 299, 299) if layout == "NCHW" else (batch_size, 299, 299, 3) + mod, params = relay.testing.inception_v3.get_workload(batch_size=batch_size, dtype=dtype) + elif name == "mxnet": + # an example for mxnet model + from mxnet.gluon.model_zoo.vision import get_model + + assert layout == "NCHW" + + block = get_model("resnet50_v1", pretrained=True) + mod, params = relay.frontend.from_mxnet(block, shape={"data": input_shape}, dtype=dtype) + net = mod["main"] + net = relay.Function( + net.params, relay.nn.softmax(net.body), None, net.type_params, net.attrs + ) + mod = tvm.IRModule.from_expr(net) + + return mod, params, input_shape, output_shape + + +################################################################# +# Start RPC Tracker +# ----------------- +# TVM uses RPC session to communicate with ARM boards. +# During tuning, the tuner will send the generated code to the board and +# measure the speed of code on the board. +# +# To scale up the tuning, TVM uses RPC Tracker to manage distributed devices. +# The RPC Tracker is a centralized controller node. We can register all devices to +# the tracker. For example, if we have 10 phones, we can register all of them +# to the tracker, and run 10 measurements in parallel, accelerating the tuning process. +# +# To start an RPC tracker, run this command on the host machine. The tracker is +# required during the whole tuning process, so we need to open a new terminal for +# this command: +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_tracker --host=0.0.0.0 --port=9190 +# +# The expected output is +# +# .. code-block:: bash +# +# INFO:RPCTracker:bind to 0.0.0.0:9190 + +################################################################# +# Register Devices to RPC Tracker +# ----------------------------------- +# Now we can register our devices to the tracker. The first step is to +# build the TVM runtime for the ARM devices. +# +# * For Linux: +# Follow this section :ref:`build-tvm-runtime-on-device` to build +# the TVM runtime on the device. Then register the device to tracker by +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_server --tracker=[HOST_IP]:9190 --key=rk3399 +# +# (replace :code:`[HOST_IP]` with the IP address of your host machine) +# +# * For Android: +# Follow this `readme page <https://github.com/apache/tvm/tree/main/apps/android_rpc>`_ to +# install the TVM RPC APK on the android device. Make sure you can pass the android rpc test. +# Then you have already registered your device. During tuning, you have to go to developer option +# and enable "Keep screen awake during changing" and charge your phone to make it stable. +# +# After registering devices, we can confirm it by querying rpc_tracker +# +# .. code-block:: bash +# +# python -m tvm.exec.query_rpc_tracker --host=0.0.0.0 --port=9190 +# +# For example, if we have 2 Huawei mate10 pro, 11 Raspberry Pi 3B and 2 rk3399, +# the output can be +# +# .. code-block:: bash +# +# Queue Status +# ---------------------------------- +# key total free pending +# ---------------------------------- +# mate10pro 2 2 0 +# rk3399 2 2 0 +# rpi3b 11 11 0 +# ---------------------------------- +# +# You can register multiple devices to the tracker to accelerate the measurement in tuning. + +########################################### +# Set Tuning Options +# ------------------ +# Before tuning, we should apply some configurations. Here I use a Raspberry Pi 3b 4GB board +# as example. In your setting, you should modify the target and device_key accordingly. Review comment: I think it would be good to add a quick note on which OS are you using, plus a vital information for the toolchain - whether the OS is 32 or 64 bits. ########## File path: tutorials/auto_scheduler/tune_network_arm.py ########## @@ -0,0 +1,420 @@ +# 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. +""" +Auto-scheduling a Neural Network for ARM CPU +============================================= +**Author**: `Thierry Moreau <https://github.com/tmoreau89, Lianmin Zheng <https://github.com/merrymercy>>`_ + +Auto-tuning for specific devices and workloads is critical for getting the +best performance. This is a tutorial on how to tune a whole neural +network for ARM CPU with the auto-scheduler via RPC. + +To auto-tune a neural network, we partition the network into small subgraphs and +tune them independently. Each subgraph is treated as one search task. +A task scheduler slices the time and dynamically allocates time resources to +these tasks. The task scheduler predicts the impact of each task on the end-to-end +execution time and prioritizes the one that can reduce the execution time the most. + +For each subgraph, we use the compute declaration in :code:`tvm/python/topi` to +get the computational DAG in the tensor expression form. +We then use the auto-scheduler to construct a search space of this DAG and search +for good schedules (low-level optimizations). + +Different from the template-based :ref:`autotvm <tutorials-autotvm-sec>` which relies on +manual templates to define the search space, the auto-scheduler does not require any +schedule templates. In other words, the auto-scheduler only uses the compute declarations +in :code:`tvm/python/topi` and does not use existing schedule templates. + +Note that this tutorial will not run on Windows or recent versions of macOS. To +get it to run, you will need to wrap the body of this tutorial in a :code:`if +__name__ == "__main__":` block. +""" + +import numpy as np + +import tvm +from tvm import relay, auto_scheduler, autotvm +import tvm.relay.testing +from tvm.contrib import graph_runtime +from tvm.contrib.utils import tempdir + +################################################################# +# Define a Network +# ---------------- +# First, we need to define the network with relay frontend API. +# We can load some pre-defined network from :code:`tvm.relay.testing`. +# We can also load models from MXNet, ONNX, PyTorch, and TensorFlow +# (see :ref:`front end tutorials<tutorial-frontend>`). +# +# For convolutional neural networks, although auto-scheduler can work correctly +# with any layout, we found the best performance is typically achieved with NHWC layout. +# We also implemented more optimizations for NHWC layout with the auto-scheduler. +# So it is recommended to convert your models to NHWC layout to use the auto-scheduler. +# You can use :ref:`ConvertLayout <convert-layout-usage>` pass to do the layout conversion in TVM. + + +def get_network(name, batch_size, layout="NHWC", dtype="float32"): + """Get the symbol definition and random weight of a network""" + + # auto-scheduler prefers NHWC layout + if layout == "NHWC": + image_shape = (224, 224, 3) + elif layout == "NCHW": + image_shape = (3, 224, 224) + else: + raise ValueError("Invalid layout: " + layout) + + input_shape = (batch_size,) + image_shape + output_shape = (batch_size, 1000) + + if name.startswith("resnet-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name.startswith("resnet3d-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "mobilenet": + mod, params = relay.testing.mobilenet.get_workload( + batch_size=batch_size, layout=layout, dtype=dtype, image_shape=image_shape + ) + elif name == "squeezenet_v1.1": + assert layout == "NCHW", "squeezenet_v1.1 only supports NCHW layout" + mod, params = relay.testing.squeezenet.get_workload( + version="1.1", + batch_size=batch_size, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "inception_v3": + input_shape = (batch_size, 3, 299, 299) if layout == "NCHW" else (batch_size, 299, 299, 3) + mod, params = relay.testing.inception_v3.get_workload(batch_size=batch_size, dtype=dtype) + elif name == "mxnet": + # an example for mxnet model + from mxnet.gluon.model_zoo.vision import get_model + + assert layout == "NCHW" + + block = get_model("resnet50_v1", pretrained=True) + mod, params = relay.frontend.from_mxnet(block, shape={"data": input_shape}, dtype=dtype) + net = mod["main"] + net = relay.Function( + net.params, relay.nn.softmax(net.body), None, net.type_params, net.attrs + ) + mod = tvm.IRModule.from_expr(net) + + return mod, params, input_shape, output_shape + + +################################################################# +# Start RPC Tracker +# ----------------- +# TVM uses RPC session to communicate with ARM boards. +# During tuning, the tuner will send the generated code to the board and +# measure the speed of code on the board. +# +# To scale up the tuning, TVM uses RPC Tracker to manage distributed devices. +# The RPC Tracker is a centralized controller node. We can register all devices to +# the tracker. For example, if we have 10 phones, we can register all of them +# to the tracker, and run 10 measurements in parallel, accelerating the tuning process. +# +# To start an RPC tracker, run this command on the host machine. The tracker is +# required during the whole tuning process, so we need to open a new terminal for +# this command: +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_tracker --host=0.0.0.0 --port=9190 +# +# The expected output is +# +# .. code-block:: bash +# +# INFO:RPCTracker:bind to 0.0.0.0:9190 + +################################################################# +# Register Devices to RPC Tracker +# ----------------------------------- +# Now we can register our devices to the tracker. The first step is to +# build the TVM runtime for the ARM devices. +# +# * For Linux: +# Follow this section :ref:`build-tvm-runtime-on-device` to build +# the TVM runtime on the device. Then register the device to tracker by +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_server --tracker=[HOST_IP]:9190 --key=rk3399 +# +# (replace :code:`[HOST_IP]` with the IP address of your host machine) +# +# * For Android: +# Follow this `readme page <https://github.com/apache/tvm/tree/main/apps/android_rpc>`_ to +# install the TVM RPC APK on the android device. Make sure you can pass the android rpc test. +# Then you have already registered your device. During tuning, you have to go to developer option +# and enable "Keep screen awake during changing" and charge your phone to make it stable. +# +# After registering devices, we can confirm it by querying rpc_tracker +# +# .. code-block:: bash +# +# python -m tvm.exec.query_rpc_tracker --host=0.0.0.0 --port=9190 +# +# For example, if we have 2 Huawei mate10 pro, 11 Raspberry Pi 3B and 2 rk3399, +# the output can be +# +# .. code-block:: bash +# +# Queue Status +# ---------------------------------- +# key total free pending +# ---------------------------------- +# mate10pro 2 2 0 +# rk3399 2 2 0 +# rpi3b 11 11 0 +# ---------------------------------- +# +# You can register multiple devices to the tracker to accelerate the measurement in tuning. + +########################################### +# Set Tuning Options +# ------------------ +# Before tuning, we should apply some configurations. Here I use a Raspberry Pi 3b 4GB board +# as example. In your setting, you should modify the target and device_key accordingly. +# set :code:`use_android` to True if you use android phone. + +#### DEVICE CONFIG #### + +# Replace "aarch64-linux-gnu" with the correct target of your board. Review comment: The mention here to `aarch64-linux-gnu` needs a bit more context, as there is no mention to it later. ########## File path: tutorials/auto_scheduler/tune_network_arm.py ########## @@ -0,0 +1,420 @@ +# 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. +""" +Auto-scheduling a Neural Network for ARM CPU +============================================= +**Author**: `Thierry Moreau <https://github.com/tmoreau89, Lianmin Zheng <https://github.com/merrymercy>>`_ + +Auto-tuning for specific devices and workloads is critical for getting the +best performance. This is a tutorial on how to tune a whole neural +network for ARM CPU with the auto-scheduler via RPC. + +To auto-tune a neural network, we partition the network into small subgraphs and +tune them independently. Each subgraph is treated as one search task. +A task scheduler slices the time and dynamically allocates time resources to +these tasks. The task scheduler predicts the impact of each task on the end-to-end +execution time and prioritizes the one that can reduce the execution time the most. + +For each subgraph, we use the compute declaration in :code:`tvm/python/topi` to +get the computational DAG in the tensor expression form. +We then use the auto-scheduler to construct a search space of this DAG and search +for good schedules (low-level optimizations). + +Different from the template-based :ref:`autotvm <tutorials-autotvm-sec>` which relies on +manual templates to define the search space, the auto-scheduler does not require any +schedule templates. In other words, the auto-scheduler only uses the compute declarations +in :code:`tvm/python/topi` and does not use existing schedule templates. + +Note that this tutorial will not run on Windows or recent versions of macOS. To +get it to run, you will need to wrap the body of this tutorial in a :code:`if +__name__ == "__main__":` block. +""" + +import numpy as np + +import tvm +from tvm import relay, auto_scheduler, autotvm +import tvm.relay.testing +from tvm.contrib import graph_runtime +from tvm.contrib.utils import tempdir + +################################################################# +# Define a Network +# ---------------- +# First, we need to define the network with relay frontend API. +# We can load some pre-defined network from :code:`tvm.relay.testing`. +# We can also load models from MXNet, ONNX, PyTorch, and TensorFlow +# (see :ref:`front end tutorials<tutorial-frontend>`). +# +# For convolutional neural networks, although auto-scheduler can work correctly +# with any layout, we found the best performance is typically achieved with NHWC layout. +# We also implemented more optimizations for NHWC layout with the auto-scheduler. +# So it is recommended to convert your models to NHWC layout to use the auto-scheduler. +# You can use :ref:`ConvertLayout <convert-layout-usage>` pass to do the layout conversion in TVM. + + +def get_network(name, batch_size, layout="NHWC", dtype="float32"): + """Get the symbol definition and random weight of a network""" + + # auto-scheduler prefers NHWC layout + if layout == "NHWC": + image_shape = (224, 224, 3) + elif layout == "NCHW": + image_shape = (3, 224, 224) + else: + raise ValueError("Invalid layout: " + layout) + + input_shape = (batch_size,) + image_shape + output_shape = (batch_size, 1000) + + if name.startswith("resnet-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name.startswith("resnet3d-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "mobilenet": + mod, params = relay.testing.mobilenet.get_workload( + batch_size=batch_size, layout=layout, dtype=dtype, image_shape=image_shape + ) + elif name == "squeezenet_v1.1": + assert layout == "NCHW", "squeezenet_v1.1 only supports NCHW layout" + mod, params = relay.testing.squeezenet.get_workload( + version="1.1", + batch_size=batch_size, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "inception_v3": + input_shape = (batch_size, 3, 299, 299) if layout == "NCHW" else (batch_size, 299, 299, 3) + mod, params = relay.testing.inception_v3.get_workload(batch_size=batch_size, dtype=dtype) + elif name == "mxnet": + # an example for mxnet model + from mxnet.gluon.model_zoo.vision import get_model + + assert layout == "NCHW" + + block = get_model("resnet50_v1", pretrained=True) + mod, params = relay.frontend.from_mxnet(block, shape={"data": input_shape}, dtype=dtype) + net = mod["main"] + net = relay.Function( + net.params, relay.nn.softmax(net.body), None, net.type_params, net.attrs + ) + mod = tvm.IRModule.from_expr(net) + + return mod, params, input_shape, output_shape + + +################################################################# +# Start RPC Tracker +# ----------------- +# TVM uses RPC session to communicate with ARM boards. +# During tuning, the tuner will send the generated code to the board and +# measure the speed of code on the board. +# +# To scale up the tuning, TVM uses RPC Tracker to manage distributed devices. +# The RPC Tracker is a centralized controller node. We can register all devices to +# the tracker. For example, if we have 10 phones, we can register all of them +# to the tracker, and run 10 measurements in parallel, accelerating the tuning process. +# +# To start an RPC tracker, run this command on the host machine. The tracker is +# required during the whole tuning process, so we need to open a new terminal for +# this command: +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_tracker --host=0.0.0.0 --port=9190 +# +# The expected output is +# +# .. code-block:: bash +# +# INFO:RPCTracker:bind to 0.0.0.0:9190 + +################################################################# +# Register Devices to RPC Tracker +# ----------------------------------- +# Now we can register our devices to the tracker. The first step is to +# build the TVM runtime for the ARM devices. +# +# * For Linux: +# Follow this section :ref:`build-tvm-runtime-on-device` to build +# the TVM runtime on the device. Then register the device to tracker by +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_server --tracker=[HOST_IP]:9190 --key=rk3399 Review comment: The key for the example device, defined on line 218 is `rasp4b-64`, so this probably needs to match? ########## File path: tutorials/auto_scheduler/tune_network_arm.py ########## @@ -0,0 +1,420 @@ +# 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. +""" +Auto-scheduling a Neural Network for ARM CPU +============================================= +**Author**: `Thierry Moreau <https://github.com/tmoreau89, Lianmin Zheng <https://github.com/merrymercy>>`_ + +Auto-tuning for specific devices and workloads is critical for getting the +best performance. This is a tutorial on how to tune a whole neural +network for ARM CPU with the auto-scheduler via RPC. + +To auto-tune a neural network, we partition the network into small subgraphs and +tune them independently. Each subgraph is treated as one search task. +A task scheduler slices the time and dynamically allocates time resources to +these tasks. The task scheduler predicts the impact of each task on the end-to-end +execution time and prioritizes the one that can reduce the execution time the most. + +For each subgraph, we use the compute declaration in :code:`tvm/python/topi` to +get the computational DAG in the tensor expression form. +We then use the auto-scheduler to construct a search space of this DAG and search +for good schedules (low-level optimizations). + +Different from the template-based :ref:`autotvm <tutorials-autotvm-sec>` which relies on +manual templates to define the search space, the auto-scheduler does not require any +schedule templates. In other words, the auto-scheduler only uses the compute declarations +in :code:`tvm/python/topi` and does not use existing schedule templates. + +Note that this tutorial will not run on Windows or recent versions of macOS. To +get it to run, you will need to wrap the body of this tutorial in a :code:`if +__name__ == "__main__":` block. +""" + +import numpy as np + +import tvm +from tvm import relay, auto_scheduler, autotvm +import tvm.relay.testing +from tvm.contrib import graph_runtime +from tvm.contrib.utils import tempdir + +################################################################# +# Define a Network +# ---------------- +# First, we need to define the network with relay frontend API. +# We can load some pre-defined network from :code:`tvm.relay.testing`. +# We can also load models from MXNet, ONNX, PyTorch, and TensorFlow +# (see :ref:`front end tutorials<tutorial-frontend>`). +# +# For convolutional neural networks, although auto-scheduler can work correctly +# with any layout, we found the best performance is typically achieved with NHWC layout. +# We also implemented more optimizations for NHWC layout with the auto-scheduler. +# So it is recommended to convert your models to NHWC layout to use the auto-scheduler. +# You can use :ref:`ConvertLayout <convert-layout-usage>` pass to do the layout conversion in TVM. + + +def get_network(name, batch_size, layout="NHWC", dtype="float32"): + """Get the symbol definition and random weight of a network""" + + # auto-scheduler prefers NHWC layout + if layout == "NHWC": + image_shape = (224, 224, 3) + elif layout == "NCHW": + image_shape = (3, 224, 224) + else: + raise ValueError("Invalid layout: " + layout) + + input_shape = (batch_size,) + image_shape + output_shape = (batch_size, 1000) + + if name.startswith("resnet-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name.startswith("resnet3d-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "mobilenet": + mod, params = relay.testing.mobilenet.get_workload( + batch_size=batch_size, layout=layout, dtype=dtype, image_shape=image_shape + ) + elif name == "squeezenet_v1.1": + assert layout == "NCHW", "squeezenet_v1.1 only supports NCHW layout" + mod, params = relay.testing.squeezenet.get_workload( + version="1.1", + batch_size=batch_size, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "inception_v3": + input_shape = (batch_size, 3, 299, 299) if layout == "NCHW" else (batch_size, 299, 299, 3) + mod, params = relay.testing.inception_v3.get_workload(batch_size=batch_size, dtype=dtype) + elif name == "mxnet": + # an example for mxnet model + from mxnet.gluon.model_zoo.vision import get_model + + assert layout == "NCHW" + + block = get_model("resnet50_v1", pretrained=True) + mod, params = relay.frontend.from_mxnet(block, shape={"data": input_shape}, dtype=dtype) + net = mod["main"] + net = relay.Function( + net.params, relay.nn.softmax(net.body), None, net.type_params, net.attrs + ) + mod = tvm.IRModule.from_expr(net) + + return mod, params, input_shape, output_shape + + +################################################################# +# Start RPC Tracker +# ----------------- +# TVM uses RPC session to communicate with ARM boards. +# During tuning, the tuner will send the generated code to the board and +# measure the speed of code on the board. +# +# To scale up the tuning, TVM uses RPC Tracker to manage distributed devices. +# The RPC Tracker is a centralized controller node. We can register all devices to +# the tracker. For example, if we have 10 phones, we can register all of them +# to the tracker, and run 10 measurements in parallel, accelerating the tuning process. +# +# To start an RPC tracker, run this command on the host machine. The tracker is +# required during the whole tuning process, so we need to open a new terminal for +# this command: +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_tracker --host=0.0.0.0 --port=9190 +# +# The expected output is +# +# .. code-block:: bash +# +# INFO:RPCTracker:bind to 0.0.0.0:9190 + +################################################################# +# Register Devices to RPC Tracker +# ----------------------------------- +# Now we can register our devices to the tracker. The first step is to +# build the TVM runtime for the ARM devices. +# +# * For Linux: +# Follow this section :ref:`build-tvm-runtime-on-device` to build +# the TVM runtime on the device. Then register the device to tracker by +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_server --tracker=[HOST_IP]:9190 --key=rk3399 +# +# (replace :code:`[HOST_IP]` with the IP address of your host machine) +# +# * For Android: +# Follow this `readme page <https://github.com/apache/tvm/tree/main/apps/android_rpc>`_ to +# install the TVM RPC APK on the android device. Make sure you can pass the android rpc test. +# Then you have already registered your device. During tuning, you have to go to developer option +# and enable "Keep screen awake during changing" and charge your phone to make it stable. +# +# After registering devices, we can confirm it by querying rpc_tracker +# +# .. code-block:: bash +# +# python -m tvm.exec.query_rpc_tracker --host=0.0.0.0 --port=9190 +# +# For example, if we have 2 Huawei mate10 pro, 11 Raspberry Pi 3B and 2 rk3399, +# the output can be +# +# .. code-block:: bash +# +# Queue Status +# ---------------------------------- +# key total free pending +# ---------------------------------- +# mate10pro 2 2 0 +# rk3399 2 2 0 +# rpi3b 11 11 0 +# ---------------------------------- +# +# You can register multiple devices to the tracker to accelerate the measurement in tuning. + +########################################### +# Set Tuning Options +# ------------------ +# Before tuning, we should apply some configurations. Here I use a Raspberry Pi 3b 4GB board +# as example. In your setting, you should modify the target and device_key accordingly. +# set :code:`use_android` to True if you use android phone. + +#### DEVICE CONFIG #### + +# Replace "aarch64-linux-gnu" with the correct target of your board. +# This target is used for cross compilation. You can query it by :code:`gcc -v` on your device. +target = tvm.target.arm_cpu("rasp4b64") + +# Also replace this with the device key in your tracker +device_key = "rasp4b-64" + Review comment: You mention Raspberry Pi 3 (line 207) in the comment above, maybe this need to encode that info, hence `rasp3b` maybe? ########## File path: tutorials/auto_scheduler/tune_network_arm.py ########## @@ -0,0 +1,420 @@ +# 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. +""" +Auto-scheduling a Neural Network for ARM CPU +============================================= +**Author**: `Thierry Moreau <https://github.com/tmoreau89, Lianmin Zheng <https://github.com/merrymercy>>`_ + +Auto-tuning for specific devices and workloads is critical for getting the +best performance. This is a tutorial on how to tune a whole neural +network for ARM CPU with the auto-scheduler via RPC. + +To auto-tune a neural network, we partition the network into small subgraphs and +tune them independently. Each subgraph is treated as one search task. +A task scheduler slices the time and dynamically allocates time resources to +these tasks. The task scheduler predicts the impact of each task on the end-to-end +execution time and prioritizes the one that can reduce the execution time the most. + +For each subgraph, we use the compute declaration in :code:`tvm/python/topi` to +get the computational DAG in the tensor expression form. +We then use the auto-scheduler to construct a search space of this DAG and search +for good schedules (low-level optimizations). + +Different from the template-based :ref:`autotvm <tutorials-autotvm-sec>` which relies on +manual templates to define the search space, the auto-scheduler does not require any +schedule templates. In other words, the auto-scheduler only uses the compute declarations +in :code:`tvm/python/topi` and does not use existing schedule templates. + +Note that this tutorial will not run on Windows or recent versions of macOS. To +get it to run, you will need to wrap the body of this tutorial in a :code:`if +__name__ == "__main__":` block. +""" + +import numpy as np + +import tvm +from tvm import relay, auto_scheduler, autotvm +import tvm.relay.testing +from tvm.contrib import graph_runtime +from tvm.contrib.utils import tempdir + +################################################################# +# Define a Network +# ---------------- +# First, we need to define the network with relay frontend API. +# We can load some pre-defined network from :code:`tvm.relay.testing`. +# We can also load models from MXNet, ONNX, PyTorch, and TensorFlow +# (see :ref:`front end tutorials<tutorial-frontend>`). +# +# For convolutional neural networks, although auto-scheduler can work correctly +# with any layout, we found the best performance is typically achieved with NHWC layout. +# We also implemented more optimizations for NHWC layout with the auto-scheduler. +# So it is recommended to convert your models to NHWC layout to use the auto-scheduler. +# You can use :ref:`ConvertLayout <convert-layout-usage>` pass to do the layout conversion in TVM. + + +def get_network(name, batch_size, layout="NHWC", dtype="float32"): + """Get the symbol definition and random weight of a network""" + + # auto-scheduler prefers NHWC layout + if layout == "NHWC": + image_shape = (224, 224, 3) + elif layout == "NCHW": + image_shape = (3, 224, 224) + else: + raise ValueError("Invalid layout: " + layout) + + input_shape = (batch_size,) + image_shape + output_shape = (batch_size, 1000) + + if name.startswith("resnet-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name.startswith("resnet3d-"): + n_layer = int(name.split("-")[1]) + mod, params = relay.testing.resnet.get_workload( + num_layers=n_layer, + batch_size=batch_size, + layout=layout, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "mobilenet": + mod, params = relay.testing.mobilenet.get_workload( + batch_size=batch_size, layout=layout, dtype=dtype, image_shape=image_shape + ) + elif name == "squeezenet_v1.1": + assert layout == "NCHW", "squeezenet_v1.1 only supports NCHW layout" + mod, params = relay.testing.squeezenet.get_workload( + version="1.1", + batch_size=batch_size, + dtype=dtype, + image_shape=image_shape, + ) + elif name == "inception_v3": + input_shape = (batch_size, 3, 299, 299) if layout == "NCHW" else (batch_size, 299, 299, 3) + mod, params = relay.testing.inception_v3.get_workload(batch_size=batch_size, dtype=dtype) + elif name == "mxnet": + # an example for mxnet model + from mxnet.gluon.model_zoo.vision import get_model + + assert layout == "NCHW" + + block = get_model("resnet50_v1", pretrained=True) + mod, params = relay.frontend.from_mxnet(block, shape={"data": input_shape}, dtype=dtype) + net = mod["main"] + net = relay.Function( + net.params, relay.nn.softmax(net.body), None, net.type_params, net.attrs + ) + mod = tvm.IRModule.from_expr(net) + + return mod, params, input_shape, output_shape + + +################################################################# +# Start RPC Tracker +# ----------------- +# TVM uses RPC session to communicate with ARM boards. +# During tuning, the tuner will send the generated code to the board and +# measure the speed of code on the board. +# +# To scale up the tuning, TVM uses RPC Tracker to manage distributed devices. +# The RPC Tracker is a centralized controller node. We can register all devices to +# the tracker. For example, if we have 10 phones, we can register all of them +# to the tracker, and run 10 measurements in parallel, accelerating the tuning process. +# +# To start an RPC tracker, run this command on the host machine. The tracker is +# required during the whole tuning process, so we need to open a new terminal for +# this command: +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_tracker --host=0.0.0.0 --port=9190 +# +# The expected output is +# +# .. code-block:: bash +# +# INFO:RPCTracker:bind to 0.0.0.0:9190 + +################################################################# +# Register Devices to RPC Tracker +# ----------------------------------- +# Now we can register our devices to the tracker. The first step is to +# build the TVM runtime for the ARM devices. +# +# * For Linux: +# Follow this section :ref:`build-tvm-runtime-on-device` to build +# the TVM runtime on the device. Then register the device to tracker by +# +# .. code-block:: bash +# +# python -m tvm.exec.rpc_server --tracker=[HOST_IP]:9190 --key=rk3399 +# +# (replace :code:`[HOST_IP]` with the IP address of your host machine) +# +# * For Android: +# Follow this `readme page <https://github.com/apache/tvm/tree/main/apps/android_rpc>`_ to +# install the TVM RPC APK on the android device. Make sure you can pass the android rpc test. +# Then you have already registered your device. During tuning, you have to go to developer option +# and enable "Keep screen awake during changing" and charge your phone to make it stable. +# +# After registering devices, we can confirm it by querying rpc_tracker +# +# .. code-block:: bash +# +# python -m tvm.exec.query_rpc_tracker --host=0.0.0.0 --port=9190 +# +# For example, if we have 2 Huawei mate10 pro, 11 Raspberry Pi 3B and 2 rk3399, +# the output can be +# +# .. code-block:: bash +# +# Queue Status +# ---------------------------------- +# key total free pending +# ---------------------------------- +# mate10pro 2 2 0 +# rk3399 2 2 0 +# rpi3b 11 11 0 +# ---------------------------------- Review comment: To keep the example coherent, it would be good also to update here when the key is updated. ---------------------------------------------------------------- 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: us...@infra.apache.org
