PyTorch-IPEX-DEMO.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Intel® Extension for PyTorch (IPEX)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Intel® Extension for PyTorch is a Python package to extend official PyTorch. It is designed to make the Out-of-Box user experience of PyTorch CPU better while achieving good performance. The extension brings new functions, but also ofurther ptimizations (for example, take advantage of Intel's new hardware features)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "import torch\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "from PIL import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/intel/inteloneapi/pytorch/latest/lib/python3.9/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead.\n",
      "  warnings.warn(\n",
      "/opt/intel/inteloneapi/pytorch/latest/lib/python3.9/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=None`.\n",
      "  warnings.warn(msg)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "ResNet(\n",
       "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (relu): ReLU(inplace=True)\n",
       "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "  (layer1): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (layer2): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (3): Bottleneck(\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (layer3): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (3): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (4): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (5): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (layer4): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
       "  (fc): Linear(in_features=2048, out_features=1000, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "net = torchvision.models.resnet50(pretrained=False)\n",
    "net.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = Image.open('content/flower.jpg').convert('RGB')\n",
    "resize = transforms.Resize([1280, 1920])\n",
    "img = resize(img)\n",
    "to_tensor = transforms.ToTensor()\n",
    "tensor = to_tensor(img)\n",
    "data = tensor.unsqueeze(0) # torch.Size([1, 3, 224, 224])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Performance evaluations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We conduct several tests with and without IPEX and for each test, we further evaluate running in Python as as well in JIT mode, that is, natively (outside Python).\n",
    "\n",
    "The following tests will be conducted:\n",
    "\n",
    "*With IPEX disabled:*\n",
    "\n",
    "1. Within PyTorch\n",
    "2. Natively\n",
    "\n",
    "*With IPEX enabled:*\n",
    "\n",
    "3. Within PyTorch\n",
    "4. Natively\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Without IPEX, in PyTorch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.42 s ± 78.3 ms per loop (mean ± std. dev. of 10 runs, 1 loop each)\n",
      "Runtime: 1.42 s ± 78.3 ms per loop (mean ± std. dev. of 10 runs, 1 loop each) s\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "time_no_ipex_obj = %timeit -r10 -o net(data)\n",
    "print('Runtime: {} s'.format(time_no_ipex_obj))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "time_no_ipex = time_no_ipex_obj.best"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Without IPEX, with TorchScript"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "TorchScript is a way to create serializable and optimizable models from PyTorch code. Any TorchScript program can be saved from a Python process and loaded in a process where there is no Python dependency.\n",
    "\n",
    "With Torchscript, a pure Python programcan be run independently from Python, such as in a standalone C++ program. This makes it possible to train models in PyTorch using familiar tools in Python and then export the model via TorchScript to a production environment where Python programs may be disadvantageous for performance and multi-threading reasons."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.48 s ± 39.4 ms per loop (mean ± std. dev. of 10 runs, 1 loop each)\n",
      "Runtime: 1.48 s ± 39.4 ms per loop (mean ± std. dev. of 10 runs, 1 loop each) s\n"
     ]
    }
   ],
   "source": [
    "net = torch.jit.trace(net, data)\n",
    "time_no_ipex_native_obj = %timeit -r10 -o net(data)\n",
    "print('Runtime: {} s'.format(time_no_ipex_native_obj))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "time_no_ipex_native = time_no_ipex_native_obj.best"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### With IPEX, in PyTorch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To explore the Intel® Extension for PyTorch, you just need to convert the model and input tensors to the extension device, then the extension will be enabled automatically. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's first reset the network from being jitted."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ResNet(\n",
       "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (relu): ReLU(inplace=True)\n",
       "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "  (layer1): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (layer2): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (3): Bottleneck(\n",
       "      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (layer3): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (3): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (4): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (5): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (layer4): Sequential(\n",
       "    (0): Bottleneck(\n",
       "      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): Bottleneck(\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "    (2): Bottleneck(\n",
       "      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
       "      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "    )\n",
       "  )\n",
       "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
       "  (fc): Linear(in_features=2048, out_features=1000, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "net = torchvision.models.resnet50(pretrained=False)\n",
    "net.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import intel_extension_for_pytorch as ipex\n",
    "model = net.to(memory_format=torch.channels_last)\n",
    "model = ipex.optimize(model)\n",
    "data = data.to(memory_format=torch.channels_last)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "788 ms ± 43.3 ms per loop (mean ± std. dev. of 10 runs, 1 loop each)\n",
      "Runtime: 788 ms ± 43.3 ms per loop (mean ± std. dev. of 10 runs, 1 loop each) s\n"
     ]
    }
   ],
   "source": [
    "time_with_ipex_obj = %timeit -r10 -o model(data)\n",
    "print('Runtime: {} s'.format(time_with_ipex_obj))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "time_with_ipex = time_with_ipex_obj.best"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### With IPEX, with TorchScript"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "830 ms ± 2.71 ms per loop (mean ± std. dev. of 10 runs, 1 loop each)\n",
      "Runtime: 830 ms ± 2.71 ms per loop (mean ± std. dev. of 10 runs, 1 loop each) s\n"
     ]
    }
   ],
   "source": [
    "model = torch.jit.trace(model, data)\n",
    "time_with_ipex_native_obj = %timeit -r10 -o model(data)\n",
    "print('Runtime: {} s'.format(time_with_ipex_native_obj))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "time_with_ipex_native = time_with_ipex_native_obj.best"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now visualize the results. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Import data visualization packages\n",
    "\n",
    "# Create arrays for the plot\n",
    "x_axis = ['Without IPEX, in PyTorch', 'Without IPEX, TorchScript', 'With IPEX, in PyTorch', 'With IPEX, TorchScript']\n",
    "x_pos = np.arange(len(x_axis))\n",
    "y_axis = [time_no_ipex, time_no_ipex_native, time_with_ipex, time_with_ipex_native]\n",
    "\n",
    "\n",
    "# Build the plot\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.bar(x_pos, y_axis, align='center', width=0.5, alpha=0.5, color=['black', 'red', 'green', 'blue'])\n",
    "ax.set_ylabel('Time (s)')\n",
    "ax.set_xticks(x_pos)\n",
    "ax.set_xticklabels(x_axis)\n",
    "plt.xticks(rotation = 90)\n",
    "ax.set_title('Timing w/ and w/o IPEX')\n",
    "ax.yaxis.grid(True)\n",
    "\n",
    "# Save the figure and show\n",
    "plt.tight_layout()\n",
    "plt.savefig('bar_plot.png')\n",
    "plt.show()\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "PyTorch (AI kit)",
   "language": "python",
   "name": "c009-intel_distribution_of_python_3_oneapi-beta05-pytorch"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  },
  "nbTranslate": {
   "displayLangs": [
    "*"
   ],
   "hotkey": "alt-t",
   "langInMainMenu": true,
   "sourceLang": "en",
   "targetLang": "fr",
   "useGoogleTranslate": true
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}