A4_Plotting_Layer_Runtime_Energy_Stack.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "A4-Plotting-Layer-Runtime-Energy-Stack.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "3ool5f9Kv2VB"
      ],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mehrotrasan16/Keras-Deep-Learning/blob/04-Plotting-Stack-Layers/A4_Plotting_Layer_Runtime_Energy_Stack.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4FEEgxP78M13"
      },
      "source": [
        "## Imports"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T8FTv4xkVUaY"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import glob"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CelpUCslWNWE"
      },
      "source": [
        "import os\n",
        "os.chdir('..')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i4ksePcz8Pf5"
      },
      "source": [
        "## Resnet Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Xtvt77CMVZfs"
      },
      "source": [
        "resnet_csv_files = glob.glob(\"**/Resnet/*.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u76HspimVtsY",
        "outputId": "9f795ddc-d976-45a5-9970-a8f4e89e1c83",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "resnet_csv_files"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['content/Resnet/Resnet50_rs.csv',\n",
              " 'content/Resnet/Resnet50_yxp_os.csv',\n",
              " 'content/Resnet/Resnet50_xp_ws.csv',\n",
              " 'content/Resnet/Resnet50_nlr.csv']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 153
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g0Az68HVVuqk",
        "outputId": "d1f01ddd-e622-4d59-a6ca-d2a7ab74a788",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "runtimes = {}\n",
        "energies = {}\n",
        "for file in resnet_csv_files:\n",
        "    with open(file,'r') as f:\n",
        "        print(f'Filename: {file}')\n",
        "        df = pd.read_csv(f)\n",
        "        df = df.set_index(' Layer Number')\n",
        "        df = df[[' Runtime (Cycles)',  ' Activity count-based Energy (nJ)']]\n",
        "        fc = df.filter(like='FC', axis=0).sum(axis = 0)\n",
        "        \n",
        "        residual = df.filter(like='_Residual', axis=0).sum(axis = 0)\n",
        "        \n",
        "        point_wise = df.filter(regex='CONV._._1', axis=0).sum(axis = 0)\n",
        "        \n",
        "        early_layer = df.filter(regex='CONV1', axis=0).sum(axis = 0)\n",
        "        \n",
        "        late_layer = df.filter(regex='CONV._._[234]', axis=0).sum(axis = 0)\n",
        "        \n",
        "        layers = [fc,residual,point_wise,early_layer,late_layer]\n",
        "        filename = file.split('/')[2].split('.')[0]\n",
        "        runtimes[filename] = [x[0] for x in layers]  \n",
        "        energies[filename] = [x[1]/10**6 for x in layers]  \n",
        "\n",
        "        # print(fc.head(),fc.count())\n",
        "        # print(residual.head(),residual.count())\n",
        "        # print(point_wise.head(),point_wise.count())\n",
        "        # print(early_layer.head(),early_layer.count())\n",
        "        # print(late_layer.head(),late_layer.count())\n",
        "        \n",
        "        # print('\\n\\n\\n\\n\\n\\n')\n",
        "print(runtimes)\n",
        "pd.DataFrame(runtimes,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Runtime(cycles)',colormap='Set1')\n",
        "pd.DataFrame(energies,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Energy(mJ)',colormap='Set2')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Filename: content/Resnet/Resnet50_rs.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/Resnet/Resnet50_yxp_os.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/Resnet/Resnet50_xp_ws.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/Resnet/Resnet50_nlr.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "{'Resnet50_rs': [4861006.0, 197152.0, 13295156.0, 2396770.0, 25355670.0], 'Resnet50_yxp_os': [20480004.0, 194087.0, 27205669.0, 1023850.0, 46658638.0], 'Resnet50_xp_ws': [12288004.0, 272400.0, 46333968.0, 297449.0, 69137275.0], 'Resnet50_nlr': [5488790.0, 198600.0, 26493138.0, 4562314.0, 42421436.0]}\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc5516decc0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 168
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RsG4M_CJvMWf"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3ool5f9Kv2VB"
      },
      "source": [
        "## Scratchpad code for ResNet"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MUicG4aTnnhg",
        "outputId": "36095e85-aaad-49ed-98c0-19d0a14e29fa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        }
      },
      "source": [
        "pd.DataFrame(runtimes,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc551ec17f0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 129
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hcwOO8fiouCu",
        "outputId": "ad248608-99d2-45a1-a1e1-ea48cd3196c4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 290
        }
      },
      "source": [
        "pd.DataFrame(runtimes).plot.bar(stacked=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc551f8fa58>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 125
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U_KWEMTblsfB",
        "outputId": "a067c2dd-c8ec-4f7a-8e23-d7905f80c029",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 363
        }
      },
      "source": [
        "pd.DataFrame(pd.concat([point_wise,residual, fc ,late_layer, early_layer]))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <td>1806352</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <td>19541</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <td>15500</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <td>335</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <td>24001</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <td>5521</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <td>6684960</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <td>38716</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <td>4562305</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <td>1093</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                         0\n",
              " Runtime (Cycles)                  1806352\n",
              " Activity count-based Energy (nJ)    19541\n",
              " Runtime (Cycles)                    15500\n",
              " Activity count-based Energy (nJ)      335\n",
              " Runtime (Cycles)                    24001\n",
              " Activity count-based Energy (nJ)     5521\n",
              " Runtime (Cycles)                  6684960\n",
              " Activity count-based Energy (nJ)    38716\n",
              " Runtime (Cycles)                  4562305\n",
              " Activity count-based Energy (nJ)     1093"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 105
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MXdHdJ-9kMlD",
        "outputId": "7ca18565-376e-4c31-a797-f2ad8d48fe4e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 447
        }
      },
      "source": [
        "pd.concat([late_layer, early_layer]).plot.bar(stacked=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc552baf630>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 102
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "haQxnvgKWe07",
        "outputId": "922df236-3db5-4376-ed56-11f2c01792e1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "df[[' Layer Number',]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                CONV1\n",
              "1            CONV2_1_1\n",
              "2            CONV2_1_2\n",
              "3            CONV2_1_3\n",
              "4     CONV2_1_Residual\n",
              "            ...       \n",
              "61           CONV5_3_1\n",
              "62           CONV5_3_2\n",
              "63           CONV5_3_3\n",
              "64    CONV5_3_Residual\n",
              "65              FC1000\n",
              "Name:  Layer Number, Length: 66, dtype: object"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "18U2BnrqWjHN",
        "outputId": "b956e539-4c52-440f-ee5b-ab65ab9b5152",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "df.columns"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['Neural Network Name', ' Layer Number', ' NumPEs', ' Runtime (Cycles)',\n",
              "       ' Activity count-based Energy (nJ)', ' Throughput (MACs/Cycle)',\n",
              "       ' Throughput Per Energy (GMACs/s*J)', ' Area', ' Power',\n",
              "       ' NoC BW (Elements/cycle)', ' Avg BW Req', ' Peak BW Req',\n",
              "       ' Vector Width', '  L2 SRAM Size (Bytes)', ' L1 SRAM Size (Bytes)',\n",
              "       ' Multicasting Factor (Weight)', ' Multicasting Factor (Input)',\n",
              "       ' Num Total Input Pixels', ' Num Total Weight Pixels', ' Ops/J',\n",
              "       ' Num MACs', ' PE power', ' L1 power', ' L2 power', ' NOC power',\n",
              "       ' input l1 read', ' input l1 write', ' input l2 read',\n",
              "       ' input l2 write', ' input reuse factor', 'filter l1 read',\n",
              "       ' filter l1 write', ' filter l2 read', ' filter l2 write',\n",
              "       ' filter reuse factor', 'output l1 read', ' output l1 write',\n",
              "       ' output l2 read', ' output l2 write', ' output reuse factor',\n",
              "       'Ingress Delay (Min)', ' Ingress Delay (Max)', ' Ingress Delay (Avg)',\n",
              "       ' Egress Delay (Min)', ' Egress Delay (Max)', '  Egress Delay (Avg)',\n",
              "       'Compute Delay (Min)', ' Compute Delay (Min)', ' Compute Delay (Avg)',\n",
              "       'Avg number of utilized PEs', ' Arithmetic Intensity'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 37
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v6Epm_rlXCcV",
        "outputId": "95ccd2f4-eb1b-4d33-8d32-e6bbc3882fc4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        }
      },
      "source": [
        "df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Neural Network Name</th>\n",
              "      <th>Layer Number</th>\n",
              "      <th>NumPEs</th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <th>Throughput (MACs/Cycle)</th>\n",
              "      <th>Throughput Per Energy (GMACs/s*J)</th>\n",
              "      <th>Area</th>\n",
              "      <th>Power</th>\n",
              "      <th>NoC BW (Elements/cycle)</th>\n",
              "      <th>Avg BW Req</th>\n",
              "      <th>Peak BW Req</th>\n",
              "      <th>Vector Width</th>\n",
              "      <th>L2 SRAM Size (Bytes)</th>\n",
              "      <th>L1 SRAM Size (Bytes)</th>\n",
              "      <th>Multicasting Factor (Weight)</th>\n",
              "      <th>Multicasting Factor (Input)</th>\n",
              "      <th>Num Total Input Pixels</th>\n",
              "      <th>Num Total Weight Pixels</th>\n",
              "      <th>Ops/J</th>\n",
              "      <th>Num MACs</th>\n",
              "      <th>PE power</th>\n",
              "      <th>L1 power</th>\n",
              "      <th>L2 power</th>\n",
              "      <th>NOC power</th>\n",
              "      <th>input l1 read</th>\n",
              "      <th>input l1 write</th>\n",
              "      <th>input l2 read</th>\n",
              "      <th>input l2 write</th>\n",
              "      <th>input reuse factor</th>\n",
              "      <th>filter l1 read</th>\n",
              "      <th>filter l1 write</th>\n",
              "      <th>filter l2 read</th>\n",
              "      <th>filter l2 write</th>\n",
              "      <th>filter reuse factor</th>\n",
              "      <th>output l1 read</th>\n",
              "      <th>output l1 write</th>\n",
              "      <th>output l2 read</th>\n",
              "      <th>output l2 write</th>\n",
              "      <th>output reuse factor</th>\n",
              "      <th>Ingress Delay (Min)</th>\n",
              "      <th>Ingress Delay (Max)</th>\n",
              "      <th>Ingress Delay (Avg)</th>\n",
              "      <th>Egress Delay (Min)</th>\n",
              "      <th>Egress Delay (Max)</th>\n",
              "      <th>Egress Delay (Avg)</th>\n",
              "      <th>Compute Delay (Min)</th>\n",
              "      <th>Compute Delay (Min)</th>\n",
              "      <th>Compute Delay (Avg)</th>\n",
              "      <th>Avg number of utilized PEs</th>\n",
              "      <th>Arithmetic Intensity</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV1</td>\n",
              "      <td>512</td>\n",
              "      <td>4562305</td>\n",
              "      <td>1093990000</td>\n",
              "      <td>24.5000</td>\n",
              "      <td>22.3951</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192924000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>4.77374</td>\n",
              "      <td>49</td>\n",
              "      <td>9</td>\n",
              "      <td>590</td>\n",
              "      <td>294</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>150528</td>\n",
              "      <td>9408</td>\n",
              "      <td>102173000</td>\n",
              "      <td>111776448</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>565.248</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>111776448</td>\n",
              "      <td>21772947</td>\n",
              "      <td>21772947</td>\n",
              "      <td>150528</td>\n",
              "      <td>5.13373</td>\n",
              "      <td>111776448</td>\n",
              "      <td>6321</td>\n",
              "      <td>6321</td>\n",
              "      <td>9408</td>\n",
              "      <td>17683.30000</td>\n",
              "      <td>111776448</td>\n",
              "      <td>111776448</td>\n",
              "      <td>3041536</td>\n",
              "      <td>760384</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>3</td>\n",
              "      <td>5.25000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV2_1_1</td>\n",
              "      <td>512</td>\n",
              "      <td>200705</td>\n",
              "      <td>316780000</td>\n",
              "      <td>63.9997</td>\n",
              "      <td>202.0320</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>63.00030</td>\n",
              "      <td>128</td>\n",
              "      <td>9</td>\n",
              "      <td>258</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>200704</td>\n",
              "      <td>4096</td>\n",
              "      <td>40548800</td>\n",
              "      <td>12845056</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>12845056</td>\n",
              "      <td>12640384</td>\n",
              "      <td>12640384</td>\n",
              "      <td>200704</td>\n",
              "      <td>1.01619</td>\n",
              "      <td>12845056</td>\n",
              "      <td>4033</td>\n",
              "      <td>4033</td>\n",
              "      <td>4096</td>\n",
              "      <td>3184.99000</td>\n",
              "      <td>12845056</td>\n",
              "      <td>12845056</td>\n",
              "      <td>200704</td>\n",
              "      <td>200704</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>1.01587</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV2_1_2</td>\n",
              "      <td>512</td>\n",
              "      <td>186625</td>\n",
              "      <td>1331090000</td>\n",
              "      <td>575.9970</td>\n",
              "      <td>432.7240</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>196.00300</td>\n",
              "      <td>1152</td>\n",
              "      <td>9</td>\n",
              "      <td>2306</td>\n",
              "      <td>54</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>200704</td>\n",
              "      <td>36864</td>\n",
              "      <td>80757200</td>\n",
              "      <td>107495424</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>107495424</td>\n",
              "      <td>36542592</td>\n",
              "      <td>36542592</td>\n",
              "      <td>200704</td>\n",
              "      <td>2.94165</td>\n",
              "      <td>107495424</td>\n",
              "      <td>36297</td>\n",
              "      <td>36297</td>\n",
              "      <td>36864</td>\n",
              "      <td>2961.55000</td>\n",
              "      <td>107495424</td>\n",
              "      <td>107495424</td>\n",
              "      <td>186624</td>\n",
              "      <td>186624</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>3.04762</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV2_1_3</td>\n",
              "      <td>512</td>\n",
              "      <td>802817</td>\n",
              "      <td>1256990000</td>\n",
              "      <td>63.9999</td>\n",
              "      <td>50.9153</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>63.00010</td>\n",
              "      <td>128</td>\n",
              "      <td>9</td>\n",
              "      <td>258</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>200704</td>\n",
              "      <td>16384</td>\n",
              "      <td>40875700</td>\n",
              "      <td>51380224</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>50561344</td>\n",
              "      <td>50561344</td>\n",
              "      <td>200704</td>\n",
              "      <td>1.01620</td>\n",
              "      <td>51380224</td>\n",
              "      <td>16129</td>\n",
              "      <td>16129</td>\n",
              "      <td>16384</td>\n",
              "      <td>3185.58000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51380224</td>\n",
              "      <td>802816</td>\n",
              "      <td>802816</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>1.01587</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV2_1_Residual</td>\n",
              "      <td>512</td>\n",
              "      <td>3137</td>\n",
              "      <td>46513400</td>\n",
              "      <td>255.9180</td>\n",
              "      <td>5502.0400</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>255.08200</td>\n",
              "      <td>512</td>\n",
              "      <td>9</td>\n",
              "      <td>1534</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>802816</td>\n",
              "      <td>256</td>\n",
              "      <td>17259900</td>\n",
              "      <td>802816</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>802816</td>\n",
              "      <td>799681</td>\n",
              "      <td>799681</td>\n",
              "      <td>802816</td>\n",
              "      <td>1.00392</td>\n",
              "      <td>802816</td>\n",
              "      <td>256</td>\n",
              "      <td>256</td>\n",
              "      <td>256</td>\n",
              "      <td>3136.00000</td>\n",
              "      <td>802816</td>\n",
              "      <td>802816</td>\n",
              "      <td>802816</td>\n",
              "      <td>799680</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>256</td>\n",
              "      <td>1.00392</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>61</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV5_3_1</td>\n",
              "      <td>512</td>\n",
              "      <td>100353</td>\n",
              "      <td>1982400000</td>\n",
              "      <td>511.9950</td>\n",
              "      <td>258.2710</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>896.00500</td>\n",
              "      <td>1024</td>\n",
              "      <td>9</td>\n",
              "      <td>2050</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>1048576</td>\n",
              "      <td>25918200</td>\n",
              "      <td>51380224</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51118592</td>\n",
              "      <td>51118592</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.00512</td>\n",
              "      <td>51380224</td>\n",
              "      <td>38797312</td>\n",
              "      <td>38797312</td>\n",
              "      <td>1048576</td>\n",
              "      <td>1.32432</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51380224</td>\n",
              "      <td>25088</td>\n",
              "      <td>100352</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>0.50000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>62</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV5_3_2</td>\n",
              "      <td>512</td>\n",
              "      <td>12801</td>\n",
              "      <td>903595000</td>\n",
              "      <td>4607.6400</td>\n",
              "      <td>5099.2300</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>2150.76000</td>\n",
              "      <td>9216</td>\n",
              "      <td>9</td>\n",
              "      <td>18434</td>\n",
              "      <td>54</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>25088</td>\n",
              "      <td>2359296</td>\n",
              "      <td>65275300</td>\n",
              "      <td>58982400</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>58982400</td>\n",
              "      <td>25170432</td>\n",
              "      <td>25170432</td>\n",
              "      <td>25088</td>\n",
              "      <td>2.34332</td>\n",
              "      <td>58982400</td>\n",
              "      <td>2359296</td>\n",
              "      <td>2359296</td>\n",
              "      <td>2359296</td>\n",
              "      <td>25.00000</td>\n",
              "      <td>58982400</td>\n",
              "      <td>58982400</td>\n",
              "      <td>12800</td>\n",
              "      <td>12800</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>3.00000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>63</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV5_3_3</td>\n",
              "      <td>512</td>\n",
              "      <td>100353</td>\n",
              "      <td>1274310000</td>\n",
              "      <td>511.9950</td>\n",
              "      <td>401.7810</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>512.00500</td>\n",
              "      <td>1024</td>\n",
              "      <td>9</td>\n",
              "      <td>2050</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>25088</td>\n",
              "      <td>1048576</td>\n",
              "      <td>40319900</td>\n",
              "      <td>51380224</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>50332160</td>\n",
              "      <td>50332160</td>\n",
              "      <td>25088</td>\n",
              "      <td>1.02082</td>\n",
              "      <td>51380224</td>\n",
              "      <td>1048576</td>\n",
              "      <td>1048576</td>\n",
              "      <td>1048576</td>\n",
              "      <td>49.00000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51380224</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>1.00000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>64</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>CONV5_3_Residual</td>\n",
              "      <td>512</td>\n",
              "      <td>197</td>\n",
              "      <td>7251180</td>\n",
              "      <td>509.4010</td>\n",
              "      <td>70250.8000</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>898.61200</td>\n",
              "      <td>1024</td>\n",
              "      <td>9</td>\n",
              "      <td>3072</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>2048</td>\n",
              "      <td>13839400</td>\n",
              "      <td>100352</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.00000</td>\n",
              "      <td>100352</td>\n",
              "      <td>75776</td>\n",
              "      <td>75776</td>\n",
              "      <td>2048</td>\n",
              "      <td>1.32432</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>0.50000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>65</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>FC1000</td>\n",
              "      <td>512</td>\n",
              "      <td>24001</td>\n",
              "      <td>5521280000</td>\n",
              "      <td>4181.1600</td>\n",
              "      <td>757.2800</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192924000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>7318.38000</td>\n",
              "      <td>8362</td>\n",
              "      <td>9</td>\n",
              "      <td>100354</td>\n",
              "      <td>294</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352000</td>\n",
              "      <td>18175500</td>\n",
              "      <td>100352000</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>565.248</td>\n",
              "      <td>16</td>\n",
              "      <td>192918000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>75289088</td>\n",
              "      <td>75289088</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.33289</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>1.00000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>1000</td>\n",
              "      <td>4000</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>512</td>\n",
              "      <td>0.50000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>66 rows × 51 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "   Neural Network Name  ...  Arithmetic Intensity\n",
              "0             Resnet50  ...               5.25000\n",
              "1             Resnet50  ...               1.01587\n",
              "2             Resnet50  ...               3.04762\n",
              "3             Resnet50  ...               1.01587\n",
              "4             Resnet50  ...               1.00392\n",
              "..                 ...  ...                   ...\n",
              "61            Resnet50  ...               0.50000\n",
              "62            Resnet50  ...               3.00000\n",
              "63            Resnet50  ...               1.00000\n",
              "64            Resnet50  ...               0.50000\n",
              "65            Resnet50  ...               0.50000\n",
              "\n",
              "[66 rows x 51 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GYS2j1xOcWHV",
        "outputId": "478583a5-09e8-4ec0-9878-b6d878d2120d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 545
        }
      },
      "source": [
        "df.set_index(' Layer Number')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Neural Network Name</th>\n",
              "      <th>NumPEs</th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <th>Throughput (MACs/Cycle)</th>\n",
              "      <th>Throughput Per Energy (GMACs/s*J)</th>\n",
              "      <th>Area</th>\n",
              "      <th>Power</th>\n",
              "      <th>NoC BW (Elements/cycle)</th>\n",
              "      <th>Avg BW Req</th>\n",
              "      <th>Peak BW Req</th>\n",
              "      <th>Vector Width</th>\n",
              "      <th>L2 SRAM Size (Bytes)</th>\n",
              "      <th>L1 SRAM Size (Bytes)</th>\n",
              "      <th>Multicasting Factor (Weight)</th>\n",
              "      <th>Multicasting Factor (Input)</th>\n",
              "      <th>Num Total Input Pixels</th>\n",
              "      <th>Num Total Weight Pixels</th>\n",
              "      <th>Ops/J</th>\n",
              "      <th>Num MACs</th>\n",
              "      <th>PE power</th>\n",
              "      <th>L1 power</th>\n",
              "      <th>L2 power</th>\n",
              "      <th>NOC power</th>\n",
              "      <th>input l1 read</th>\n",
              "      <th>input l1 write</th>\n",
              "      <th>input l2 read</th>\n",
              "      <th>input l2 write</th>\n",
              "      <th>input reuse factor</th>\n",
              "      <th>filter l1 read</th>\n",
              "      <th>filter l1 write</th>\n",
              "      <th>filter l2 read</th>\n",
              "      <th>filter l2 write</th>\n",
              "      <th>filter reuse factor</th>\n",
              "      <th>output l1 read</th>\n",
              "      <th>output l1 write</th>\n",
              "      <th>output l2 read</th>\n",
              "      <th>output l2 write</th>\n",
              "      <th>output reuse factor</th>\n",
              "      <th>Ingress Delay (Min)</th>\n",
              "      <th>Ingress Delay (Max)</th>\n",
              "      <th>Ingress Delay (Avg)</th>\n",
              "      <th>Egress Delay (Min)</th>\n",
              "      <th>Egress Delay (Max)</th>\n",
              "      <th>Egress Delay (Avg)</th>\n",
              "      <th>Compute Delay (Min)</th>\n",
              "      <th>Compute Delay (Min)</th>\n",
              "      <th>Compute Delay (Avg)</th>\n",
              "      <th>Avg number of utilized PEs</th>\n",
              "      <th>Arithmetic Intensity</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>CONV1</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>4562305</td>\n",
              "      <td>1093990000</td>\n",
              "      <td>24.5000</td>\n",
              "      <td>22.3951</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192924000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>4.77374</td>\n",
              "      <td>49</td>\n",
              "      <td>9</td>\n",
              "      <td>590</td>\n",
              "      <td>294</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>150528</td>\n",
              "      <td>9408</td>\n",
              "      <td>102173000</td>\n",
              "      <td>111776448</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>565.248</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>111776448</td>\n",
              "      <td>21772947</td>\n",
              "      <td>21772947</td>\n",
              "      <td>150528</td>\n",
              "      <td>5.13373</td>\n",
              "      <td>111776448</td>\n",
              "      <td>6321</td>\n",
              "      <td>6321</td>\n",
              "      <td>9408</td>\n",
              "      <td>17683.30000</td>\n",
              "      <td>111776448</td>\n",
              "      <td>111776448</td>\n",
              "      <td>3041536</td>\n",
              "      <td>760384</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>3</td>\n",
              "      <td>5.25000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_1_1</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>200705</td>\n",
              "      <td>316780000</td>\n",
              "      <td>63.9997</td>\n",
              "      <td>202.0320</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>63.00030</td>\n",
              "      <td>128</td>\n",
              "      <td>9</td>\n",
              "      <td>258</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>200704</td>\n",
              "      <td>4096</td>\n",
              "      <td>40548800</td>\n",
              "      <td>12845056</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>12845056</td>\n",
              "      <td>12640384</td>\n",
              "      <td>12640384</td>\n",
              "      <td>200704</td>\n",
              "      <td>1.01619</td>\n",
              "      <td>12845056</td>\n",
              "      <td>4033</td>\n",
              "      <td>4033</td>\n",
              "      <td>4096</td>\n",
              "      <td>3184.99000</td>\n",
              "      <td>12845056</td>\n",
              "      <td>12845056</td>\n",
              "      <td>200704</td>\n",
              "      <td>200704</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>1.01587</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_1_2</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>186625</td>\n",
              "      <td>1331090000</td>\n",
              "      <td>575.9970</td>\n",
              "      <td>432.7240</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>196.00300</td>\n",
              "      <td>1152</td>\n",
              "      <td>9</td>\n",
              "      <td>2306</td>\n",
              "      <td>54</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>200704</td>\n",
              "      <td>36864</td>\n",
              "      <td>80757200</td>\n",
              "      <td>107495424</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>107495424</td>\n",
              "      <td>36542592</td>\n",
              "      <td>36542592</td>\n",
              "      <td>200704</td>\n",
              "      <td>2.94165</td>\n",
              "      <td>107495424</td>\n",
              "      <td>36297</td>\n",
              "      <td>36297</td>\n",
              "      <td>36864</td>\n",
              "      <td>2961.55000</td>\n",
              "      <td>107495424</td>\n",
              "      <td>107495424</td>\n",
              "      <td>186624</td>\n",
              "      <td>186624</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>3.04762</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_1_3</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>802817</td>\n",
              "      <td>1256990000</td>\n",
              "      <td>63.9999</td>\n",
              "      <td>50.9153</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>63.00010</td>\n",
              "      <td>128</td>\n",
              "      <td>9</td>\n",
              "      <td>258</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>200704</td>\n",
              "      <td>16384</td>\n",
              "      <td>40875700</td>\n",
              "      <td>51380224</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>50561344</td>\n",
              "      <td>50561344</td>\n",
              "      <td>200704</td>\n",
              "      <td>1.01620</td>\n",
              "      <td>51380224</td>\n",
              "      <td>16129</td>\n",
              "      <td>16129</td>\n",
              "      <td>16384</td>\n",
              "      <td>3185.58000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51380224</td>\n",
              "      <td>802816</td>\n",
              "      <td>802816</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>1.01587</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_1_Residual</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>3137</td>\n",
              "      <td>46513400</td>\n",
              "      <td>255.9180</td>\n",
              "      <td>5502.0400</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>255.08200</td>\n",
              "      <td>512</td>\n",
              "      <td>9</td>\n",
              "      <td>1534</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>802816</td>\n",
              "      <td>256</td>\n",
              "      <td>17259900</td>\n",
              "      <td>802816</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>802816</td>\n",
              "      <td>799681</td>\n",
              "      <td>799681</td>\n",
              "      <td>802816</td>\n",
              "      <td>1.00392</td>\n",
              "      <td>802816</td>\n",
              "      <td>256</td>\n",
              "      <td>256</td>\n",
              "      <td>256</td>\n",
              "      <td>3136.00000</td>\n",
              "      <td>802816</td>\n",
              "      <td>802816</td>\n",
              "      <td>802816</td>\n",
              "      <td>799680</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>256</td>\n",
              "      <td>1.00392</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_1</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>100353</td>\n",
              "      <td>1982400000</td>\n",
              "      <td>511.9950</td>\n",
              "      <td>258.2710</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>896.00500</td>\n",
              "      <td>1024</td>\n",
              "      <td>9</td>\n",
              "      <td>2050</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>1048576</td>\n",
              "      <td>25918200</td>\n",
              "      <td>51380224</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51118592</td>\n",
              "      <td>51118592</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.00512</td>\n",
              "      <td>51380224</td>\n",
              "      <td>38797312</td>\n",
              "      <td>38797312</td>\n",
              "      <td>1048576</td>\n",
              "      <td>1.32432</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51380224</td>\n",
              "      <td>25088</td>\n",
              "      <td>100352</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>0.50000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_2</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>12801</td>\n",
              "      <td>903595000</td>\n",
              "      <td>4607.6400</td>\n",
              "      <td>5099.2300</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>2150.76000</td>\n",
              "      <td>9216</td>\n",
              "      <td>9</td>\n",
              "      <td>18434</td>\n",
              "      <td>54</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>25088</td>\n",
              "      <td>2359296</td>\n",
              "      <td>65275300</td>\n",
              "      <td>58982400</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>58982400</td>\n",
              "      <td>25170432</td>\n",
              "      <td>25170432</td>\n",
              "      <td>25088</td>\n",
              "      <td>2.34332</td>\n",
              "      <td>58982400</td>\n",
              "      <td>2359296</td>\n",
              "      <td>2359296</td>\n",
              "      <td>2359296</td>\n",
              "      <td>25.00000</td>\n",
              "      <td>58982400</td>\n",
              "      <td>58982400</td>\n",
              "      <td>12800</td>\n",
              "      <td>12800</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>3.00000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_3</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>100353</td>\n",
              "      <td>1274310000</td>\n",
              "      <td>511.9950</td>\n",
              "      <td>401.7810</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>512.00500</td>\n",
              "      <td>1024</td>\n",
              "      <td>9</td>\n",
              "      <td>2050</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>25088</td>\n",
              "      <td>1048576</td>\n",
              "      <td>40319900</td>\n",
              "      <td>51380224</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>50332160</td>\n",
              "      <td>50332160</td>\n",
              "      <td>25088</td>\n",
              "      <td>1.02082</td>\n",
              "      <td>51380224</td>\n",
              "      <td>1048576</td>\n",
              "      <td>1048576</td>\n",
              "      <td>1048576</td>\n",
              "      <td>49.00000</td>\n",
              "      <td>51380224</td>\n",
              "      <td>51380224</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>1.00000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_Residual</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>197</td>\n",
              "      <td>7251180</td>\n",
              "      <td>509.4010</td>\n",
              "      <td>70250.8000</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192923000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>898.61200</td>\n",
              "      <td>1024</td>\n",
              "      <td>9</td>\n",
              "      <td>3072</td>\n",
              "      <td>6</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>2048</td>\n",
              "      <td>13839400</td>\n",
              "      <td>100352</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>113.050</td>\n",
              "      <td>4</td>\n",
              "      <td>192918000</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.00000</td>\n",
              "      <td>100352</td>\n",
              "      <td>75776</td>\n",
              "      <td>75776</td>\n",
              "      <td>2048</td>\n",
              "      <td>1.32432</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>512</td>\n",
              "      <td>0.50000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>FC1000</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>24001</td>\n",
              "      <td>5521280000</td>\n",
              "      <td>4181.1600</td>\n",
              "      <td>757.2800</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192924000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>7318.38000</td>\n",
              "      <td>8362</td>\n",
              "      <td>9</td>\n",
              "      <td>100354</td>\n",
              "      <td>294</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352000</td>\n",
              "      <td>18175500</td>\n",
              "      <td>100352000</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>565.248</td>\n",
              "      <td>16</td>\n",
              "      <td>192918000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>75289088</td>\n",
              "      <td>75289088</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.33289</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>1.00000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>1000</td>\n",
              "      <td>4000</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>512</td>\n",
              "      <td>0.50000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>66 rows × 50 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                 Neural Network Name  ...   Arithmetic Intensity\n",
              " Layer Number                         ...                       \n",
              "CONV1                       Resnet50  ...                5.25000\n",
              "CONV2_1_1                   Resnet50  ...                1.01587\n",
              "CONV2_1_2                   Resnet50  ...                3.04762\n",
              "CONV2_1_3                   Resnet50  ...                1.01587\n",
              "CONV2_1_Residual            Resnet50  ...                1.00392\n",
              "...                              ...  ...                    ...\n",
              "CONV5_3_1                   Resnet50  ...                0.50000\n",
              "CONV5_3_2                   Resnet50  ...                3.00000\n",
              "CONV5_3_3                   Resnet50  ...                1.00000\n",
              "CONV5_3_Residual            Resnet50  ...                0.50000\n",
              "FC1000                      Resnet50  ...                0.50000\n",
              "\n",
              "[66 rows x 50 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 58
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S-d0JQ9LXKTh"
      },
      "source": [
        "fully_connected = df.loc[df[' Layer Number'] == 'FC1000']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "keNSapXlZN4_",
        "outputId": "06b9dc11-ce34-43f3-8c5e-822c93227438",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 219
        }
      },
      "source": [
        "df.set_index(' Layer Number').filter(like='FC', axis=0)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Neural Network Name</th>\n",
              "      <th>NumPEs</th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "      <th>Throughput (MACs/Cycle)</th>\n",
              "      <th>Throughput Per Energy (GMACs/s*J)</th>\n",
              "      <th>Area</th>\n",
              "      <th>Power</th>\n",
              "      <th>NoC BW (Elements/cycle)</th>\n",
              "      <th>Avg BW Req</th>\n",
              "      <th>Peak BW Req</th>\n",
              "      <th>Vector Width</th>\n",
              "      <th>L2 SRAM Size (Bytes)</th>\n",
              "      <th>L1 SRAM Size (Bytes)</th>\n",
              "      <th>Multicasting Factor (Weight)</th>\n",
              "      <th>Multicasting Factor (Input)</th>\n",
              "      <th>Num Total Input Pixels</th>\n",
              "      <th>Num Total Weight Pixels</th>\n",
              "      <th>Ops/J</th>\n",
              "      <th>Num MACs</th>\n",
              "      <th>PE power</th>\n",
              "      <th>L1 power</th>\n",
              "      <th>L2 power</th>\n",
              "      <th>NOC power</th>\n",
              "      <th>input l1 read</th>\n",
              "      <th>input l1 write</th>\n",
              "      <th>input l2 read</th>\n",
              "      <th>input l2 write</th>\n",
              "      <th>input reuse factor</th>\n",
              "      <th>filter l1 read</th>\n",
              "      <th>filter l1 write</th>\n",
              "      <th>filter l2 read</th>\n",
              "      <th>filter l2 write</th>\n",
              "      <th>filter reuse factor</th>\n",
              "      <th>output l1 read</th>\n",
              "      <th>output l1 write</th>\n",
              "      <th>output l2 read</th>\n",
              "      <th>output l2 write</th>\n",
              "      <th>output reuse factor</th>\n",
              "      <th>Ingress Delay (Min)</th>\n",
              "      <th>Ingress Delay (Max)</th>\n",
              "      <th>Ingress Delay (Avg)</th>\n",
              "      <th>Egress Delay (Min)</th>\n",
              "      <th>Egress Delay (Max)</th>\n",
              "      <th>Egress Delay (Avg)</th>\n",
              "      <th>Compute Delay (Min)</th>\n",
              "      <th>Compute Delay (Min)</th>\n",
              "      <th>Compute Delay (Avg)</th>\n",
              "      <th>Avg number of utilized PEs</th>\n",
              "      <th>Arithmetic Intensity</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>FC1000</th>\n",
              "      <td>Resnet50</td>\n",
              "      <td>512</td>\n",
              "      <td>24001</td>\n",
              "      <td>5521280000</td>\n",
              "      <td>4181.16</td>\n",
              "      <td>757.28</td>\n",
              "      <td>24547400000000</td>\n",
              "      <td>192924000</td>\n",
              "      <td>32000000</td>\n",
              "      <td>7318.38</td>\n",
              "      <td>8362</td>\n",
              "      <td>9</td>\n",
              "      <td>100354</td>\n",
              "      <td>294</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>100352</td>\n",
              "      <td>100352000</td>\n",
              "      <td>18175500</td>\n",
              "      <td>100352000</td>\n",
              "      <td>5632.36</td>\n",
              "      <td>565.248</td>\n",
              "      <td>16</td>\n",
              "      <td>192918000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>75289088</td>\n",
              "      <td>75289088</td>\n",
              "      <td>100352</td>\n",
              "      <td>1.33289</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>1.0</td>\n",
              "      <td>100352000</td>\n",
              "      <td>100352000</td>\n",
              "      <td>1000</td>\n",
              "      <td>4000</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>6</td>\n",
              "      <td>512</td>\n",
              "      <td>0.5</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "              Neural Network Name  ...   Arithmetic Intensity\n",
              " Layer Number                      ...                       \n",
              "FC1000                   Resnet50  ...                    0.5\n",
              "\n",
              "[1 rows x 50 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 54
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4cd_JUoLYezx",
        "outputId": "53ea3f14-1805-4c9c-ae65-bc787b0931df",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 583
        }
      },
      "source": [
        "x = df.filter(like='_Residual', axis=0)\n",
        "x\n",
        "x.loc[x[' Layer Number'] not in ]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>CONV2_1_Residual</th>\n",
              "      <td>3137</td>\n",
              "      <td>46513400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_2_Residual</th>\n",
              "      <td>3137</td>\n",
              "      <td>46513400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_3_Residual</th>\n",
              "      <td>3137</td>\n",
              "      <td>46513400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_1_Residual</th>\n",
              "      <td>785</td>\n",
              "      <td>23325300</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_1_Residual</th>\n",
              "      <td>785</td>\n",
              "      <td>23325300</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_3_Residual</th>\n",
              "      <td>785</td>\n",
              "      <td>23325300</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_4_Residual</th>\n",
              "      <td>785</td>\n",
              "      <td>23325300</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_1_Residual</th>\n",
              "      <td>393</td>\n",
              "      <td>13521000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_2_Residual</th>\n",
              "      <td>393</td>\n",
              "      <td>13521000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_3_Residual</th>\n",
              "      <td>393</td>\n",
              "      <td>13521000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_4_Residual</th>\n",
              "      <td>393</td>\n",
              "      <td>13521000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_5_Residual</th>\n",
              "      <td>393</td>\n",
              "      <td>13521000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_6_Residual</th>\n",
              "      <td>393</td>\n",
              "      <td>13521000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_1_Residual</th>\n",
              "      <td>197</td>\n",
              "      <td>7251180</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_2_Residual</th>\n",
              "      <td>197</td>\n",
              "      <td>7251180</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_Residual</th>\n",
              "      <td>197</td>\n",
              "      <td>7251180</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                   Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                         \n",
              "CONV2_1_Residual               3137                           46513400\n",
              "CONV2_2_Residual               3137                           46513400\n",
              "CONV2_3_Residual               3137                           46513400\n",
              "CONV3_1_Residual                785                           23325300\n",
              "CONV3_1_Residual                785                           23325300\n",
              "CONV3_3_Residual                785                           23325300\n",
              "CONV3_4_Residual                785                           23325300\n",
              "CONV4_1_Residual                393                           13521000\n",
              "CONV4_2_Residual                393                           13521000\n",
              "CONV4_3_Residual                393                           13521000\n",
              "CONV4_4_Residual                393                           13521000\n",
              "CONV4_5_Residual                393                           13521000\n",
              "CONV4_6_Residual                393                           13521000\n",
              "CONV5_1_Residual                197                            7251180\n",
              "CONV5_2_Residual                197                            7251180\n",
              "CONV5_3_Residual                197                            7251180"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 80
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EBku1quHYtDn",
        "outputId": "c5dbfab9-f5ff-4a7d-e006-b8a0b0245925",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 583
        }
      },
      "source": [
        "df.set_index(' Layer Number').filter(regex='CONV._._1', axis=0)[[' Runtime (Cycles)',  ' Activity count-based Energy (nJ)']]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>CONV2_1_1</th>\n",
              "      <td>200705</td>\n",
              "      <td>316780000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_2_1</th>\n",
              "      <td>200705</td>\n",
              "      <td>1268030000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_3_1</th>\n",
              "      <td>200705</td>\n",
              "      <td>1268030000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_1_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>641186000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_2_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>631057000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_3_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>631057000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_4_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>631057000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_1_1</th>\n",
              "      <td>50177</td>\n",
              "      <td>637088000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_2_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1735250000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_3_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1735250000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_4_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1735250000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_5_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1735250000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_6_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1735250000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_1_1</th>\n",
              "      <td>50177</td>\n",
              "      <td>875933000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_2_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1982400000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_1</th>\n",
              "      <td>100353</td>\n",
              "      <td>1982400000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                      \n",
              "CONV2_1_1                 200705                          316780000\n",
              "CONV2_2_1                 200705                         1268030000\n",
              "CONV2_3_1                 200705                         1268030000\n",
              "CONV3_1_1                 100353                          641186000\n",
              "CONV3_2_1                 100353                          631057000\n",
              "CONV3_3_1                 100353                          631057000\n",
              "CONV3_4_1                 100353                          631057000\n",
              "CONV4_1_1                  50177                          637088000\n",
              "CONV4_2_1                 100353                         1735250000\n",
              "CONV4_3_1                 100353                         1735250000\n",
              "CONV4_4_1                 100353                         1735250000\n",
              "CONV4_5_1                 100353                         1735250000\n",
              "CONV4_6_1                 100353                         1735250000\n",
              "CONV5_1_1                  50177                          875933000\n",
              "CONV5_2_1                 100353                         1982400000\n",
              "CONV5_3_1                 100353                         1982400000"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 63
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e53g-TUAb5RZ",
        "outputId": "b6a75b06-3ed7-47a2-fb56-9f573158a21a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "df.filter(regex='CONV._._[234]', axis=0)[[' Runtime (Cycles)',  ' Activity count-based Energy (nJ)']]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>CONV2_1_2</th>\n",
              "      <td>186625</td>\n",
              "      <td>1331090000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_1_3</th>\n",
              "      <td>802817</td>\n",
              "      <td>1256990000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_2_2</th>\n",
              "      <td>186625</td>\n",
              "      <td>1331090000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_2_3</th>\n",
              "      <td>802817</td>\n",
              "      <td>1256990000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_3_2</th>\n",
              "      <td>186625</td>\n",
              "      <td>1331090000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2_3_3</th>\n",
              "      <td>802817</td>\n",
              "      <td>1256990000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_1_2</th>\n",
              "      <td>86529</td>\n",
              "      <td>1262400000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_1_3</th>\n",
              "      <td>401409</td>\n",
              "      <td>1256740000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_2_2</th>\n",
              "      <td>86529</td>\n",
              "      <td>1262400000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_2_3</th>\n",
              "      <td>401409</td>\n",
              "      <td>1256740000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_3_2</th>\n",
              "      <td>86529</td>\n",
              "      <td>1262400000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_3_3</th>\n",
              "      <td>401409</td>\n",
              "      <td>1256740000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_4_2</th>\n",
              "      <td>86529</td>\n",
              "      <td>1262400000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3_4_3</th>\n",
              "      <td>401409</td>\n",
              "      <td>1256740000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_1_2</th>\n",
              "      <td>36865</td>\n",
              "      <td>1130880000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_1_3</th>\n",
              "      <td>200705</td>\n",
              "      <td>1259510000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_2_2</th>\n",
              "      <td>36865</td>\n",
              "      <td>1130880000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_2_3</th>\n",
              "      <td>200705</td>\n",
              "      <td>1259510000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_3_2</th>\n",
              "      <td>36865</td>\n",
              "      <td>1130880000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_3_3</th>\n",
              "      <td>200705</td>\n",
              "      <td>1259510000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_4_2</th>\n",
              "      <td>36865</td>\n",
              "      <td>1130880000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_4_3</th>\n",
              "      <td>200705</td>\n",
              "      <td>1259510000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_5_2</th>\n",
              "      <td>36865</td>\n",
              "      <td>1130880000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_5_3</th>\n",
              "      <td>200705</td>\n",
              "      <td>1259510000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_6_2</th>\n",
              "      <td>36865</td>\n",
              "      <td>1130880000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4_6_3</th>\n",
              "      <td>200705</td>\n",
              "      <td>1259510000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_1_2</th>\n",
              "      <td>12801</td>\n",
              "      <td>903595000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_1_3</th>\n",
              "      <td>100353</td>\n",
              "      <td>1274310000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_2_2</th>\n",
              "      <td>12801</td>\n",
              "      <td>903595000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_2_3</th>\n",
              "      <td>100353</td>\n",
              "      <td>1274310000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_2</th>\n",
              "      <td>12801</td>\n",
              "      <td>903595000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5_3_3</th>\n",
              "      <td>100353</td>\n",
              "      <td>1274310000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                      \n",
              "CONV2_1_2                 186625                         1331090000\n",
              "CONV2_1_3                 802817                         1256990000\n",
              "CONV2_2_2                 186625                         1331090000\n",
              "CONV2_2_3                 802817                         1256990000\n",
              "CONV2_3_2                 186625                         1331090000\n",
              "CONV2_3_3                 802817                         1256990000\n",
              "CONV3_1_2                  86529                         1262400000\n",
              "CONV3_1_3                 401409                         1256740000\n",
              "CONV3_2_2                  86529                         1262400000\n",
              "CONV3_2_3                 401409                         1256740000\n",
              "CONV3_3_2                  86529                         1262400000\n",
              "CONV3_3_3                 401409                         1256740000\n",
              "CONV3_4_2                  86529                         1262400000\n",
              "CONV3_4_3                 401409                         1256740000\n",
              "CONV4_1_2                  36865                         1130880000\n",
              "CONV4_1_3                 200705                         1259510000\n",
              "CONV4_2_2                  36865                         1130880000\n",
              "CONV4_2_3                 200705                         1259510000\n",
              "CONV4_3_2                  36865                         1130880000\n",
              "CONV4_3_3                 200705                         1259510000\n",
              "CONV4_4_2                  36865                         1130880000\n",
              "CONV4_4_3                 200705                         1259510000\n",
              "CONV4_5_2                  36865                         1130880000\n",
              "CONV4_5_3                 200705                         1259510000\n",
              "CONV4_6_2                  36865                         1130880000\n",
              "CONV4_6_3                 200705                         1259510000\n",
              "CONV5_1_2                  12801                          903595000\n",
              "CONV5_1_3                 100353                         1274310000\n",
              "CONV5_2_2                  12801                          903595000\n",
              "CONV5_2_3                 100353                         1274310000\n",
              "CONV5_3_2                  12801                          903595000\n",
              "CONV5_3_3                 100353                         1274310000"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 83
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "M3LQwfeOey9o",
        "outputId": "61f62e47-d250-42c1-e5c9-ddac303823e1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "df.filter(regex='CONV._._[234]', axis=0).sum(axis = 0)[1]/10**6"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "38716.855"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 92
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "diNqme99v9ar"
      },
      "source": [
        "## Scratchpad for VGG"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "chwOk4Suv8LT"
      },
      "source": [
        "vgg_csv_files = glob.glob(\"**/VGG/*.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cSGF7GLkwrrc",
        "outputId": "a6f2a412-48da-4307-c376-7286b903bdc7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "vgg_csv_files"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['content/VGG/vgg16_yxp_os.csv',\n",
              " 'content/VGG/vgg16_xp_ws.csv',\n",
              " 'content/VGG/vgg16_nlr.csv',\n",
              " 'content/VGG/vgg16_rs.csv']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yXHw8wMRJNd8",
        "outputId": "4c080a5f-4b06-4722-f429-edc127482e95",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 394
        }
      },
      "source": [
        "df.filter(regex='CONV([456789]|1[0-9])', axis=0)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>CONV4</th>\n",
              "      <td>9388248</td>\n",
              "      <td>1.184600e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5</th>\n",
              "      <td>4764716</td>\n",
              "      <td>1.428090e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV6</th>\n",
              "      <td>9531180</td>\n",
              "      <td>1.141830e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV7</th>\n",
              "      <td>9531180</td>\n",
              "      <td>1.686230e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV8</th>\n",
              "      <td>4621334</td>\n",
              "      <td>1.368780e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV9</th>\n",
              "      <td>9190934</td>\n",
              "      <td>1.061790e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV10</th>\n",
              "      <td>9190934</td>\n",
              "      <td>1.566550e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV11</th>\n",
              "      <td>2187786</td>\n",
              "      <td>1.135110e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV12</th>\n",
              "      <td>2187786</td>\n",
              "      <td>3.408010e+09</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV13</th>\n",
              "      <td>2187786</td>\n",
              "      <td>3.408010e+09</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                      \n",
              "CONV4                    9388248                       1.184600e+10\n",
              "CONV5                    4764716                       1.428090e+10\n",
              "CONV6                    9531180                       1.141830e+10\n",
              "CONV7                    9531180                       1.686230e+10\n",
              "CONV8                    4621334                       1.368780e+10\n",
              "CONV9                    9190934                       1.061790e+10\n",
              "CONV10                   9190934                       1.566550e+10\n",
              "CONV11                   2187786                       1.135110e+10\n",
              "CONV12                   2187786                       3.408010e+09\n",
              "CONV13                   2187786                       3.408010e+09"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_987dRowwton",
        "outputId": "289e6a70-a427-4c03-f741-2ae0c9b15600",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "runtimes = {}\n",
        "energies = {}\n",
        "for file in vgg_csv_files:\n",
        "    with open(file,'r') as f:\n",
        "        print(f'Filename: {file}')\n",
        "        df = pd.read_csv(f)\n",
        "        df = df.set_index(' Layer Number')\n",
        "        # print(df.head())\n",
        "        df = df[[' Runtime (Cycles)',  ' Activity count-based Energy (nJ)']]\n",
        "        fc = df.filter(like='FC', axis=0).sum(axis = 0)\n",
        "        \n",
        "        residual = df.filter(like='_Residual', axis=0).sum(axis = 0)\n",
        "        \n",
        "        point_wise = df.filter(regex='CONV._._1', axis=0).sum(axis = 0)\n",
        "        \n",
        "        early_layer = df.filter(regex='CONV[1-3]$', axis=0).sum(axis=0)\n",
        "        \n",
        "        late_layer = df.filter(regex='CONV([456789]|1[0-9])', axis=0).sum(axis = 0)\n",
        "        \n",
        "        layers = [fc,residual,point_wise,early_layer,late_layer]\n",
        "        filename = file.split('/')[2].split('.')[0]\n",
        "        runtimes[filename] = [x[0] for x in layers]  \n",
        "        energies[filename] = [x[1]/10**6 for x in layers]  \n",
        "\n",
        "        print(fc.head(),fc.count())\n",
        "        print(residual.head(),residual.count())\n",
        "        print(point_wise.head(),point_wise.count())\n",
        "        print(early_layer.head(),early_layer.count())\n",
        "        print(late_layer.head(),late_layer.count())\n",
        "        \n",
        "        print('\\n\\n\\n\\n\\n\\n')\n",
        "# print(runtimes)\n",
        "pd.DataFrame(runtimes,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Runtime(cycles)',colormap='Set1')\n",
        "pd.DataFrame(energies,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Energy(mJ)',colormap='Set2')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Filename: content/VGG/vgg16_yxp_os.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    2.943407e+07\n",
            " Activity count-based Energy (nJ)    3.359842e+10\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    1.182189e+08\n",
            " Activity count-based Energy (nJ)    1.600440e+11\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/VGG/vgg16_xp_ws.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    1.038702e+07\n",
            " Activity count-based Energy (nJ)    2.806511e+10\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    5.423896e+07\n",
            " Activity count-based Energy (nJ)    1.133989e+11\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/VGG/vgg16_nlr.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    3.171153e+07\n",
            " Activity count-based Energy (nJ)    3.402793e+10\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    1.224878e+08\n",
            " Activity count-based Energy (nJ)    1.394456e+11\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/VGG/vgg16_rs.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    1.511748e+07\n",
            " Activity count-based Energy (nJ)    2.199977e+10\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    6.278188e+07\n",
            " Activity count-based Energy (nJ)    1.125458e+11\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f02c4d92470>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mi5kIg_zw4-4",
        "outputId": "e529b6d9-62c9-495a-d650-7e1c0d56a13a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 489
        }
      },
      "source": [
        "df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>CONV1</th>\n",
              "      <td>447431</td>\n",
              "      <td>2.942000e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV2</th>\n",
              "      <td>9867527</td>\n",
              "      <td>7.025470e+09</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV3</th>\n",
              "      <td>4802520</td>\n",
              "      <td>1.468010e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV4</th>\n",
              "      <td>9388248</td>\n",
              "      <td>1.184600e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV5</th>\n",
              "      <td>4764716</td>\n",
              "      <td>1.428090e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV6</th>\n",
              "      <td>9531180</td>\n",
              "      <td>1.141830e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV7</th>\n",
              "      <td>9531180</td>\n",
              "      <td>1.686230e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV8</th>\n",
              "      <td>4621334</td>\n",
              "      <td>1.368780e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV9</th>\n",
              "      <td>9190934</td>\n",
              "      <td>1.061790e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV10</th>\n",
              "      <td>9190934</td>\n",
              "      <td>1.566550e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV11</th>\n",
              "      <td>2187786</td>\n",
              "      <td>1.135110e+10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV12</th>\n",
              "      <td>2187786</td>\n",
              "      <td>3.408010e+09</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>CONV13</th>\n",
              "      <td>2187786</td>\n",
              "      <td>3.408010e+09</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                      \n",
              "CONV1                     447431                       2.942000e+08\n",
              "CONV2                    9867527                       7.025470e+09\n",
              "CONV3                    4802520                       1.468010e+10\n",
              "CONV4                    9388248                       1.184600e+10\n",
              "CONV5                    4764716                       1.428090e+10\n",
              "CONV6                    9531180                       1.141830e+10\n",
              "CONV7                    9531180                       1.686230e+10\n",
              "CONV8                    4621334                       1.368780e+10\n",
              "CONV9                    9190934                       1.061790e+10\n",
              "CONV10                   9190934                       1.566550e+10\n",
              "CONV11                   2187786                       1.135110e+10\n",
              "CONV12                   2187786                       3.408010e+09\n",
              "CONV13                   2187786                       3.408010e+09"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 155
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KKtUR1u8C4Lo"
      },
      "source": [
        "## Squeezenet Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W14SrZCxC-74"
      },
      "source": [
        "squeezenet_csv_files = glob.glob(\"**/Squeezenet/*.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qRDYSnUeC-8D",
        "outputId": "82eb3074-2f8b-474a-d441-619ddc0edeca",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "squeezenet_csv_files"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['content/Squeezenet/squeezenet1_0_xp_ws.csv',\n",
              " 'content/Squeezenet/squeezenet1_0_rs.csv',\n",
              " 'content/Squeezenet/squeezenet1_0_kcp_ws.csv',\n",
              " 'content/Squeezenet/squeezenet1_0_ykp_os.csv']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 172
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0ntjv7P-C-8Q",
        "outputId": "a97d9b3f-d59c-4da3-e216-e00116525c3e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "runtimes = {}\n",
        "energies = {}\n",
        "for file in squeezenet_csv_files:\n",
        "    with open(file,'r') as f:\n",
        "        print(f'Filename: {file}')\n",
        "        df = pd.read_csv(f)\n",
        "        df = df.set_index(' Layer Number')\n",
        "        # print(df.head())\n",
        "        df = df[[' Runtime (Cycles)',  ' Activity count-based Energy (nJ)']]\n",
        "        fc = df.filter(like='FC', axis=0).sum(axis = 0)\n",
        "        \n",
        "        residual = df.filter(like='_Residual', axis=0).sum(axis = 0)\n",
        "        \n",
        "        point_wise = df.filter(regex='CONV._._1', axis=0).sum(axis = 0)\n",
        "        \n",
        "        early_layer = df.filter(regex='\\d?', axis=0).sum(axis = 0)\n",
        "        \n",
        "        late_layer = df.filter(regex='CONV._._[234]', axis=0).sum(axis = 0)\n",
        "        \n",
        "        layers = [fc,residual,point_wise,early_layer,late_layer]\n",
        "        filename = file.split('/')[2].split('.')[0]\n",
        "        runtimes[filename] = [x[0] for x in layers]  \n",
        "        energies[filename] = [x[1]/10**6 for x in layers]  \n",
        "\n",
        "        print(fc.head(),fc.count())\n",
        "        print(residual.head(),residual.count())\n",
        "        print(point_wise.head(),point_wise.count())\n",
        "        print(early_layer.head(),early_layer.count())\n",
        "        print(late_layer.head(),late_layer.count())\n",
        "        \n",
        "        print('\\n\\n\\n\\n\\n\\n')\n",
        "# print(runtimes)\n",
        "pd.DataFrame(runtimes,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Runtime(cycles)',colormap='Set1')\n",
        "pd.DataFrame(energies,index=['fc','residual','point_wise','early','late']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Energy(mJ)',colormap='Set2')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Filename: content/Squeezenet/squeezenet1_0_xp_ws.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    1.246738e+07\n",
            " Activity count-based Energy (nJ)    1.387753e+10\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/Squeezenet/squeezenet1_0_rs.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    3.100330e+06\n",
            " Activity count-based Energy (nJ)    9.393642e+09\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/Squeezenet/squeezenet1_0_kcp_ws.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    5.496319e+06\n",
            " Activity count-based Energy (nJ)    6.099731e+09\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/Squeezenet/squeezenet1_0_ykp_os.csv\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    2.545827e+06\n",
            " Activity count-based Energy (nJ)    1.036101e+10\n",
            "dtype: float64 2\n",
            " Runtime (Cycles)                    0.0\n",
            " Activity count-based Energy (nJ)    0.0\n",
            "dtype: float64 2\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc5512fff60>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 175
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KqNJhJrYC-8X",
        "outputId": "6d8ec4f4-236f-417a-fe61-a98a9c40ccd5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 896
        }
      },
      "source": [
        "df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>137347</td>\n",
              "      <td>4.099870e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>41473</td>\n",
              "      <td>1.126320e+09</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>27649</td>\n",
              "      <td>8.579810e+07</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>26626</td>\n",
              "      <td>1.226990e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>55297</td>\n",
              "      <td>2.769280e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>27649</td>\n",
              "      <td>9.517760e+07</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>26626</td>\n",
              "      <td>1.226990e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>110593</td>\n",
              "      <td>3.910950e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>110593</td>\n",
              "      <td>3.045800e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>106498</td>\n",
              "      <td>4.932920e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>55297</td>\n",
              "      <td>7.437280e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>27649</td>\n",
              "      <td>9.495570e+07</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>25602</td>\n",
              "      <td>1.170970e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>82945</td>\n",
              "      <td>3.194530e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>62209</td>\n",
              "      <td>1.853070e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>57602</td>\n",
              "      <td>2.639060e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>124417</td>\n",
              "      <td>5.879150e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>62209</td>\n",
              "      <td>2.134450e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>57602</td>\n",
              "      <td>2.639060e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>165889</td>\n",
              "      <td>6.736460e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>110593</td>\n",
              "      <td>3.417570e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>22</th>\n",
              "      <td>102402</td>\n",
              "      <td>4.695550e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23</th>\n",
              "      <td>53249</td>\n",
              "      <td>6.877790e+08</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24</th>\n",
              "      <td>26625</td>\n",
              "      <td>8.829240e+07</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25</th>\n",
              "      <td>22529</td>\n",
              "      <td>9.905230e+07</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>26</th>\n",
              "      <td>838657</td>\n",
              "      <td>1.782640e+09</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                      \n",
              "1                         137347                       4.099870e+08\n",
              "2                          41473                       1.126320e+09\n",
              "3                          27649                       8.579810e+07\n",
              "4                          26626                       1.226990e+08\n",
              "5                          55297                       2.769280e+08\n",
              "6                          27649                       9.517760e+07\n",
              "7                          26626                       1.226990e+08\n",
              "8                         110593                       3.910950e+08\n",
              "9                         110593                       3.045800e+08\n",
              "10                        106498                       4.932920e+08\n",
              "11                         55297                       7.437280e+08\n",
              "12                         27649                       9.495570e+07\n",
              "13                         25602                       1.170970e+08\n",
              "14                         82945                       3.194530e+08\n",
              "15                         62209                       1.853070e+08\n",
              "16                         57602                       2.639060e+08\n",
              "17                        124417                       5.879150e+08\n",
              "18                         62209                       2.134450e+08\n",
              "19                         57602                       2.639060e+08\n",
              "20                        165889                       6.736460e+08\n",
              "21                        110593                       3.417570e+08\n",
              "22                        102402                       4.695550e+08\n",
              "23                         53249                       6.877790e+08\n",
              "24                         26625                       8.829240e+07\n",
              "25                         22529                       9.905230e+07\n",
              "26                        838657                       1.782640e+09"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 174
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VF2DD_WpRrbn"
      },
      "source": [
        "## MobileNetV2 Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "83w5WaeRR3JQ"
      },
      "source": [
        "mbnet_csv_files = glob.glob(\"**/mbnet/*.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AJGGshQER3KI",
        "outputId": "bac54dd7-144b-45a0-dc40-7b291305b6f7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "mbnet_csv_files"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['content/mbnet/MobileNetV2_yxp_os.csv',\n",
              " 'content/mbnet/MobileNetV2_xp_ws.csv',\n",
              " 'content/mbnet/MobileNetV2_kcp_ws.csv']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 192
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Df-f_lbkRzhQ",
        "outputId": "ffd32623-95e5-44ee-a7f0-1ad43c3cea4a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "runtimes = {}\n",
        "energies = {}\n",
        "for file in mbnet_csv_files:\n",
        "    with open(file,'r') as f:\n",
        "        print(f'Filename: {file}')\n",
        "        df = pd.read_csv(f)\n",
        "        df = df.set_index(' Layer Number')\n",
        "        # print(df.head())\n",
        "        df = df[[' Runtime (Cycles)',  ' Activity count-based Energy (nJ)']]\n",
        "        # fc = df.filter(like='FC', axis=0).sum(axis = 0)\n",
        "        \n",
        "        residual = df.filter(like='_Residual', axis=0).sum(axis = 0)\n",
        "        \n",
        "        point_wise = df.filter(regex='Bottleneck._._1', axis=0).sum(axis = 0)\n",
        "        \n",
        "        early_layer = df.filter(regex='CONV1', axis=0).sum(axis = 0)\n",
        "        \n",
        "        late_layer = df.filter(regex='CONV[23]', axis=0).sum(axis = 0)\n",
        "\n",
        "        depth_wise = df.filter(regex='Bottleneck._._\\d',axis=0).sum(axis=0)\n",
        "        \n",
        "        layers = [fc,residual,point_wise,early_layer,late_layer,depth_wise]\n",
        "        filename = file.split('/')[2].split('.')[0]\n",
        "        runtimes[filename] = [x[0] for x in layers]  \n",
        "        energies[filename] = [x[1]/10**6 for x in layers]  \n",
        "\n",
        "        # print(fc.head(),fc.count())\n",
        "        # print(residual.head(),residual.count())\n",
        "        # print(point_wise.head(),point_wise.count())\n",
        "        # print(early_layer.head(),early_layer.count())\n",
        "        # print(late_layer.head(),late_layer.count())\n",
        "        \n",
        "        print('\\n\\n\\n\\n\\n\\n')\n",
        "# print(runtimes)\n",
        "pd.DataFrame(runtimes,index=['fc','residual','point_wise','early','late','depth_wise']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Runtime(cycles)',colormap='Set1')\n",
        "pd.DataFrame(energies,index=['fc','residual','point_wise','early','late','depth_wise']).T.plot.bar(stacked=True,figsize=(10,10),ylabel='Energy(mJ)',colormap='Set2')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Filename: content/mbnet/MobileNetV2_yxp_os.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/mbnet/MobileNetV2_xp_ws.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "Filename: content/mbnet/MobileNetV2_kcp_ws.csv\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc550e08710>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 207
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oslrAr6FVR5h",
        "outputId": "239b424d-068a-47a8-ea8e-f08cec6d3b4c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "df.filter(regex='Bottleneck._._\\d',axis=0)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Runtime (Cycles)</th>\n",
              "      <th>Activity count-based Energy (nJ)</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Layer Number</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Bottleneck1_1_1</th>\n",
              "      <td>200705</td>\n",
              "      <td>75334200.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck1_1_2</th>\n",
              "      <td>12212</td>\n",
              "      <td>55937000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck1_1_3</th>\n",
              "      <td>100353</td>\n",
              "      <td>60123900.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck2_1_1</th>\n",
              "      <td>602113</td>\n",
              "      <td>106242000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck2_1_2</th>\n",
              "      <td>9134</td>\n",
              "      <td>73251400.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck2_1_3</th>\n",
              "      <td>301058</td>\n",
              "      <td>196154000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck2_2_1</th>\n",
              "      <td>301058</td>\n",
              "      <td>367825000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck2_2_2</th>\n",
              "      <td>5944</td>\n",
              "      <td>41035200.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck2_2_3</th>\n",
              "      <td>75266</td>\n",
              "      <td>64829800.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_1_1</th>\n",
              "      <td>225793</td>\n",
              "      <td>102104000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_1_2</th>\n",
              "      <td>2976</td>\n",
              "      <td>26902900.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_1_3</th>\n",
              "      <td>169346</td>\n",
              "      <td>106256000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_2_1</th>\n",
              "      <td>169346</td>\n",
              "      <td>206568000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_2_2</th>\n",
              "      <td>1462</td>\n",
              "      <td>14661900.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_2_3</th>\n",
              "      <td>169346</td>\n",
              "      <td>105342000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_3_1</th>\n",
              "      <td>169346</td>\n",
              "      <td>100397000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_3_2</th>\n",
              "      <td>1462</td>\n",
              "      <td>14661900.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck3_3_3</th>\n",
              "      <td>37634</td>\n",
              "      <td>29399800.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_1_1</th>\n",
              "      <td>75265</td>\n",
              "      <td>47179200.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_1_2</th>\n",
              "      <td>879</td>\n",
              "      <td>8618610.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_1_3</th>\n",
              "      <td>56450</td>\n",
              "      <td>36194700.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_2_1</th>\n",
              "      <td>56450</td>\n",
              "      <td>80798800.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_2_2</th>\n",
              "      <td>488</td>\n",
              "      <td>4443140.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_2_3</th>\n",
              "      <td>56450</td>\n",
              "      <td>35772200.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_3_1</th>\n",
              "      <td>56450</td>\n",
              "      <td>34397000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_3_2</th>\n",
              "      <td>488</td>\n",
              "      <td>4443140.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_3_3</th>\n",
              "      <td>56450</td>\n",
              "      <td>35772200.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_4_1</th>\n",
              "      <td>56450</td>\n",
              "      <td>34397000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_4_2</th>\n",
              "      <td>488</td>\n",
              "      <td>4443140.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck4_4_3</th>\n",
              "      <td>18818</td>\n",
              "      <td>13454300.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_1_1</th>\n",
              "      <td>37634</td>\n",
              "      <td>38023700.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_1_2</th>\n",
              "      <td>842</td>\n",
              "      <td>8891120.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_1_3</th>\n",
              "      <td>225794</td>\n",
              "      <td>138600000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_2_1</th>\n",
              "      <td>225794</td>\n",
              "      <td>322038000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_2_2</th>\n",
              "      <td>842</td>\n",
              "      <td>8891120.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_2_3</th>\n",
              "      <td>56450</td>\n",
              "      <td>38092300.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_1_1</th>\n",
              "      <td>112898</td>\n",
              "      <td>113600000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_1_2</th>\n",
              "      <td>567</td>\n",
              "      <td>6036400.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck5_2_3</th>\n",
              "      <td>127010</td>\n",
              "      <td>85478600.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_2_1</th>\n",
              "      <td>127010</td>\n",
              "      <td>189554000.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_2_2</th>\n",
              "      <td>266</td>\n",
              "      <td>2796020.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_2_3</th>\n",
              "      <td>127010</td>\n",
              "      <td>85019300.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_3_1</th>\n",
              "      <td>127010</td>\n",
              "      <td>84347500.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_3_2</th>\n",
              "      <td>266</td>\n",
              "      <td>2796020.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck6_3_3</th>\n",
              "      <td>35282</td>\n",
              "      <td>24600800.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck7_1_1</th>\n",
              "      <td>70562</td>\n",
              "      <td>75203600.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck7_1_2</th>\n",
              "      <td>445</td>\n",
              "      <td>4660030.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Bottleneck7_1_3</th>\n",
              "      <td>117602</td>\n",
              "      <td>79787900.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                  Runtime (Cycles)   Activity count-based Energy (nJ)\n",
              " Layer Number                                                        \n",
              "Bottleneck1_1_1             200705                         75334200.0\n",
              "Bottleneck1_1_2              12212                         55937000.0\n",
              "Bottleneck1_1_3             100353                         60123900.0\n",
              "Bottleneck2_1_1             602113                        106242000.0\n",
              "Bottleneck2_1_2               9134                         73251400.0\n",
              "Bottleneck2_1_3             301058                        196154000.0\n",
              "Bottleneck2_2_1             301058                        367825000.0\n",
              "Bottleneck2_2_2               5944                         41035200.0\n",
              "Bottleneck2_2_3              75266                         64829800.0\n",
              "Bottleneck3_1_1             225793                        102104000.0\n",
              "Bottleneck3_1_2               2976                         26902900.0\n",
              "Bottleneck3_1_3             169346                        106256000.0\n",
              "Bottleneck3_2_1             169346                        206568000.0\n",
              "Bottleneck3_2_2               1462                         14661900.0\n",
              "Bottleneck3_2_3             169346                        105342000.0\n",
              "Bottleneck3_3_1             169346                        100397000.0\n",
              "Bottleneck3_3_2               1462                         14661900.0\n",
              "Bottleneck3_3_3              37634                         29399800.0\n",
              "Bottleneck4_1_1              75265                         47179200.0\n",
              "Bottleneck4_1_2                879                          8618610.0\n",
              "Bottleneck4_1_3              56450                         36194700.0\n",
              "Bottleneck4_2_1              56450                         80798800.0\n",
              "Bottleneck4_2_2                488                          4443140.0\n",
              "Bottleneck4_2_3              56450                         35772200.0\n",
              "Bottleneck4_3_1              56450                         34397000.0\n",
              "Bottleneck4_3_2                488                          4443140.0\n",
              "Bottleneck4_3_3              56450                         35772200.0\n",
              "Bottleneck4_4_1              56450                         34397000.0\n",
              "Bottleneck4_4_2                488                          4443140.0\n",
              "Bottleneck4_4_3              18818                         13454300.0\n",
              "Bottleneck5_1_1              37634                         38023700.0\n",
              "Bottleneck5_1_2                842                          8891120.0\n",
              "Bottleneck5_1_3             225794                        138600000.0\n",
              "Bottleneck5_2_1             225794                        322038000.0\n",
              "Bottleneck5_2_2                842                          8891120.0\n",
              "Bottleneck5_2_3              56450                         38092300.0\n",
              "Bottleneck6_1_1             112898                        113600000.0\n",
              "Bottleneck6_1_2                567                          6036400.0\n",
              "Bottleneck5_2_3             127010                         85478600.0\n",
              "Bottleneck6_2_1             127010                        189554000.0\n",
              "Bottleneck6_2_2                266                          2796020.0\n",
              "Bottleneck6_2_3             127010                         85019300.0\n",
              "Bottleneck6_3_1             127010                         84347500.0\n",
              "Bottleneck6_3_2                266                          2796020.0\n",
              "Bottleneck6_3_3              35282                         24600800.0\n",
              "Bottleneck7_1_1              70562                         75203600.0\n",
              "Bottleneck7_1_2                445                          4660030.0\n",
              "Bottleneck7_1_3             117602                         79787900.0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 206
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fNL-WnEGSIfX"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}