inference-times.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook is a simple experiment with git-cloning the DeepWeeds repo and run the `deepweeds.py` script to measure inference times."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cloning into 'DeepWeeds'...\n",
      "remote: Enumerating objects: 33, done.\u001b[K\n",
      "remote: Total 33 (delta 0), reused 0 (delta 0), pack-reused 33\n",
      "Unpacking objects: 100% (33/33), done.\n"
     ]
    }
   ],
   "source": [
    "!git clone https://github.com/AlexOlsen/DeepWeeds.git"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/storage/deep-weeds-experiments/DeepWeeds\n"
     ]
    }
   ],
   "source": [
    "%cd DeepWeeds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n",
      "Downloading DeepWeeds images to ./images/images.zip\n",
      "Finished downloading images.\n",
      "Unzipping ./images/images.zip\n",
      "Finished unzipping images.\n",
      "2019-11-04 19:05:57.258291: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1\n",
      "2019-11-04 19:05:57.278429: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.279237: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties: \n",
      "name: Quadro M4000 major: 5 minor: 2 memoryClockRate(GHz): 0.7725\n",
      "pciBusID: 0000:00:05.0\n",
      "2019-11-04 19:05:57.279545: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0\n",
      "2019-11-04 19:05:57.281194: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0\n",
      "2019-11-04 19:05:57.282403: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0\n",
      "2019-11-04 19:05:57.282826: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0\n",
      "2019-11-04 19:05:57.284389: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0\n",
      "2019-11-04 19:05:57.285532: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0\n",
      "2019-11-04 19:05:57.289752: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7\n",
      "2019-11-04 19:05:57.289938: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.290871: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.291649: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0\n",
      "2019-11-04 19:05:57.291957: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA\n",
      "2019-11-04 19:05:57.298304: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2600010000 Hz\n",
      "2019-11-04 19:05:57.299016: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x4734290 executing computations on platform Host. Devices:\n",
      "2019-11-04 19:05:57.299038: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Host, Default Version\n",
      "2019-11-04 19:05:57.357397: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.358055: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x45ae8c0 executing computations on platform CUDA. Devices:\n",
      "2019-11-04 19:05:57.358083: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Quadro M4000, Compute Capability 5.2\n",
      "2019-11-04 19:05:57.358286: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.358776: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties: \n",
      "name: Quadro M4000 major: 5 minor: 2 memoryClockRate(GHz): 0.7725\n",
      "pciBusID: 0000:00:05.0\n",
      "2019-11-04 19:05:57.358855: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0\n",
      "2019-11-04 19:05:57.358894: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0\n",
      "2019-11-04 19:05:57.358923: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0\n",
      "2019-11-04 19:05:57.358938: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0\n",
      "2019-11-04 19:05:57.358959: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0\n",
      "2019-11-04 19:05:57.358980: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0\n",
      "2019-11-04 19:05:57.359003: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7\n",
      "2019-11-04 19:05:57.359072: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.359587: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.360088: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0\n",
      "2019-11-04 19:05:57.360309: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0\n",
      "2019-11-04 19:05:57.361694: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:\n",
      "2019-11-04 19:05:57.361722: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165]      0 \n",
      "2019-11-04 19:05:57.361736: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0:   N \n",
      "2019-11-04 19:05:57.361935: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.362488: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2019-11-04 19:05:57.362979: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1304] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 7654 MB memory) -> physical GPU (device: 0, name: Quadro M4000, pci bus id: 0000:00:05.0, compute capability: 5.2)\n",
      "2019-11-04 19:06:48.236978: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0\n",
      "2019-11-04 19:06:48.419793: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7\n"
     ]
    }
   ],
   "source": [
    "!python3 deepweeds.py inference --model models/resnet.hdf5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "outputs\n",
      "└── 20191104-190645\n",
      "    └── tf_inference_times.csv\n",
      "\n",
      "1 directory, 1 file\n"
     ]
    }
   ],
   "source": [
    "!tree outputs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Some simple statistics of inference_times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/storage/deep-weeds-experiments/DeepWeeds\n"
     ]
    }
   ],
   "source": [
    "%cd DeepWeeds\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('outputs/20191104-190645/tf_inference_times.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(17509, 3)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "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>Filename</th>\n",
       "      <th>Preprocessing time (ms)</th>\n",
       "      <th>Inference time (ms)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>20160928-140314-0.jpg</td>\n",
       "      <td>27.733088</td>\n",
       "      <td>4564.514637</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20160928-140337-0.jpg</td>\n",
       "      <td>14.467001</td>\n",
       "      <td>28.810501</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>20160928-140731-0.jpg</td>\n",
       "      <td>11.768818</td>\n",
       "      <td>31.310081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>20160928-140747-0.jpg</td>\n",
       "      <td>11.573315</td>\n",
       "      <td>30.301332</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20160928-141107-0.jpg</td>\n",
       "      <td>14.033556</td>\n",
       "      <td>27.382612</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                Filename  Preprocessing time (ms)  Inference time (ms)\n",
       "0  20160928-140314-0.jpg                27.733088          4564.514637\n",
       "1  20160928-140337-0.jpg                14.467001            28.810501\n",
       "2  20160928-140731-0.jpg                11.768818            31.310081\n",
       "3  20160928-140747-0.jpg                11.573315            30.301332\n",
       "4  20160928-141107-0.jpg                14.033556            27.382612"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first inference takes much more time than subsequent ones. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# TODO should include file size as well?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "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>Preprocessing time (ms)</th>\n",
       "      <th>Inference time (ms)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>17509.000000</td>\n",
       "      <td>17509.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>12.302010</td>\n",
       "      <td>27.833134</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>2.772527</td>\n",
       "      <td>34.326025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>10.621071</td>\n",
       "      <td>25.703192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>11.795521</td>\n",
       "      <td>27.024508</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>12.087345</td>\n",
       "      <td>27.384758</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>12.443066</td>\n",
       "      <td>27.845383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>226.429462</td>\n",
       "      <td>4564.514637</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Preprocessing time (ms)  Inference time (ms)\n",
       "count             17509.000000         17509.000000\n",
       "mean                 12.302010            27.833134\n",
       "std                   2.772527            34.326025\n",
       "min                  10.621071            25.703192\n",
       "25%                  11.795521            27.024508\n",
       "50%                  12.087345            27.384758\n",
       "75%                  12.443066            27.845383\n",
       "max                 226.429462          4564.514637"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "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>Preprocessing time (ms)</th>\n",
       "      <th>Inference time (ms)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Preprocessing time (ms)</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.045208</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Inference time (ms)</th>\n",
       "      <td>0.045208</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                         Preprocessing time (ms)  Inference time (ms)\n",
       "Preprocessing time (ms)                 1.000000             0.045208\n",
       "Inference time (ms)                     0.045208             1.000000"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# correlation matrix \n",
    "df.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = sns.boxenplot(data=df)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}