Measuring_energy_consumption.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "003b72ad-df96-4dcc-a4f9-474bbaf6f05a",
   "metadata": {},
   "source": [
    "# How to use EnergyMeter\n",
    "EnergyMeter allows us to measure the energy consumption of code on Python, segregating their energy usage per component (CPU, DRAM, GPU and \n",
    "Hard Disk).\n",
    "\n",
    "We will first introduce how to use EnergyMeter with a minimal example in which we will use numpy to perform some calculations. We create the EnergyMeter by giving it the parameters of the energy consumption of the hard drive (TODO: set reasonable default params so these are optional) and we also give it a label, that will be used to report the results later on.\n",
    "\n",
    "All we need to do afterwards, is to run meter.begin() before running the code we want to measure and meter.end() after running the code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "e21dfdbb-c2d8-4f80-9b96-55332ad1f948",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:imports error \n",
      " You need to install pymongo>=3.9.0 in order to use MongoOutput \n"
     ]
    }
   ],
   "source": [
    "from energy_meter import EnergyMeter\n",
    "import time\n",
    "import numpy as np\n",
    "\n",
    "meter = EnergyMeter(disk_avg_speed=1600*1e6, \n",
    "                            disk_active_power=6, \n",
    "                            disk_idle_power=1.42, \n",
    "                            label=\"Matrix Multiplication\", include_idle=False)\n",
    "meter.begin()\n",
    "(np.random.rand(300,300)**5)*1/np.random.rand(300,300)\n",
    "meter.end()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a1de0ebc-b38f-41ae-815a-f456723c4b34",
   "metadata": {},
   "source": [
    "We can now see how the energy was spent while running our code. For this, we can use the methods get_total_jules_per_component() or plot_total_jules_per_component() to check how much energy was spent while running the code and how the energy was distributed in the computer components."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2e9a0690-2d8b-4a9a-8dc0-23bd4d0fbda4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "meter.plot_total_jules_per_component()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4861befc-fc21-4e8c-bc4f-2c8420c00e36",
   "metadata": {},
   "source": [
    "Note that even if our code did not write or read anything to disk, the plot indicates that the disk used some energy. This is because we also measure the energy used by the disk while being idle.\n",
    "\n",
    "## A more interesting example\n",
    "Let's test EnergyMeter in a more interesting example: measuring the energy consumption of performing inference with a Convolutional Neural Network, more specifically VGG16."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a953a674-7c53-4149-a45b-d476dd39cbea",
   "metadata": {},
   "outputs": [],
   "source": [
    "from urllib.request import urlopen\n",
    "from PIL import Image\n",
    "import timm\n",
    "import torch\n",
    "import json\n",
    "import torchvision\n",
    "\n",
    "transforms = torchvision.transforms.Compose([\n",
    "    torchvision.transforms.transforms.Resize(size=256),\n",
    "    torchvision.transforms.transforms.CenterCrop(size=(224, 224)),\n",
    "    torchvision.transforms.transforms.ToTensor(),\n",
    "    torchvision.transforms.transforms.Normalize(mean=torch.Tensor([0.4850, 0.4560, 0.4060]), \n",
    "                                                 std=torch.Tensor([0.2290, 0.2240, 0.2250])),\n",
    "])\n",
    "\n",
    "# the json file where the output must be stored\n",
    "json_file = open(\"imagenet_labels.json\", \"r\")\n",
    "imagenet_dict = json.load(json_file)\n",
    "json_file.close()\n",
    "\n",
    "def predict(model, img):\n",
    "    output = model(img)\n",
    "    top5_probabilities, top5_class_indices = torch.topk(output.softmax(dim=1) * 100, k=5)\n",
    "    return top5_probabilities, top5_class_indices.tolist()[0]\n",
    "\n",
    "def preprocess_data(img):\n",
    "    # unsqueeze single image into batch of 1.\n",
    "    return transforms(img).unsqueeze(0)\n",
    "\n",
    "model = timm.create_model('vgg16.tv_in1k', pretrained=True)\n",
    "model = model.eval()\n",
    "model\n",
    "\n",
    "# If we run this twice, the GPU seems to be preprocessing the image when\n",
    "# we run the CPU test. In order to avoid it, it's necessary to load the\n",
    "# model again, so it seems something remains in the GPU even after running\n",
    "# model.to(\"cpu\").\n",
    "\n",
    "img = Image.open(urlopen(\n",
    "    'https://unsplash.com/photos/DwxlhTvC16Q/download?ixid=M3wxMjA3fDB8MXxzZWFyY2h8MXx8bWVyY2VkZXMlMjBjYXJ8ZW58MHx8fHwxNjg2NzA1MzUyfDA&force=true&w=640'\n",
    "))\n",
    "\n",
    "# CPU\n",
    "# Disk in LSD15 has an avg i/o speed of 1600MB/s and \n",
    "# a power consumption of 6W when active and 1.42W when idle.\n",
    "meter_cpu = EnergyMeter(disk_avg_speed=1600*1e6, \n",
    "                        disk_active_power=6, \n",
    "                        disk_idle_power=1.42, \n",
    "                        label=\"CPU Inference\")\n",
    "model = model.to(\"cpu\")\n",
    "\n",
    "meter_cpu.begin()\n",
    "for _ in range(200):\n",
    "    p_img = preprocess_data(img).to(\"cpu\")\n",
    "    probs, labels = predict(model, p_img)\n",
    "meter_cpu.end()\n",
    "\n",
    "# To ensure samples from GPU power stats are not mixed up.\n",
    "time.sleep(1)\n",
    "\n",
    "meter_gpu = EnergyMeter(disk_avg_speed=1600*1e6, \n",
    "                        disk_active_power=6, \n",
    "                        disk_idle_power=1.42, \n",
    "                        label=\"GPU Inference\")\n",
    "model = model.to(\"cuda:0\")\n",
    "meter_gpu.begin()\n",
    "for _ in range(200):\n",
    "    p_img = preprocess_data(img).to(\"cuda:0\")\n",
    "    probs, labels = predict(model, p_img)\n",
    "meter_gpu.end()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "dccbd7a6-516d-4585-9368-486bcdf1079e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "meter_gpu.plot_total_jules_per_component()\n",
    "meter_cpu.plot_total_jules_per_component()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9d2d01c5-74d0-4e2e-9f31-1865f3edc36e",
   "metadata": {},
   "source": [
    "# The overhead of EnergyMeter\n",
    "We test the overhead of using the EnergyMeter. For this, we run inference for 100 and 1000 times with and without the EnergyMeter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "176283e6-0dde-4470-a9fe-790d3d94de6e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculating the overhead of the meter.\n",
    "def inference_with_meter(model, img, n_inference):\n",
    "    meter_overhead = EnergyMeter(disk_avg_speed=1600*1e6, \n",
    "                            disk_active_power=6, \n",
    "                            disk_idle_power=1.42, \n",
    "                            label=\"Inference\")\n",
    "    meter_overhead.begin()\n",
    "    for _ in range(n_inference):\n",
    "        p_img = preprocess_data(img).to(\"cuda:0\")\n",
    "        probs, labels = predict(model, p_img)\n",
    "    meter_overhead.end()\n",
    "\n",
    "def inference_without_meter(model, img, n_inference):\n",
    "    for _ in range(n_inference):\n",
    "        p_img = preprocess_data(img).to(\"cuda:0\")\n",
    "        probs, labels = predict(model, p_img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "677374a8-76dd-42fa-ac38-a512fa303e41",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "952 ms ± 21.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%timeit inference_with_meter(model, img, n_inference=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "139fcfee-6280-40c6-a2c6-3aea24680de1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "629 ms ± 806 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%timeit inference_without_meter(model, img, n_inference=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "09c0ddc5-750f-4157-ab3a-ee5a9318bea8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.74 s ± 250 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%timeit inference_with_meter(model, img, n_inference=1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "dd6993f1-87c3-474b-be67-4be86388233f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.31 s ± 12.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%timeit inference_without_meter(model, img, n_inference=1000)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c9f2260-9f93-4508-b658-358cb0e5c968",
   "metadata": {},
   "source": [
    "We see that the EnergyMeter adds an overhead of around 300ms to the original task and that this overhead is constant.\n",
    "\n",
    "## How does this overhead affect the code between meter.begin() and meter.end()?\n",
    "Even though we have measured the overhead of the meter, this measurement included everything from creating the meter, starting and finishing it. However, the overhead between meter.begin() and meter.end() is the most important one because it's affecting the normal functioning of the code that we want to profile.\n",
    "\n",
    "We will perform the same measurements as before but only timing the model's inference (leaving out the creation of the energy meter and its begin() and end() methods)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c5f9b1d3-ebd0-4ac3-9269-e4a624cb0a18",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "634 ms ± 7.77 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "meter_overhead = EnergyMeter(disk_avg_speed=1600*1e6, \n",
    "                            disk_active_power=6, \n",
    "                            disk_idle_power=1.42, \n",
    "                            label=\"Inference\")\n",
    "meter_overhead.begin()\n",
    "%timeit inference_without_meter(model, img, n_inference=100)\n",
    "meter_overhead.end()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "6e2f7d10-1fb3-4509-985b-4ab26aa8ff46",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.32 s ± 12.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "meter_overhead = EnergyMeter(disk_avg_speed=1600*1e6, \n",
    "                            disk_active_power=6, \n",
    "                            disk_idle_power=1.42, \n",
    "                            label=\"Inference\")\n",
    "meter_overhead.begin()\n",
    "%timeit inference_without_meter(model, img, n_inference=1000)\n",
    "meter_overhead.end()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b29967c6-207f-4963-ab64-d39f53025a9b",
   "metadata": {},
   "source": [
    "We see that the time measured for inference while running our meter for 100 and 1000 images is (almost) the same as without it. This is because the main overhead is in the begin() and end() methods, so we can assume that the code inside these two calls is a fair representation of what we would see without our meter."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}