ex_04_mult_pool.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A Guided Tour of Ray Core: Multiprocessing Pool"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[*Distributed multiprocessing.Pool*](https://docs.ray.io/en/latest/multiprocessing.html) make it easy to scale existing Python applications that use [`multiprocessing.Pool`](https://docs.python.org/3/library/multiprocessing.html) by leveraging *actors*.\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, let's start Ray…"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import logging\n",
    "import ray\n",
    "\n",
    "ray.init(\n",
    "    ignore_reinit_error=True,\n",
    "    logging_level=logging.ERROR,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multiprocessing Pool example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following is a simple Python function with a slight delay added (to make it behave like a more complex calculation)..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "\n",
    "def my_function (x):\n",
    "    time.sleep(1)\n",
    "    return x ** 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "\n",
    "my_function(11)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, let's try using this with Python's built-in [multiprocessing.Pool](https://docs.python.org/3/library/multiprocessing.html):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "\n",
    "from multiprocessing import Pool\n",
    "\n",
    "pool = Pool()\n",
    "\n",
    "for result in pool.map(my_function, range(50)):\n",
    "    print(result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we'll create a *Pool* using and distribute its tasks across a cluster (or across the available cores on a laptop):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "\n",
    "from ray.util.multiprocessing import Pool\n",
    "\n",
    "pool = Pool()\n",
    "\n",
    "for result in pool.map(my_function, range(50)):\n",
    "    print(result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The distributed version has the trade-off of increased overhead, although now it can scale-out horizontally across a cluster. The benefits would be more pronounced with a more computationally expensive calculation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, shutdown Ray"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ray.shutdown()"
   ]
  }
 ],
 "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}