cupy-dask.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cupy\n",
    "from dask.diagnostics import ProgressBar\n",
    "import dask.array as da"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "rs = da.random.RandomState(RandomState=cupy.random.RandomState)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = rs.normal(10, 1, size=(500000, 500000), chunks=(10000, 10000))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2000.0"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.nbytes / 1e9  # 2TB"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Single GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[########################################] | 100% Completed |  1min 37.4s\n"
     ]
    }
   ],
   "source": [
    "with ProgressBar():\n",
    "    (x + 1)[::2, ::2].sum().compute(scheduler='single-threaded')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multiple GPUs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from dask.distributed import LocalCUDACluster, Client\n",
    "cluster = LocalCUDACluster()\n",
    "client = Client(cluster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 9.02 s, sys: 832 ms, total: 9.85 s\n",
      "Wall time: 19.2 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array(6.87500349e+11)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time (x + 1)[::2, ::2].sum().compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Try on the CPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "rs = da.random.RandomState()\n",
    "x = rs.normal(10, 1, size=(500000, 500000), chunks=(10000, 10000))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Single CPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[########################################] | 100% Completed |  2hr 39min 28.2s\n"
     ]
    }
   ],
   "source": [
    "with ProgressBar():\n",
    "    (x + 1)[::2, ::2].sum().compute(scheduler='single-threaded')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 80 cores with multi-threading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[########################################] | 100% Completed | 11min 31.3s\n"
     ]
    }
   ],
   "source": [
    "from dask.diagnostics import ProgressBar\n",
    "\n",
    "with ProgressBar():\n",
    "    (x + 1)[::2, ::2].sum().compute(scheduler='threads')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}

The multiple GPU section should begin

from dask_cuda import LocalCUDACluster
from dask.distributed import Client