Timing multi-stream dask-cuda

dask-cuda-multistream-timing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from dask_cuda import LocalCUDACluster\n",
    "from dask.distributed import Client\n",
    "import numpy as np\n",
    "import cupy\n",
    "import dask.array as da\n",
    "import asyncio\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_svd(client):\n",
    "    x = cupy.random.random((100000, 1000))\n",
    "    d = da.from_array(x, chunks=(10000, 1000), asarray=False)\n",
    "    u, s, v = np.linalg.svd(d)\n",
    "    return client.compute(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def without_streams(iterations=1):\n",
    "    async with LocalCUDACluster(asynchronous=True) as cluster:\n",
    "        async with Client(cluster, asynchronous=True) as client:\n",
    "            t = time.time()\n",
    "\n",
    "            for i in range(iterations):\n",
    "                res = await compute_svd(client)\n",
    "\n",
    "            print(\"Single (default) stream time for\", iterations, \"iterations:\", time.time() - t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "async def with_streams(iterations=1):\n",
    "    async with LocalCUDACluster(asynchronous=True) as cluster:\n",
    "        async with Client(cluster, asynchronous=True) as client:\n",
    "            streams = [cupy.cuda.stream.Stream() for _ in range(iterations)]\n",
    "\n",
    "            t = time.time()\n",
    "\n",
    "            svd = []\n",
    "            for stream in streams:\n",
    "                with stream:\n",
    "                    svd.append(compute_svd(client))\n",
    "\n",
    "            cupy.cuda.device.Device().synchronize()\n",
    "\n",
    "            res = []\n",
    "            for s in svd:\n",
    "                res.append(await s)\n",
    "\n",
    "            print(iterations, \"streams time for\", iterations, \"iterations:\", time.time() - t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Single (default) stream time for 1 iterations: 12.272627592086792\n"
     ]
    }
   ],
   "source": [
    "await without_streams(iterations=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 streams time for 1 iterations: 11.929816484451294\n"
     ]
    }
   ],
   "source": [
    "await with_streams(iterations=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Single (default) stream time for 2 iterations: 22.586686849594116\n"
     ]
    }
   ],
   "source": [
    "await without_streams(iterations=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 streams time for 2 iterations: 14.74821138381958\n"
     ]
    }
   ],
   "source": [
    "await with_streams(iterations=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Single (default) stream time for 4 iterations: 44.295682430267334\n"
     ]
    }
   ],
   "source": [
    "await without_streams(iterations=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4 streams time for 4 iterations: 19.735482931137085\n"
     ]
    }
   ],
   "source": [
    "await with_streams(iterations=4)"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}