{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Devices"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-info\">\n",
    "\n",
    "This tutorial is available as an IPython notebook at [malaya-speech/example/devices](https://github.com/huseinzol05/malaya-speech/tree/master/example/devices).\n",
    "    \n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### List available devices supported to run Malaya-Speech model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import malaya_speech\n",
    "import logging\n",
    "logging.basicConfig(level = logging.INFO)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('CPU:0', '0.268 GB'),\n",
       " ('XLA_CPU:0', '17.18 GB'),\n",
       " ('XLA_GPU:0', '17.18 GB'),\n",
       " ('XLA_GPU:1', '17.18 GB'),\n",
       " ('XLA_GPU:2', '17.18 GB'),\n",
       " ('XLA_GPU:3', '17.18 GB'),\n",
       " ('GPU:0', '30.486 GB'),\n",
       " ('GPU:1', '30.489 GB'),\n",
       " ('GPU:2', '30.489 GB'),\n",
       " ('GPU:3', '30.489 GB')]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "malaya_speech.utils.available_device()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Use specific device for specific model\n",
    "\n",
    "To do that, pass `device` parameter to any load model function in Malaya, default is `CPU:0`.\n",
    "\n",
    "```python\n",
    "malaya_speech.gender.deep_model(model = 'vggvox-v2', device = 'CPU:0')\n",
    "```\n",
    "\n",
    "Or if you want to use XLA,\n",
    "\n",
    "```python\n",
    "malaya_speech.gender.deep_model(model = 'vggvox-v2', device = 'XLA_CPU:0')\n",
    "```\n",
    "\n",
    "By default, `device` will automatically set to a gpu with the most empty memory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:running gender/vggvox-v2 using device /device:GPU:1\n"
     ]
    }
   ],
   "source": [
    "gender = malaya_speech.gender.deep_model(model = 'vggvox-v2', device = 'CPU:0')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Disable auto GPU\n",
    "\n",
    "Let say you do not want to use auto allocate to gpu, simply set `auto_gpu` to `False`, or set,\n",
    "\n",
    "```bash\n",
    "export CUDA_VISIBLE_DEVICES=''\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:running gender/vggvox-v2 using device /device:CPU:0\n"
     ]
    }
   ],
   "source": [
    "gender_cpu = malaya_speech.gender.deep_model(model = 'vggvox-v2', device = 'CPU:0', auto_gpu = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:running gender/vggvox-v2 using device /device:XLA_CPU:0\n"
     ]
    }
   ],
   "source": [
    "gender_xla_cpu = malaya_speech.gender.deep_model(model = 'vggvox-v2', device = 'XLA_CPU:0', auto_gpu = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8000, 16000)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y, sr = malaya_speech.load('speech/video/The-Singaporean-White-Boy.wav')\n",
    "y = y[:int(sr * 0.5)]\n",
    "len(y), sr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 923 ms, sys: 429 ms, total: 1.35 s\n",
      "Wall time: 1.22 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[0.21194158, 0.2119417 , 0.5761167 ]], dtype=float32)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "gender_cpu.predict_proba([y])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.61 s, sys: 503 ms, total: 5.11 s\n",
      "Wall time: 4.74 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[0.21194158, 0.2119417 , 0.5761167 ]], dtype=float32)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "gender_xla_cpu.predict_proba([y])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Again, not all Tensorflow operation support XLA**."
   ]
  }
 ],
 "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}