benchmark_kernels.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "4c9a23e3-7965-4072-bdde-d3b575fdd54a",
   "metadata": {},
   "source": [
    "# GCM Filters Kernel Benchmarking"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5160201e-f86d-48f9-8fd1-4a6c8c30d262",
   "metadata": {},
   "source": [
    "## Baseline: `scipy.ndimage.uniform_filter`\n",
    "\n",
    "This is considered a \"good\" implementation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f31c7849-18a0-481d-a793-0a4eef8a052a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.ndimage import uniform_filter\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "%load_ext ptime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "b7efec79-0557-4b16-8c83-38091992e448",
   "metadata": {},
   "outputs": [],
   "source": [
    "shape = (1024, 1024)\n",
    "data = np.random.rand(*shape)\n",
    "data_u = np.random.rand(*shape)\n",
    "data_v = np.random.rand(*shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e1265c33-30b5-4e67-a59a-a668774550f2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "11.9 ms ± 545 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "uf_call = %timeit -o -n10 uniform_filter(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "550336ac-52f2-441e-8c27-5a7787e9753b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10.7 ms ± 126 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "output = np.zeros_like(data)\n",
    "uf_call_w_output = %timeit -o -n10 uniform_filter(data, output=output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b151394c-4de9-4a71-8511-00c628372435",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total serial time:   0.08 s\n",
      "Total parallel time: 0.02 s\n",
      "For a 4.05X speedup across 6 threads\n"
     ]
    }
   ],
   "source": [
    "uf_ptime = %ptime -o -n6 uniform_filter(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "6b2ddee9-6204-4c6f-a406-d85ab4fb1e7c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total serial time:   0.07 s\n",
      "Total parallel time: 0.01 s\n",
      "For a 5.92X speedup across 6 threads\n"
     ]
    }
   ],
   "source": [
    "uf_ptime_w_output = %ptime -o -n6 uniform_filter(data, output=output)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8ac1032-d1ac-4256-a1c2-c4db52639265",
   "metadata": {},
   "source": [
    "## Time GCM-Filters Functions in the same way\n",
    "- How fast are our filters in single-threaded mode compared to the baseline?\n",
    "- How do the scale in comparison?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "cfe7eafb-ccbe-4374-a694-c4f93b39098c",
   "metadata": {},
   "outputs": [],
   "source": [
    "import gcm_filters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "a6d34cc0-8800-498d-9d1c-210599631183",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12.1 ms ± 595 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "regular_laplacian = gcm_filters.kernels.RegularLaplacian()\n",
    "\n",
    "%timeit -n10 regular_laplacian(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "7933953e-2db8-42bd-a193-62e4f6fd6529",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total serial time:   0.08 s\n",
      "Total parallel time: 0.05 s\n",
      "For a 1.68X speedup across 6 threads\n"
     ]
    }
   ],
   "source": [
    "%ptime -n6 regular_laplacian(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "79377d76-d652-47c7-8ac6-84da8154fabc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# convenience function to get a filter function\n",
    "\n",
    "def get_filter_function(grid_type, grid_vars):\n",
    "\n",
    "    filter = gcm_filters.Filter(\n",
    "        filter_scale=4,\n",
    "        dx_min=1,\n",
    "        filter_shape=gcm_filters.FilterShape.TAPER,\n",
    "        grid_type=grid_type,\n",
    "        grid_vars=grid_vars,\n",
    "    )\n",
    "    \n",
    "    return gcm_filters.filter._create_filter_func(\n",
    "        filter.filter_spec,\n",
    "        gcm_filters.kernels.ALL_KERNELS[grid_type]\n",
    "    ), filter.filter_spec.n_steps_total\n",
    "\n",
    "filter_function, nsteps_total = get_filter_function(gcm_filters.kernels.GridType.REGULAR, {})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "60e06888-d6fb-4e14-8c38-e6fa01b8cd12",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "243 ms ± 3.34 ms per loop (mean ± std. dev. of 7 runs, 5 loops each)\n"
     ]
    }
   ],
   "source": [
    "%timeit -n5 filter_function(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "abfe7fbd-b7fe-49ea-aa5d-430acdcb7b0f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total serial time:   1.29 s\n",
      "Total parallel time: 1.04 s\n",
      "For a 1.23X speedup across 6 threads\n"
     ]
    }
   ],
   "source": [
    "%ptime -n6 filter_function(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "16cf03e7-6fae-4554-b729-217aa66e6979",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "15.428571428571429"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "216 / nsteps_total"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "269b1252-b55b-487f-8b27-7db380585e60",
   "metadata": {},
   "source": [
    "We just saw that the speed of the filter function is very nearly equal to the speed of the kernel (14 ms) times the number of steps in the filter.\n",
    "The scaling is also similar. So from here on we will just focus on the kernels themselves, assuming whatever improvements we make to the filters will carry over to the kernels."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3b86fc5-78f2-42a5-b391-2b4f3ca7b499",
   "metadata": {},
   "source": [
    "## Test All Kernels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "44967324-ad99-4e0f-89ee-edae82fcc935",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GridType.REGULAR\n",
      " 0 extra args\n",
      "247 ns ± 76.2 ns per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "13.9 ms ± 281 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "Total serial time:   0.10 s\n",
      "Total parallel time: 0.06 s\n",
      "For a 1.58X speedup across 6 threads\n",
      "GridType.REGULAR_WITH_LAND\n",
      " 1 extra args\n",
      "10.8 ms ± 71.3 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "24.1 ms ± 366 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "Total serial time:   0.15 s\n",
      "Total parallel time: 0.11 s\n",
      "For a 1.33X speedup across 6 threads\n",
      "GridType.IRREGULAR_WITH_LAND\n",
      " 8 extra args\n",
      "5.5 ms ± 511 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "28.2 ms ± 2.68 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "Total serial time:   0.19 s\n",
      "Total parallel time: 0.14 s\n",
      "For a 1.40X speedup across 6 threads\n",
      "GridType.TRIPOLAR_REGULAR_WITH_LAND\n",
      " 1 extra args\n",
      "6.69 ms ± 437 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "19.7 ms ± 132 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "Total serial time:   0.13 s\n",
      "Total parallel time: 0.11 s\n",
      "For a 1.16X speedup across 6 threads\n",
      "GridType.TRIPOLAR_POP_WITH_LAND\n",
      " 6 extra args\n",
      "13.6 ms ± 96.8 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "29.5 ms ± 419 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "Total serial time:   0.18 s\n",
      "Total parallel time: 0.13 s\n",
      "For a 1.43X speedup across 6 threads\n",
      "GridType.VECTOR_C_GRID\n",
      " 14 extra args\n",
      "25.4 ms ± 2.19 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "85 ms ± 7.64 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "Total serial time:   0.61 s\n",
      "Total parallel time: 0.44 s\n",
      "For a 1.38X speedup across 6 threads\n"
     ]
    }
   ],
   "source": [
    "results = {}\n",
    "\n",
    "import warnings\n",
    "warnings.simplefilter(\"ignore\")\n",
    "\n",
    "for name, kernel in gcm_filters.kernels.ALL_KERNELS.items():\n",
    "    print(name)\n",
    "    results[name.name] = {}\n",
    "    extra_inputs = {arg: np.zeros_like(data) for arg in kernel.required_grid_args()}\n",
    "    print(f' {len(extra_inputs)} extra args')\n",
    "    results[name.name]['init'] = %timeit -o -n10 kernel_func = kernel(**extra_inputs)\n",
    "    kernel_func = kernel(**extra_inputs)\n",
    "    if 'VECTOR' in name.name:\n",
    "        results[name.name]['call'] = %timeit -o -n10 kernel_func(data_u, data_v)\n",
    "        results[name.name]['call_ptime'] = %ptime -o -n6 kernel_func(data_u, data_v)\n",
    "    else:\n",
    "        results[name.name]['call'] = %timeit -o -n10 kernel_func(data)\n",
    "        results[name.name]['call_ptime'] = %ptime -o -n6 kernel_func(data)\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8256f912-eaf6-4029-9baa-03e4fcaab19b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>init_avg</th>\n",
       "      <th>init_std</th>\n",
       "      <th>call_avg</th>\n",
       "      <th>call_std</th>\n",
       "      <th>parallel_speedup</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>REGULAR</th>\n",
       "      <td>2.468571e-07</td>\n",
       "      <td>7.623177e-08</td>\n",
       "      <td>0.013928</td>\n",
       "      <td>0.000281</td>\n",
       "      <td>1.580123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>REGULAR_WITH_LAND</th>\n",
       "      <td>1.076826e-02</td>\n",
       "      <td>7.134474e-05</td>\n",
       "      <td>0.024120</td>\n",
       "      <td>0.000366</td>\n",
       "      <td>1.329384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>IRREGULAR_WITH_LAND</th>\n",
       "      <td>5.503437e-03</td>\n",
       "      <td>5.108730e-04</td>\n",
       "      <td>0.028195</td>\n",
       "      <td>0.002679</td>\n",
       "      <td>1.397791</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TRIPOLAR_REGULAR_WITH_LAND</th>\n",
       "      <td>6.685203e-03</td>\n",
       "      <td>4.371660e-04</td>\n",
       "      <td>0.019675</td>\n",
       "      <td>0.000132</td>\n",
       "      <td>1.158377</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TRIPOLAR_POP_WITH_LAND</th>\n",
       "      <td>1.363606e-02</td>\n",
       "      <td>9.677481e-05</td>\n",
       "      <td>0.029463</td>\n",
       "      <td>0.000419</td>\n",
       "      <td>1.428361</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>VECTOR_C_GRID</th>\n",
       "      <td>2.537815e-02</td>\n",
       "      <td>2.190825e-03</td>\n",
       "      <td>0.084961</td>\n",
       "      <td>0.007639</td>\n",
       "      <td>1.380396</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ndimage.uniform_filter</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.011904</td>\n",
       "      <td>0.000545</td>\n",
       "      <td>4.046563</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ndimage.uniform_filter (output)</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.010684</td>\n",
       "      <td>0.000126</td>\n",
       "      <td>5.920890</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     init_avg      init_std  call_avg  \\\n",
       "REGULAR                          2.468571e-07  7.623177e-08  0.013928   \n",
       "REGULAR_WITH_LAND                1.076826e-02  7.134474e-05  0.024120   \n",
       "IRREGULAR_WITH_LAND              5.503437e-03  5.108730e-04  0.028195   \n",
       "TRIPOLAR_REGULAR_WITH_LAND       6.685203e-03  4.371660e-04  0.019675   \n",
       "TRIPOLAR_POP_WITH_LAND           1.363606e-02  9.677481e-05  0.029463   \n",
       "VECTOR_C_GRID                    2.537815e-02  2.190825e-03  0.084961   \n",
       "ndimage.uniform_filter                    NaN           NaN  0.011904   \n",
       "ndimage.uniform_filter (output)           NaN           NaN  0.010684   \n",
       "\n",
       "                                 call_std  parallel_speedup  \n",
       "REGULAR                          0.000281          1.580123  \n",
       "REGULAR_WITH_LAND                0.000366          1.329384  \n",
       "IRREGULAR_WITH_LAND              0.002679          1.397791  \n",
       "TRIPOLAR_REGULAR_WITH_LAND       0.000132          1.158377  \n",
       "TRIPOLAR_POP_WITH_LAND           0.000419          1.428361  \n",
       "VECTOR_C_GRID                    0.007639          1.380396  \n",
       "ndimage.uniform_filter           0.000545          4.046563  \n",
       "ndimage.uniform_filter (output)  0.000126          5.920890  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_for_pandas = {\n",
    "    name: {\n",
    "        'init_avg': res['init'].average,\n",
    "        'init_std': res['init'].stdev,\n",
    "        'call_avg': res['call'].average,\n",
    "        'call_std': res['call'].stdev,\n",
    "        'parallel_speedup': res['call_ptime'].speedup\n",
    "    }\n",
    "    for name, res in results.items()\n",
    "}\n",
    "\n",
    "results_for_pandas['ndimage.uniform_filter'] = {\n",
    "    'call_avg': uf_call.average,\n",
    "    'call_std': uf_call.stdev,\n",
    "    'parallel_speedup': uf_ptime.speedup\n",
    "}\n",
    "\n",
    "results_for_pandas['ndimage.uniform_filter (output)'] = {\n",
    "    'call_avg': uf_call_w_output.average,\n",
    "    'call_std': uf_call_w_output.stdev,\n",
    "    'parallel_speedup': uf_ptime_w_output.speedup\n",
    "}\n",
    "\n",
    "df = pd.DataFrame(results_for_pandas).transpose()\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "a5565cc2-54de-4c32-98d8-f045d84b5523",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot(kind='bar', y='call_avg', yerr='call_std')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "c4b50c0d-24dc-470e-930a-a66bceffa976",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot(kind='bar', y='parallel_speedup')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b4b77a9f-2dca-4dfa-ba53-802d9f8763d8",
   "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": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}