(Additional) Profiling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Profiling for extract_dHdl() In Detail\n",
    "\n",
    "## Overview\n",
    "\n",
    "Based on the last discussion and the previous profiling results, we identified that reading a large number of small files is a bottleneck and there is more to investigate for the long processing time. This ipython notebook discusses the further investigation with the two experiments, \n",
    "- profiling results for the alchemlyb GROMACS parsing function, ``extract_dHdl()`` \n",
    "   - pd.DataFrame conversion is expensive\n",
    "- cost of writing/reading hdf5 structured binary format\n",
    "\n",
    "### Profiling pd.DataFrame inside of ``extract_dHdl()``\n",
    "\n",
    "Processing time is not entirely dependent on a large number of small files to read, conversion to Pandas DataFrame is another big portion of the bottleneck. Profiling result indicates that ``_extract_dataframe()`` function below consumes 88.3% time of ``extract_dHdl()`` function call, and what it does is that parses headers and builds pd.DataFrame from a GROMACS xvg file.\n",
    "\n",
    "The simple conversion to pd.DataFrame consumes 47.1% of time (See Line 188) and this is where concurrent execution can reduce the processing time. The practical issue is a Lustre filesystem locking for accessing files in a same directory. Opening many files in the same directory creates contention according to the documentation†, and this experiment shows that there are 25,000 files dividied by 106 sims parent directories resulting in average 236 files in the same directory to read. We need a test run to verify if this is true for file reading, not writing.\n",
    "\n",
    "†Lustre Best Practices: https://www.nas.nasa.gov/hecc/support/kb/lustre-best-practices_226.html\n",
    "\n",
    "```\n",
    "(alchemlyb) [hrlee@login018 ~]$ python -m line_profiler analysis.nodask.py.lprof\n",
    "Timer unit: 1e-06 s\n",
    "\n",
    "Total time: 14278.7 s\n",
    "File: /home/hrlee/.conda/envs/alchemlyb/lib/python3.7/site-packages/alchemlyb/parsing/gmx.py\n",
    "Function: extract_dHdl at line 82\n",
    "\n",
    "Line #      Hits         Time  Per Hit   % Time  Line Contents\n",
    "==============================================================\n",
    "    82\n",
    "    83                                           def extract_dHdl(xvg, T):\n",
    "...(omitted)...\n",
    "    97     25089      54164.0      2.2      0.0      beta = 1/(k_b * T)\n",
    "    98\n",
    "    99     25089 1262148551.0  50306.8      8.8      state, lambdas, statevec = _extract_state(xvg)\n",
    "   100\n",
    "   101                                               # extract a DataFrame from XVG data\n",
    "   102     25089 12611389601.0 502666.1     88.3      df = _extract_dataframe(xvg)\n",
    "\n",
    "...(omitted)...\n",
    "\n",
    "```\n",
    "\n",
    "```\n",
    "Total time: 11287.7 s\n",
    "File: /home/hrlee/.conda/envs/alchemlyb/lib/python3.7/site-packages/alchemlyb/parsing/gmx.py\n",
    "Function: _extract_dataframe at line 153\n",
    "\n",
    "Line #      Hits         Time  Per Hit   % Time  Line Contents\n",
    "==============================================================\n",
    "   153\n",
    "   154                                           def _extract_dataframe(xvg):\n",
    "   155                                               \"\"\"Extract a DataFrame from XVG data.\n",
    "   156\n",
    "   157                                               \"\"\"\n",
    "   158     25089    7650896.0    305.0      0.1      with anyopen(xvg, 'r') as f:\n",
    "   159     25089      44973.0      1.8      0.0          names = []\n",
    "   160     25089      23741.0      0.9      0.0          rows = []\n",
    "   161 206530593 1768891547.0      8.6     15.7          for line in f:\n",
    "   162 206505504  176606311.0      0.9      1.6              line = line.strip()\n",
    "   163 206505504  151686186.0      0.7      1.3              if len(line) == 0:\n",
    "   164                                                           continue\n",
    "   165\n",
    "   166 206505504  164058596.0      0.8      1.5              if \"label\" in line and \"xaxis\" in line:\n",
    "   167     25089      57774.0      2.3      0.0                  xaxis = line.split('\"')[-2]\n",
    "   168\n",
    "   169 206505504  564111193.0      2.7      5.0              if line.startswith(\"@ s\") and \"subtitle\" not in line:\n",
    "   170   2116821    2960341.0      1.4      0.0                  name = line.split(\"legend \")[-1].replace('\"','').strip()\n",
    "   171   2116821    1563486.0      0.7      0.0                  names.append(name)\n",
    "   172\n",
    "   173                                                       # should catch non-numeric lines so we don't proceed in parsing\n",
    "   174                                                       # here\n",
    "   175 206505504  150227213.0      0.7      1.3              if line.startswith(('#', '@')) :\n",
    "   176   2693868    1679231.0      0.6      0.0                  continue\n",
    "   177\n",
    "   178 203811636  144638777.0      0.7      1.3              if line.startswith('&'):\n",
    "   179                                                           raise NotImplementedError('{}: Multi-data not supported,'\n",
    "   180                                                                                     'only simple NXY format.'.format(xvg))\n",
    "   181                                                       # parse line as floats\n",
    "   182 203811636 2696231020.0     13.2     23.9              row = map(float, line.split())\n",
    "   183 203811636  143656049.0      0.7      1.3              rows.append(row)\n",
    "   184\n",
    "   185     25089      58698.0      2.3      0.0      cols = [xaxis]\n",
    "   186     25089     189198.0      7.5      0.0      cols.extend(names)\n",
    "   187\n",
    "   188     25089 5313358786.0 211780.4     47.1      return pd.DataFrame(rows, columns=cols)\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### HDF5 Write/Read (In Progress)\n",
    "\n",
    "Alchemlyb provides convenient data storage and retrieval in HDF5 hierarchical data format (see the details: [datreant](https://github.com/datreant/datreant)), this experiment is to measure an elapsed time for writing and reading hdf5 in the ``get_dHdl()`` function of [the original process](https://github.com/hpcanalytics/midas-alchemlyb/blob/master/analysis.ipynb). The following results indicate that some extra time is required to write hdf5 files but reading hdf5 instead of parsing xvg files again is a huge improvement for reducing the processing time. In detail, the implementation is:\n",
    "\n",
    "```python\n",
    "def get_dHdl(sim, lower=None, upper=None, step=None):\n",
    "    try:\n",
    "        # Reading hdf5, if exists\n",
    "        dHdl = sim.data.retrieve('dHdl') \n",
    "        ...(omitted)...\n",
    "    except:\n",
    "        dHdl = get_dHdl_XVG(sim, lower=lower, upper=upper, step=step)\n",
    "        # NEW: Writing hdf5 of the pandas DataFrame, dHdl, concatenated.\n",
    "        sim.data['dHdl'] = dHdl\n",
    "    \n",
    "    return dHdl\n",
    "```\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_h5_write = pd.read_csv(open(\"logs/get_dHdl.nodask.hdf5_write.log\"), sep=\",\",quotechar = '\\'')\n",
    "df_h5_read = pd.read_csv(open(\"logs/get_dHdl.nodask.hdf5_read.log\"), sep=\",\",quotechar = '\\'')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Log files are available at [hdf5_read.log](https://github.com/hpcanalytics/midas-alchemlyb/blob/master/logs/get_dHdl.nodask.hdf5_read.log), [hdf5_write.log](https://github.com/hpcanalytics/midas-alchemlyb/blob/master/logs/get_dHdl.nodask.hdf5_write.log)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### HDF5 Write\n",
    "\n",
    "Storing h5 file is useful for later use, and the following bar chart tells us how much time it needs to store additionally."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0,0.5,u'Elapsed Time (Second)')"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa2078a8ad0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_h5_write = df_h5_write.sort_values(by=['elapsed_time']).reset_index(drop=True)\n",
    "df_h5_write = df_h5_write[df_h5_write.type == \"hdf5_write\"]\n",
    "ax = df_h5_write.plot.bar(x='sim',y='elapsed_time', figsize=(20,4))\n",
    "ax.set_title(\"Writing HDF5\")\n",
    "ax.set_xlabel(\"Sim Name\")\n",
    "ax.set_ylabel(\"Elapsed Time (Second)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that each sim concatenates pd.DataFrame extracted from about 10 to 280 xvg files and store a single h5 file. The distance of elapsed time among them is too large, it can be due to Lustre contention or other unknown issues."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### HDF5 Read\n",
    "\n",
    "Our first run (without Dask) completed in 3 hours and 30 minutes for processing 25k files, and it can be reduced in 2 minutes, just reading 106 (sim size) h5 files.\n",
    "\n",
    "```\n",
    "5105885.ext+ extern COMPLETED 2019-03-21T00:36:49 2019-03-21T00:38:52   00:02:03 808K 140656K 1 28 r079 \n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0,0.5,u'Elapsed Time (Second)')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa1d7e13590>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_h5_read = df_h5_read.sort_values(by=['elapsed_time']).reset_index(drop=True)\n",
    "\n",
    "ax = df_h5_read.plot.bar(x='sim',y='elapsed_time', figsize=(20,4))\n",
    "ax.set_title(\"Reading HDF5\")\n",
    "ax.set_xlabel(\"Sim Name\")\n",
    "ax.set_ylabel(\"Elapsed Time (Second)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Conclusion\n",
    "\n",
    "We are still in a exploring mode to find an optimal solution in dealing with a large number of intermediate files (typically small sizes). Code refactoring can be an easy answer to improve (if code is available to fix and implemented without performance consideration) but our approach needs to be a higher level to suggest a general resolution.\n",
    "\n",
    "Distributed file access performance is dependent on a hardware specification (e.g. SSD, NVMe) and a file system (e.g. Lustre) along with operating system configuration. Best practice to achieve optimal performance starts from understanding existing system and configuration, otherwise a simple mistake can create a huge bottleneck easily.\n",
    "\n",
    "The couple of experiments indicate that 1) concurrent data file processing divided by multiple directories will reduce the enrie processing time, and 2) conversion to hdf5 format will provide better performance for later use over the ascii plain text file.\n",
    "\n",
    "Future work includes the following experiments: 1) file re-distribution on Lustre, 2) conversion cost to hdf5, 3) data moving cost for reading in memdisk or SSD, 4) replacing file output by memory using a Python API, e.g. [gmxapi](https://github.com/kassonlab/gmxapi) or using a queueing system e.g. ZMQ."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}