xref: /petsc/src/benchmarks/streams/process.py (revision 293a2e3a936a6e3794ad5cab930a69f1b4ef35ab)

#!/usr/bin/env python
#!/bin/env python
#
#    Computers speed up of Streams benchmark results generated by make streams and plots
#
#    matplotlib can switch between different backends hence this needs to be run
#    twice to first generate a file and then display a window
#
import os
#
def process(fileoutput = 1):
  import re
  ff = open('scaling.log')
  data = ff.read()
  ff.close()

  hosts  = {}
  triads = {}
  speedups = {}
  match = data.split('Number of MPI processes ')
  for i in match:
    if i:
      fields = i.split('\n')
      size = int(fields[0])
      hosts[size] = fields[1:size+1]
      triads[size] = float(fields[size+5].split()[1])

  if len(hosts) < 2: return

  ff = open('scaling.log','a')
  if fileoutput: print 'np  speedup'
  if fileoutput: ff.write('np  speedup\n')
  for sizes in hosts:
    speedups[sizes] = triads[sizes]/triads[1]
    if fileoutput: print sizes,round(triads[sizes]/triads[1],2)
    if fileoutput: ff.write(str(sizes)+' '+str(round(triads[sizes]/triads[1],2))+'\n')

  if fileoutput: print "Estimation of possible speedup of MPI programs based on Streams benchmark."
  if fileoutput: ff.write("Estimation of possible speedup of MPI programs based on Streams benchmark.\n")

  if fileoutput:
    import re
    last = max(hosts.keys())
    lasthosts = hosts[last]
    for i in range(0,len(lasthosts)):
      lasthosts[i] = re.sub(r"Process [0-9]*", "", lasthosts[i])
    ulasthosts = list(set(lasthosts))
    print "It appears you have "+str(len(ulasthosts))+" nodes"
    ff.write("It appears you have "+str(len(ulasthosts))+" nodes\n")

    testhosts = []
    for i in range(0,len(lasthosts)):
      testhosts.append(ulasthosts[i % len(ulasthosts)])
    if testhosts == lasthosts:
      print "   distributed in a round robin order"
      ff.write("   distributed in a round robin order\n")
    else:
      print "   NOT distributed in a round robin order"
      ff.write("   NOT distributed in a round robin order\n")

  try:
    import matplotlib
  except:
    print "Unable to open matplotlib to plot speedup"
    return

  try:
    if fileoutput: matplotlib.use('Agg')
    import matplotlib.pyplot as plt
  except:
    print "Unable to open matplotlib to plot speedup"
    return

  try:
    fig, ax1 = plt.subplots()
    plt.title('MPI Perfect and Streams Speedup')
    ax2 = ax1.twinx()
    ax1.set_autoscaley_on(False)

    # make sure that actual bandwidth values (as opposed to perfect speedup) takes
    # at least a third of the y axis
    ymax = min(max(hosts.keys()), 3*max(triads.values())/min(triads.values()) - 2)

    ax1.set_xlim([min(hosts.keys()),max(hosts.keys())])
    ax1.set_ylim([min(hosts.keys()),ymax])
    ax1.set_xlabel('Number of MPI processes')
    ax1.set_ylabel('Memory Bandwidth Speedup')
    ax1.plot(hosts.keys(),hosts.keys(),'b',hosts.keys(),speedups.values(),'r-o')
    ax2.set_autoscaley_on(False)
    ax2.set_xlim([min(hosts.keys()),max(hosts.keys())])
    ax2.set_ylim([min(triads.values())/1000.,min(triads.values())*ymax/1000.])
    ax2.set_ylabel("Achieved Bandwidth. Gigabytes per Second")

    plt.show()
    if fileoutput: plt.savefig('scaling.png')
    if fileoutput: print "See graph in the file src/benchmarks/streams/scaling.png"
    if fileoutput: ff.write("See graph in the file src/benchmarks/streams/scaling.png\n")
  except Exception, e:
    if fileoutput: print "Unable to plot speedup to a file"
    else: print "Unable to display speedup plot"
    return

  ff.close()

#
#
if __name__ ==  '__main__':
  import sys
  process(len(sys.argv)-1)