adding linear and utilizationn plotter

linear/linear_run.py

@@ -0,0 +1,68 @@
+import time
+from ptflops import get_model_complexity_info
+
+def run(point):
+   start = time.time()
+   try:
+      batch_size = point['batch_size']
+      in_features = point['in_features']
+      out_features = point['out_features']
+      bias = int(point['bias']) == 1
+      omp_num_threads = point['omp_num_threads']
+      print(point)
+
+      import os
+      os.environ['OMP_NUM_THREADS'] = str(omp_num_threads)
+      os.environ['MKL_NUM_THREADS'] = str(omp_num_threads)
+      os.environ['KMP_HW_SUBSET'] = '1s,%sc,2t' % str(omp_num_threads)
+      os.environ['KMP_AFFINITY'] = 'granularity=fine,verbose,compact,1,0'
+      os.environ['KMP_BLOCKTIME'] = str(0)
+      os.environ['MKLDNN_VERBOSE'] = str(1)
+      os.environ['MKL_VERBOSE'] = str(1)
+      import torch
+
+      print('torch version: ',torch.__version__,' torch file: ',torch.__file__)
+
+      inputs = torch.arange(batch_size * in_features,dtype=torch.float).view((batch_size,in_features))
+
+      layer = torch.nn.Linear(in_features,out_features,bias=bias)
+      flops, params = get_model_complexity_info(layer, tuple(inputs.shape),as_strings=False)
+      print(flops)
+
+      outputs = layer(inputs)
+
+      runs = 5
+      tot_time = 0.
+      tt = time.time()
+      for _ in range(runs):
+         outputs = layer(inputs)
+         tot_time += time.time() - tt
+         tt = time.time()
+
+      ave_time = tot_time / runs
+
+      print('flop = ',flops,'ave_time = ',ave_time)
+
+      ave_flops = flops / ave_time  #* batch_size
+
+      print('runtime=',time.time() - start)
+      return ave_flops
+   except Exception as e:
+      import traceback
+      print('received exception: ',str(e))
+      print(traceback.print_exc())
+      print('runtime=',time.time() - start)
+      return 0.
+
+
+if __name__ == '__main__':
+   point = {
+      'batch_size': 10,
+      'in_features': 512,
+      'out_features': 512,
+      'bias': 1,
+      'omp_num_threads':64,
+   }
+
+   print('flops for this setting =',run(point))
+

linear/problem.py

@@ -0,0 +1,13 @@
+from deephyper.benchmark import HpProblem
+
+Problem = HpProblem()
+Problem.add_dim('batch_size',(1,8192))
+Problem.add_dim('in_features',(128,8192))
+Problem.add_dim('out_features',(128,8192))
+Problem.add_dim('omp_num_threads',(8,64))
+Problem.add_dim('bias',[0,1])
+
+Problem.add_starting_point(batch_size=128,in_features=1024,out_features=512,omp_num_threads=64,bias=0)
+
+if __name__ == '__main__':
+   print(Problem)

utilization.py

@@ -0,0 +1,44 @@
+#!/usr/bin/env python
+import argparse,logging
+from balsam.core.models import utilization_report, BalsamJob, process_job_times
+import matplotlib.pyplot as plt
+logger = logging.getLogger(__name__)
+
+
+def main():
+   ''' simple starter program that can be copied for use when starting a new script. '''
+   logging_format = '%(asctime)s %(levelname)s:%(name)s:%(message)s'
+   logging_datefmt = '%Y-%m-%d %H:%M:%S'
+   logging_level = logging.INFO
+
+   parser = argparse.ArgumentParser(description='')
+   parser.add_argument('-w','--workflow',help='workflow name to analyze',required=True)
+   parser.add_argument('--debug', dest='debug', default=False, action='store_true', help="Set Logger to DEBUG")
+   parser.add_argument('--error', dest='error', default=False, action='store_true', help="Set Logger to ERROR")
+   parser.add_argument('--warning', dest='warning', default=False, action='store_true', help="Set Logger to ERROR")
+   parser.add_argument('--logfilename',dest='logfilename',default=None,help='if set, logging information will go to file')
+   args = parser.parse_args()
+
+   if args.debug and not args.error and not args.warning:
+      logging_level = logging.DEBUG
+   elif not args.debug and args.error and not args.warning:
+      logging_level = logging.ERROR
+   elif not args.debug and not args.error and args.warning:
+      logging_level = logging.WARNING
+
+   logging.basicConfig(level=logging_level,
+                       format=logging_format,
+                       datefmt=logging_datefmt,
+                       filename=args.logfilename)
+
+   qs = BalsamJob.objects.filter(workflow=args.workflow)
+   dat = process_job_times(qs)
+   times, utils = utilization_report(dat)
+
+   plt.step(times, utils, where="post")
+   plt.waitforbuttonpress()
+
+
+
+if __name__ == "__main__":
+   main()