'dif_l': 2.170269e-06,
'dif_w': 4.396577e-06,
'load_l': 2.800742e-06,
'load_w': 6.228898e-05,
'mirr_l': 3.378882e-07,
'mirr_ld': 2.072842e-06,
'mirr_w': 5.311666e-05,
'mirr_wd': 1.979695e-06,
'mirr_wo': 4.983195e-05,
'out_l': 6.257517e-07,
'out_w': 3.441845e-05,
'r_out': 1.402169e+05
}
# Prepare parallel environment
cOS.setVM(MPI(mirrorMap={'*':'.'}))
# 3-sigma target yield
yt=YieldTargeting(
designParams, statParams, opParams,
heads, analyses, measures, variables=variables,
beta=3.0, wcSpecs=wcList,
# Comment out to use default initial point (lo+hi)/2
# initial=atDesign,
initialNominalDesign=True,
# Norms for measures with zero goal
norms={ 'area': 100e-12, 'vgs_drv': 1e-3, 'vds_drv':1e-3 },
tradeoffs=1e-6, # Tradeoff optimization weight, can be overridden in *CbdOptions
stopWhenAllSatisfied=True,
# Initial nominal optimization
initialCbdOptions={
from pyopus.parallel.cooperative import cOS
from pyopus.parallel.mpi import MPI
from funclib import jobProcessor
# Result at which we stop
stopAtResult = 150
# Minimal and maximal number of parallel tasks
# The maximal number of parallel tasks can be infinite (set maxTasks to None)
minTasks = 1
maxTasks = 1000
if __name__ == '__main__':
# Set up MPI
cOS.setVM(MPI())
# Thsi list will hold the jobs (values that are doubled)
jobs = []
# This list will be filled with results
results = []
# Stop the loop
stop = False
# Running task status storage
running = {}
# Job index of next job
atJob = 0
# Outsources tasks, run this example as
# mpirun -n 4 python 02-remote.py
from pyopus.parallel.cooperative import cOS
from pyopus.parallel.mpi import MPI
from funclib import printMsgMPI
if __name__=='__main__':
# Set up MPI
cOS.setVM(MPI())
# Spawn two tasks (locally)
tidA=cOS.Spawn(printMsgMPI, kwargs={'msg': 'Hello A', 'n': 10})
tidB=cOS.Spawn(printMsgMPI, kwargs={'msg': 'Hello B', 'n': 20})
# Spawn two remote tasks
tidC=cOS.Spawn(printMsgMPI, kwargs={'msg': 'Hello C', 'n': 15}, remote=True)
tidD=cOS.Spawn(printMsgMPI, kwargs={'msg': 'Hello D', 'n': 18}, remote=True)
# IDs of running tasks
running=set([tidA,tidB,tidC,tidD])
# Wait for all tasks to finish
while len(running)>0:
# Wait for any task
retval=cOS.Join()
# Wait for specified tasks
# retval=cOS.Join(running)
# Remove IDs of finished tasks
for tid in retval.keys():
atDesign = {
'dif_l': 6.816424e-07,
'dif_w': 3.332037e-06,
'nl_l': 2.655088e-06,
'nl_w': 4.977226e-05,
'nm_l': 3.665018e-06,
'nm_w': 7.507191e-05,
'pm_l': 1.487570e-06,
'pm_w0': 2.871096e-05,
'pm_w1': 2.871096e-05,
'pm_w2': 6.389441e-05,
'pm_w3': 8.310102e-05,
}
# Prepare parallel environment
cOS.setVM(MPI(mirrorMap={'*': '.'}))
# 3-sigma target yield
yt = YieldTargeting(
designParams,
statParams,
opParams,
heads,
analyses,
measures,
variables=variables,
beta=3.0,
wcSpecs=wcList,
# Comment out to use default initial point (lo+hi)/2
# initial=atDesign,
initialNominalDesign=True,
print "Received func=%2d, run=%2d, popsize=%3d "
if writeFhist:
fp=open("fhist_f%d_p%d_r%d.pck" % (funcIndicesList[iFunc], popSizeList[iPopSize], iRun), "wb")
dump(fHistory, fp, protocol=-1)
fp.close()
finalF[iFunc][iPopSize][iRun]=fBest
except:
print "Finished"
if __name__=='__main__':
cOS.setVM(MPI(startupDir=os.getcwd()))
# Prepare results storage
finalF=zeros((len(funcIndicesList), len(popSizeList), nRun))
# Dispatch jobs
cOS.dispatch(
jobList=jobGenerator(),
collector=jobCollector(finalF),
remote=True
)
# Prepare function names
names=[]
dims=[]
for i in funcIndicesList:
shutil.copy2(
'../_MAIN_work/globalVars.py', datadirname +
'/globalVars.py') #copy current globalVars script for logging reasons
shutil.copy2('../_MAIN_work/scoreFunctions.py',
datadirname + '/scoreFunctions.py')
output = open("data.pkl", "wb")
os.chdir("../_MAIN_work")
# Set up MPI for parallel computing
cOS.setVM(
MPI(
mirrorMap={
'models.inc': '.',
'topdc.cir': '.',
'topdc_psrr.cir': '.',
'mosmm.inc': '.',
'cmos180n.lib': '.',
#'testTopCirc_hspice_AF.cir':'.',
#'testTopCirc_hspice_AF_outimped.cir':'.'
}))
generationNum = 0
optimizedIndivids = [] #PSADE parameter-optimized individuals
hotGen = population() #hot generation (current)
oldGen = population() #previous generation (old)
currentBestScore = np.Inf
bestScoresList = []
averageScoresList = []
