cvw/synthDC/extractSummary.py

266 lines
9.3 KiB
Python
Raw Normal View History

2022-03-03 00:22:14 +00:00
#!/usr/bin/python3
2022-06-24 00:21:39 +00:00
# Madeleine Masser-Frye (mmmasserfrye@hmc.edu) 06/2022
from collections import namedtuple
import re
import csv
2022-06-24 00:21:39 +00:00
import subprocess
from matplotlib.cbook import flatten
import matplotlib.pyplot as plt
import matplotlib.lines as lines
2022-07-07 16:08:21 +00:00
import numpy as np
from ppa.ppaAnalyze import noOutliers
from matplotlib import ticker
import argparse
2022-07-17 16:06:30 +00:00
import os
2022-06-24 00:21:39 +00:00
def synthsintocsv():
''' writes a CSV with one line for every available synthesis
each line contains the module, tech, width, target freq, and resulting metrics
'''
print("This takes a moment...")
bashCommand = "find . -path '*runs/wallypipelinedcore_*' -prune"
output = subprocess.check_output(['bash','-c', bashCommand])
allSynths = output.decode("utf-8").split('\n')[:-1]
specReg = re.compile('[a-zA-Z0-9]+')
metricReg = re.compile('-?\d+\.\d+[e]?[-+]?\d*')
file = open("Summary.csv", "w")
writer = csv.writer(file)
writer.writerow(['Width', 'Config', 'Special', 'Tech', 'Target Freq', 'Delay', 'Area'])
for oneSynth in allSynths:
2022-07-07 16:08:21 +00:00
descrip = specReg.findall(oneSynth)
2022-06-24 00:21:39 +00:00
width = descrip[2][:4]
config = descrip[2][4:]
if descrip[3][-2:] == 'nm':
special = ''
else:
special = descrip[3]
descrip = descrip[1:]
tech = descrip[3][:-2]
freq = descrip[4]
metrics = []
for phrase in ['Path Slack', 'Design Area']:
bashCommand = 'grep "{}" '+ oneSynth[2:]+'/reports/*qor*'
bashCommand = bashCommand.format(phrase)
try:
output = subprocess.check_output(['bash','-c', bashCommand])
nums = metricReg.findall(str(output))
nums = [float(m) for m in nums]
metrics += nums
except:
2022-06-28 02:28:13 +00:00
print(width + config + tech + '_' + freq + " doesn't have reports")
2022-06-24 00:21:39 +00:00
if metrics == []:
pass
else:
delay = 1000/int(freq) - metrics[0]
area = metrics[1]
writer.writerow([width, config, special, tech, freq, delay, area])
file.close()
2022-07-17 18:00:44 +00:00
2022-06-24 00:21:39 +00:00
def synthsfromcsv(filename):
2022-06-28 02:28:13 +00:00
Synth = namedtuple("Synth", "width config special tech freq delay area")
2022-06-24 00:21:39 +00:00
with open(filename, newline='') as csvfile:
csvreader = csv.reader(csvfile)
global allSynths
allSynths = list(csvreader)[1:]
for i in range(len(allSynths)):
for j in range(len(allSynths[0])):
try: allSynths[i][j] = int(allSynths[i][j])
except:
try: allSynths[i][j] = float(allSynths[i][j])
except: pass
allSynths[i] = Synth(*allSynths[i])
return allSynths
2022-07-17 18:00:44 +00:00
2022-06-24 00:21:39 +00:00
def freqPlot(tech, width, config):
''' plots delay, area for syntheses with specified tech, module, width
'''
current_directory = os.getcwd()
final_directory = os.path.join(current_directory, 'plots/wally')
2022-07-17 18:00:44 +00:00
if not os.path.exists(final_directory):
os.makedirs(final_directory)
2022-06-24 00:21:39 +00:00
freqsL, delaysL, areasL = ([[], []] for i in range(3))
for oneSynth in allSynths:
if (width == oneSynth.width) & (config == oneSynth.config) & (tech == oneSynth.tech) & ('' == oneSynth.special):
2022-06-24 00:21:39 +00:00
ind = (1000/oneSynth.delay < oneSynth.freq) # when delay is within target clock period
freqsL[ind] += [oneSynth.freq]
delaysL[ind] += [oneSynth.delay]
areasL[ind] += [oneSynth.area]
2022-07-07 16:08:21 +00:00
fig, (ax1, ax2) = plt.subplots(2, 1, sharex=True)
allFreqs = list(flatten(freqsL))
if allFreqs != []:
median = np.median(allFreqs)
else:
median = 0
2022-06-24 00:21:39 +00:00
for ind in [0,1]:
areas = areasL[ind]
delays = delaysL[ind]
freqs = freqsL[ind]
2022-07-07 16:08:21 +00:00
freqs, delays, areas = noOutliers(median, freqs, delays, areas)
2022-06-24 00:21:39 +00:00
c = 'blue' if ind else 'green'
2022-07-07 16:08:21 +00:00
targs = [1000/f for f in freqs]
ax1.scatter(targs, delays, color=c)
ax2.scatter(targs, areas, color=c)
2022-06-24 00:21:39 +00:00
freqs = list(flatten(freqsL))
delays = list(flatten(delaysL))
areas = list(flatten(areasL))
legend_elements = [lines.Line2D([0], [0], color='green', ls='', marker='o', label='timing achieved'),
lines.Line2D([0], [0], color='blue', ls='', marker='o', label='slack violated')]
ax1.legend(handles=legend_elements)
2022-06-24 01:14:53 +00:00
ytop = ax2.get_ylim()[1]
ax2.set_ylim(ymin=0, ymax=1.1*ytop)
2022-07-07 16:08:21 +00:00
ax2.set_xlabel("Target Cycle Time (ns)")
ax1.set_ylabel('Cycle Time Achieved (ns)')
2022-06-24 00:21:39 +00:00
ax2.set_ylabel('Area (sq microns)')
2022-07-07 16:08:21 +00:00
ax1.set_title(tech + ' ' + width + config)
ax2.yaxis.set_major_formatter(ticker.StrMethodFormatter('{x:,.0f}'))
addFO4axis(fig, ax1, tech)
2022-06-24 01:14:53 +00:00
plt.savefig('./plots/wally/freqSweep_' + tech + '_' + width + config + '.png')
2022-06-24 00:21:39 +00:00
2022-07-07 16:08:21 +00:00
def areaDelay(tech, delays, areas, labels, fig, ax, norm=False):
2022-07-07 16:08:21 +00:00
plt.subplots_adjust(left=0.18)
2022-06-24 00:21:39 +00:00
fo4 = techdict[tech].fo4
add32area = techdict[tech].add32area
marker = techdict[tech].shape
color = techdict[tech].color
2022-07-07 16:08:21 +00:00
if norm:
delays = [d/fo4 for d in delays]
areas = [a/add32area for a in areas]
2022-06-24 00:21:39 +00:00
plt.scatter(delays, areas, marker=marker, color=color)
2022-07-07 16:08:21 +00:00
plt.xlabel('Cycle time (ns)')
2022-06-24 00:21:39 +00:00
plt.ylabel('Area (sq microns)')
2022-07-07 16:08:21 +00:00
ytop = ax.get_ylim()[1]
2022-06-24 01:14:53 +00:00
plt.ylim(ymin=0, ymax=1.1*ytop)
2022-07-07 16:08:21 +00:00
ax.yaxis.set_major_formatter(ticker.StrMethodFormatter('{x:,.0f}'))
2022-06-24 00:21:39 +00:00
for i in range(len(labels)):
plt.annotate(labels[i], (delays[i], areas[i]), textcoords="offset points", xytext=(0,10), ha='center')
return fig
2022-07-17 18:00:44 +00:00
def plotFeatures(tech, width, config):
delays, areas, labels = ([] for i in range(3))
freq = techdict[tech].targfreq
for oneSynth in allSynths:
if (tech == oneSynth.tech) & (freq == oneSynth.freq):
if (oneSynth.config == config) & (width == oneSynth.width):
delays += [oneSynth.delay]
areas += [oneSynth.area]
labels += [oneSynth.special]
fig, (ax) = plt.subplots(1, 1)
fig = areaDelay(tech, delays, areas, labels, fig, ax)
titlestr = tech+'_'+width+config
plt.title(titlestr)
plt.savefig('./plots/wally/features_'+titlestr+'.png')
2022-07-17 18:00:44 +00:00
def plotConfigs(tech, special=''):
delays, areas, labels = ([] for i in range(3))
freq = techdict[tech].targfreq
for oneSynth in allSynths:
if (tech == oneSynth.tech) & (freq == oneSynth.freq) & (oneSynth.special == special):
delays += [oneSynth.delay]
areas += [oneSynth.area]
labels += [oneSynth.width + oneSynth.config]
fig, (ax) = plt.subplots(1, 1)
fig = areaDelay(tech, delays, areas, labels, fig, ax)
titleStr = tech+'_'+special
plt.title(titleStr)
plt.savefig('./plots/wally/configs_' + titleStr + '.png')
def normAreaDelay(special=''):
fig, (ax) = plt.subplots(1, 1)
fullLeg = []
for tech in list(techdict.keys()):
delays, areas, labels = ([] for i in range(3))
spec = techdict[tech]
freq = spec.targfreq
for oneSynth in allSynths:
if (tech == oneSynth.tech) & (freq == oneSynth.freq) & (oneSynth.special == special):
delays += [oneSynth.delay]
areas += [oneSynth.area]
labels += [oneSynth.width + oneSynth.config]
areaDelay(tech, delays, areas, labels, fig, ax, norm=True)
fullLeg += [lines.Line2D([0], [0], markerfacecolor=spec.color, label=tech, marker=spec.shape, markersize=10, color='w')]
2022-07-07 16:08:21 +00:00
ax.set_title('Normalized Area & Cycle Time by Configuration')
ax.set_xlabel('Cycle Time (FO4)')
ax.set_ylabel('Area (add32)')
2022-07-07 16:08:21 +00:00
ax.legend(handles = fullLeg, loc='upper left')
plt.savefig('./plots/wally/normAreaDelay.png')
2022-07-17 18:00:44 +00:00
2022-07-07 16:08:21 +00:00
def addFO4axis(fig, ax, tech):
fo4 = techdict[tech].fo4
2022-07-07 16:08:21 +00:00
ax3 = fig.add_axes((0.125,0.14,0.775,0.0))
ax3.yaxis.set_visible(False) # hide the yaxis
fo4Range = [x/fo4 for x in ax.get_xlim()]
dif = fo4Range[1] - fo4Range[0]
for n in [0.02, 0.05, 0.1, 0.25, 0.5, 1, 2, 5, 10, 20, 50, 100, 200, 500, 1000]:
d = dif/n
if d > 3 and d < 10:
r = [int(x/n) for x in fo4Range]
nsTicks = [round(x*n, 2) for x in range(r[0], r[1]+1)]
break
new_tick_locations = [fo4*float(x) for x in nsTicks]
ax3.set_xticks(new_tick_locations)
ax3.set_xticklabels(nsTicks)
ax3.set_xlim(ax.get_xlim())
ax3.set_xlabel("FO4 delays")
plt.subplots_adjust(left=0.125, bottom=0.25, right=0.9, top=0.9)
2022-06-24 00:21:39 +00:00
if __name__ == '__main__':
2022-07-07 16:08:21 +00:00
parser = argparse.ArgumentParser()
parser.add_argument("-s", "--skyfreq", type=int, default=3000, help = "Target frequency used for sky90 syntheses")
parser.add_argument("-t", "--tsmcfreq", type=int, default=10000, help = "Target frequency used for tsmc28 syntheses")
args = parser.parse_args()
TechSpec = namedtuple("TechSpec", "color shape targfreq fo4 add32area add32lpower add32denergy")
techdict = {}
techdict['sky90'] = TechSpec('green', 'o', args.skyfreq, 43.2e-3, 1440.600027, 714.057, 0.658023)
techdict['tsmc28'] = TechSpec('blue', 's', args.tsmcfreq, 12.2e-3, 209.286002, 1060.0, .081533)
2022-07-07 16:08:21 +00:00
synthsintocsv()
2022-06-24 00:21:39 +00:00
synthsfromcsv('Summary.csv')
2022-06-28 02:28:13 +00:00
freqPlot('tsmc28', 'rv32', 'e')
freqPlot('sky90', 'rv32', 'e')
plotFeatures('sky90', 'rv64', 'gc')
plotFeatures('tsmc28', 'rv64', 'gc')
plotConfigs('sky90', special='orig')
plotConfigs('tsmc28', special='orig')
normAreaDelay(special='orig')