Extracts relevant data from the bunch of benchmark simulations we had and saves the data into an excel spreadsheet

431_p3_dataextract.py

benchmarkFilesDir = "/Users/georgepetrou/Desktop/431_pe3/raw"
outputExcelFilePath = "/Users/georgepetrou/Desktop/431_pe3/output.xlsx"

## Dont edit below here
import os
import mmap
import xlsxwriter

def processData():
	processedData = {}
	benchmarkFiles = [file for file in os.listdir(benchmarkFilesDir) if os.path.isfile(os.path.join(benchmarkFilesDir, file)) and file.endswith(".simout")]
	for fileName in sorted(benchmarkFiles):
		fullBenchmarkFilePath = os.path.join(benchmarkFilesDir, fileName)
		extractDataFromFile(fullBenchmarkFilePath, fileName, processedData)
	return processedData

infoWeWant = ["bpred_nottaken.misses", "bpred_bimod.misses", "bpred_2lev.misses", "bpred_comb.misses", "sim_CPI", "sim_num_insn"]
def extractDataFromFile(path, fileNameOnly, processedData):
	processedData[fileNameOnly] = {}

	with open(path, 'r') as file:
		numMisses = 0
		numInstructions = 0

		for line in file:
			if any(x in line for x in infoWeWant):
				info = line.strip().split(None, 2)[:-1]

				if "misses" in info[0]:
					numMisses = float(info[1])
					processedData[fileNameOnly]["Misses"] = numMisses
				elif "sim_CPI" in info[0]:
					numInstructions = float(info[1])
					processedData[fileNameOnly]["CPI"] = numInstructions

		if numMisses > 0 and numInstructions > 0:
			missPerInstr = "{0:.6f}".format(numInstructions/numMisses)
			processedData[fileNameOnly]["MissPerInstr"] = missPerInstr
		else:
			processedData[fileNameOnly]["MissPerInstr"] = 0
		if numMisses <= 0:
			processedData[fileNameOnly]["Misses"] = 0

def handleExcelOutput():
	workbook = xlsxwriter.Workbook(outputExcelFilePath)
	worksheet = workbook.add_worksheet()
	curRow = 1

	for fileName, fileData in sorted(processedData.items()):
		worksheet.write(curRow, 0, fileName[:-7])
		worksheet.write(curRow, 1, fileData.get("CPI"))
		worksheet.write(curRow, 2, fileData.get("Misses"))
		worksheet.write(curRow, 3, fileData.get("MissPerInstr"))
		curRow += 1
	workbook.close()

if __name__ == "__main__":
	processedData = processData()
	handleExcelOutput()