Euler 50

gistfile1.txt

# set up prime sieve to n- track running sum in separate sum array
total = 0
n=1000000
myPrimes = [True]*(n+1)
mySum = [0]*(n+1)
last = 0
for i in range (2,n):
	if myPrimes[i] == True:
		j = 2*i
		while j<=n:
			myPrimes[j]=False
			j=j+i	
		mySum[i]=last+i
		last = mySum[i] 

# notice that after mySum[3943], you've gone over a million- it's composite.
# so start there and subtract the lowest sum, mySum[2]
# if it's prime and below 1MM, stop- if it's not, try subtracting mySum[3]...

s = set()
for l in range (3900,3943):
		for k in range (2,100): 
			if myPrimes[mySum[k]]!=True:
				test = mySum[l]-mySum[k]
				if myPrimes[test] == True:
					s.add(test)
print max(s)

Still a little manual- but it does run in about 2s, as opposed to 55s for the last version. And you could do it for any length n- simply by looking up what number the sum goes over n at.