Project 8 Summary
November 22, 2009
Predicted number of hours: 10hrs
Actual number of hours: 10hrs
Project 7 Summary
November 15, 2009
Project 6 Summary
November 1, 2009
Project 5 Summary
October 18, 2009
Project 4 Summary
October 4, 2009
Project 3 Summary
September 27, 2009
Predicted number of hours: 5hrs
Actual number of hours: 4hrs
Execution Time: 0.196
Rank: 35
Project 2 Summary
September 10, 2009
Predicted hours to complete: 2hrs
Actual hours to complete: 3hrs
Execution time: 0.020
Rank: 59
Project 1 Summary
September 3, 2009
Predicted hours to complete: 10hrs
Actual hours to complete: 15hrs
Execution time: 0.008
Rank: 144
Yay!
September 2, 2009
Got it all working in Windows + Eclipse + Cygwin. Cygwin is fairly convenient due to it coming with cppunit and the dl library and stuff. The main problem that I ran into was that the unit testing code given to us wasn’t working. Turns out that I just had to add -c -fmessage-length=0 to the compiler flags that were there by default but I removed them earlier. Now I need to ensure that all this stuff works with the CS labs’ computers.
Finished a smarter version of the program too by using a precompiled metacache that cached the cycles from 1-100000 in slices of 100. Made an automated slicer that given a max and an interval, spit out to the console a string that allowed me to change the slices metacache easily. For example, for my metacache, it spit out ” { 119, 125, ….” and so on. Also played around with the metacache size and interval and I have not found a more efficient maximum and interval.
Afterwards, I tried a lot of other optimizations to see if I could improve the speed any further. My first one was a precompiled cache of power of twos and their cycle lengths. That one actually slowed it down. The second was a precache that calculated cycles from 1-1000. Even after playing with the range, the best I got was to not adversely affect the running time. Then I added a cache with an offset. Because I had a precache and a metacache, I tried to see if caching values larger than my slice max of 100000 would increase the running time. It didn’t but it didn’t adversely affect it either. Lastly, I tried to precache the slices metacache instead of precompiling it in order to get a bigger range. But even with a larger range, the run time slowed down instead of speeding up. This leads me to believe that UVa does not feed it values much about 100000.
Current run time is 0.008.
Windows + Stuff = Pain.
September 1, 2009
I might have to up my estimate to 10hrs or so. Having trouble setting up my environment due to some problem when tests are enabled. Using Eclipse + MinGW. Weird thing is that a simple program I found for cppUnit seems to work. Oh well, I’ll look into it later.
Finished the dumb version of the program too. Not too sure if it works since the UVa Online Judging thing doesn’t seem to want to send me a confirmation email.