|Week||What We Plan To Do||What We Actually Did|
|Apr 1-7||Develop the proposal||Finished the proposal|
|Apr 8-14||Write and test a serial implementation||Successfully wrote a serial implementation.|
|Apr 15-21||Begin writing and testing a parallel implementation||Wrote global memory implementations|
|Apr 22-28||Analyze our current results, and implement optimizations||Analyzed global memory performance, implemented shared memory version|
|Apr 29-May 5||Test using real world data, compare results and performance to existing alternatives||Analyzed results of both versions|
|May 6-11||Write the final presentation||Added minor optimizations, wrote final presentation and report|
[Keep a log of work you have done here. You may wish to list optimizations you tried, what failed, etc. Or you can just rant. Keeping a good log will make it easy to put together your final writeup.]We wrote a serial implementation which is very slow. We met and discussed how to decompose our existing code using CUDA and Thrust, and tried trivially parallelizing our code using OpenMP (unsuccessfully). Also, we discussed making our problem more interesting and practical by using data from previous stocks and running our algorithm to predict prices, and then seeing how the stocks actually turned out.