Quick update

It's been a while since my last post and I thought a short update was in order. I have been very busy this week, planning a vacation, working on a summer class, and learning CUDA. I spent two days in the lab and finished up my matrix multiplication code. I'm ready to start Monday on a new application that will do minimal and maximal array reductions. This is an important next step because it requires my kernel code to make logical decisions that will directly impact the efficiency of the algorithms. I have learned, and will explain in detail, that "for loops" and "if else statements" can dramatically increase computation time. Also I am learning the importance of using shared memory and the device registers to make code more efficient. Thankfully the CUDA SDK come with a very useful visual profiler that plots data showing how efficiently my code is running. I will have screen shots and full details this Tuesday so stay tuned!