Working on our final project, we have been running into this exact issue. The overhead involved in setting up a multithreaded program doesn't outweigh the performance improvements, especially in artificial intelligence. It is a sobering thought when thinking about how applicable parallelization actually is in our every day lives.
The results in the right column are from the BID Data Suite paper by Canny and Zhao: Big Data Analytics with Small Footprint: Squaring the Cloud (KDD 13).
Working on our final project, we have been running into this exact issue. The overhead involved in setting up a multithreaded program doesn't outweigh the performance improvements, especially in artificial intelligence. It is a sobering thought when thinking about how applicable parallelization actually is in our every day lives.