How to compare MapReduce on GPUs and Apache Spark?
Hey Ananya, GPU’s have definitely extended beyond their traditional base in scientific HPC centers. The reason for using GPUs is that they get into the underlining server and whatever is running on a single node runs faster. So, it doesn’t change the fundamental approach to solving a problem, but the algorithm behind the problem can be more complex with the added performance which leads to better accuracy for everything from pattern matching to fine tuned classification. When MapReduce is coupled with the speedup on GPU nodes, the performance permits far better accuracy with the added algorithmic complexity the system can handle. So in essence the advantage of using GPU’s and MapReduce together is that the overall speed and accuracy of the MapReduce job improves. But MapReduce still remains suitable for Batch Processing scenarios and Spark for real time analytics. This is due to the way data is proccessed by each of them.