Dell PowerEdge R940xa GPU Database Acceleration on - Page 6

2 What is GPU acceleration and how does it apply to databases? "Think of the GPU as a coin press machine, which can punch out 100 coins with a single operation every four seconds, whereas a CPU is a coin press which can punch out 1 coin per operation every one second. While the CPU has a faster "punch time", the GPU can punch more coins per minute. This is the key difference between the GPU and CPU. The GPU is throughput oriented, while the CPU is latency oriented." The GPU is therefore well suited for operations that perform the same instruction on large amounts of data at once. Put it simply, a GPU database is a database, relational or non-relational, that uses a GPU (graphical processing unit) to perform some database operations and because they are throughput orientated they are typically very fast. GPU databases are flexible and can process many different types of data, or much larger amounts of data. 2.1 Why GPU? GPU Flop Comparison vs. CPU [source: NVIDIA] GPUs are highly parallel hardware accelerators originally designed to accelerate the creation of computer graphics. More recently, folks have been looking at GPUs to accelerate other workloads like Database analytics and On Line Analytic Processing (OLAP). Although GPUs are great for accelerating analytics, they have little or no use for transactional (OLTP) style workloads. GPUs are faster at large numbers of numerical computations than CPUs, whereas CPUs outperform GPUs for tasks that are hard to parallelize or that involve complex control flow instructions. The GPUs small, efficient cores are better suited to performing similar, repeated instructions in parallel, making it suitable for accelerating process-intensive workloads in data analysis applications. Another characteristic of GPUs is their memory bandwidth is much higher, up to 900GB/second when compared to CPUs which is around 68GB/second. The combination of linking together several GPU devices, each with super-fast fast I/O and several-thousand cores means very high rates of single-precision performance can be achieved. 6 GPU Database Acceleration on PowerEdge R940xa