IOPS is mentioned quite often when benchmarking and testing storage systems, but what does it really mean? We discuss and explain what an IOPS is, and how to measure it.

I/O Characterization of TensorFlow with Darshan

Understanding how applications perform I/O is important not only because of the volume of data being written and read, but because the performance of some applications is dependent on how I/O is conducted. In this article we profile I/O at the block layer to help you make the best storage decisions.

The language you choose to use affects I/O patterns and performance. We track a simple write I/O pattern with C and look at how to improve performance.

HPC Storage is arguably one of the most pressing issues in HPC. Selecting various HPC Storage solutions is a problem that requires some research, study, and planning to be effective – particularly cost-effective. Getting this process started usually means understanding how your applications perform I/O. This article presents some techniques for examining I/O patterns in HPC applications.