On 09.02.23 11:06, Hans Holmberg wrote: > It takes a significant amount of time and trouble to build, run and understand > benchmarks for these applications. Modeling the workloads using fio > minimizes the set-up work and would enable more developers to actually > run these things. The workload definitions could also help developers > understanding what sort of IO that these use cases generate. True, but I think Adam has a point here. IIRC mmtests comes with some scripts to download, build and run the desired applications and then do the maths. In this day and age people would probably want to use a container with the application inside and some automation around it to run the benchmark and present the results. Don't get me wrong. Simple fio snippets are definitively what I would prefer as well, but creating these does require a lot of insight into the inner workings of a desired user-space workload. Of cause once that is achieved we can easily add this workload to fsperf and have some sort of CI testing for it to make sure we don't regress here.
