True, for a definition of micro-benchmark you have decided for yourself rather than asked me to clarify....
On 7 April 2013 09:37, Patricia Shanahan <p...@acm.org> wrote: > On 4/7/2013 1:04 AM, Dan Creswell wrote: > >> On 7 April 2013 05:24, Patricia Shanahan <p...@acm.org> wrote: >> >> On 4/6/2013 7:26 PM, Greg Trasuk wrote: >>> ... >>> >>> Once we have a stable set of regression tests, then OK, we could >>> >>>> think about improving performance or using Maven repositories as the >>>> codebase server. >>>> >>>> ... >>> >>> I think there is something else you need before it would be a good idea >>> to >>> release any changes for the sake of performance - some examples of >>> workloads whose performance you want to improve, and that are in fact >>> improved by the changes. >>> >>> >> Indeed. And if we do these changes in a confined, one small step at a time >> type fashion, we can build micro-benchmarks as we go along. >> >> Later, if it makes sense we can look at more macro tests such as e.g. >> Lookup or JavaSpace or Transaction etc. >> >> > Programmer constructed micro-benchmarks are good at answering the > question "How am I doing at optimizing X?". They are useless for > answering the question that needs to be asked first: "What, if anything, > should I be optimizing?". > > To find out what, if anything, needs optimization one has to start from > real workloads that are running slower than desired, and measure them. > Or, in some cases, from industry benchmarks that are known to be > correlated with real workloads in some field. > > Patricia > >