Echo commit hash used for cloning IOR and MDREAL

[vnobody]

Hello.
Probably it would be good to echo commit hashes while cloning IOR and MDREAL repositories in prepare.sh? Currently it clones the repos and checks out the latest commits by default. Reason for this change might be, when testing multiple systems in different days/months and installing fresh version, there might be a situation, that results are with big difference and users might wonder why this happened. In that case, the reason will be, that different versions of IOR or MDREAL were used.

[adilger]

It also makes sense to add a git tag to the repository before when IO-500 list is published at ISC and SC. While this does not guarantee which version of the script or tools was run for a given benchmark result (the actual commit hashes for IOR and io-500-dev should be included in the result submission) it does give an approximate state of affairs at each publication date for future reference.

At some point, when the IO-500 scripts have stabilized more, and there are more participants submitting results, it might make sense to mandate a specific version of the scripts and IOR for a given list, so that everyone has a level playing field.

[JulianKunkel]

Absolutely true. Inside the script we initially to clone a particular version. However, as we worked on improving IOR there were too many changes. I believe the problem will be simply fixed by using only "release" versions. These versions of mdtest/ior also output there version information hence we have it. Hopefully, we resolve the stuff soon for the mdtest/IOR versions and make a new release and simply use it. 2018-08-09 21:59 GMT+01:00 adilger <[email protected]>:

…

It also makes sense to add a git tag to the repository before when IO-500 list is published at ISC and SC. While this does not guarantee which version of the script or tools was run for a given benchmark result (the actual commit hashes for IOR and io-500-dev should be included in the result submission) it does give an approximate state of affairs at each publication date for future reference. At some point, when the IO-500 scripts have stabilized more, and there are more participants submitting results, it might make sense to mandate a specific version of the scripts and IOR for a given list, so that everyone has a level playing field. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#29 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE1uyplKMrZWWtnSwturWfNFCwQEfMMTks5uPKK-gaJpZM4V2C1x> .

-- Dr. Julian Kunkel Lecturer, Department of Computer Science +44 (0) 118 378 8218 http://www.cs.reading.ac.uk/ https://hps.vi4io.org/