Cray User Group: Day Two
May. 10th, 2016 11:44 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
sawylShamelessly skipped the morning's plenary sessions — the others went so we had coverage and that sort of high level stuff really isn't my thing any more.
Talked to Altair over lunch and towards the end, WW, TL and I all tried to explain variational assimilation to the scheduling guys. I'm not sure we did a terribly good job, but anything we can do to help the, understand the nature of our work has got to be a good thing.
After lunch I attended the first two IO sessions which convinced me we ought to be running all our jobs with darshan to allow routine characterisation of job IO geometries. While it may not be necessary for everyone, I think it would be useful to have to option to append that type of information to the standard end of job report.
The next couple of talks were on schedulers — both PBS and SLURM — and an interesting example of data collction and visualisation using d3.js and various open tools to make it to easy for users to spot problems with their own jobs.
In the evening we all went to the Folly in Gracechurch Street where we ended up staying until chucking out time. Chatting to some of the others, I discovered that C from the Bureau of Meteorology was a serious trad climber and we had a good chat about all things climbing related.
Talked to Altair over lunch and towards the end, WW, TL and I all tried to explain variational assimilation to the scheduling guys. I'm not sure we did a terribly good job, but anything we can do to help the, understand the nature of our work has got to be a good thing.
After lunch I attended the first two IO sessions which convinced me we ought to be running all our jobs with darshan to allow routine characterisation of job IO geometries. While it may not be necessary for everyone, I think it would be useful to have to option to append that type of information to the standard end of job report.
The next couple of talks were on schedulers — both PBS and SLURM — and an interesting example of data collction and visualisation using d3.js and various open tools to make it to easy for users to spot problems with their own jobs.
In the evening we all went to the Folly in Gracechurch Street where we ended up staying until chucking out time. Chatting to some of the others, I discovered that C from the Bureau of Meteorology was a serious trad climber and we had a good chat about all things climbing related.
