EEHPCWG / PowerMeasurementMethodology

TeX for PMM spec
8 stars 0 forks source link

Core phase and Measurement Reporting Frequency #80

Closed nbates closed 9 years ago

nbates commented 9 years ago

5) I think the requirement of covering a significant temporal part of the core phase is very important. Figure 2.1 motivates this - but right now the way I read the document only the whole run number also needs to contain the whole core phase. Minimum interval length of the measured part of the core phase is only mentioned for Level 1 and only in Figure 2.3. For Level 2 and 3 I would suggest to require that no more than a total of two seconds (one each at begin and end) of the core phase are not covered by the total average power / energy. This matches the requirement of 1 samples/second. Also the reported average power / energy values should not only be required to be evenly spaced, but evenly cover the core phase, e.g. the both time between the start of core phase and the fist reported measurement as well as the the last reported measurement and the end of the core phase should not be larger than the time between two successive reported measurements. I recall this was also a discuss ion point at SC14, but I'm not sure how it was concluded.

Thomas Ilsche email 10 July 2015