Choose timezone
Your profile timezone:
Developer's meeting
→
Europe/Zurich
Present: Enrico, Milo, Frank
EHN1 (Enrico):
- (see slides)
- nice diagrams elaborating a bit on the various steps and interactions, in the schemes both without and with WIB
- there has not yet been an opportunity to test with the latter; apparently there is not really a WIB available to us at present (in spite of earlier conclusions) and it is to be seen when there will be
- first tests with the former have been done, but upon reading back the data stored to disk it seems that there is still something incorrect (as only a single frame is read back while multiple frames should have been stored)
- 100 Gb/s Mellanox cards:
- 94 Gb/s achieved by Giovanna in a separate test (using servers 17, 18)
- however, attempting to use both the Mellanox cards and the BNL-711 card in server 19 resulted in a system crash, necessitating a system re-install
- hope is that there no clash between resource requirements from the sides of both cards
- idea is to test once more, but this needs "quiet time" (no ongoing other tests) and Giovanna's availability
Data re-ordering and compression (Milo):
- (see slides)
- nice comparison of compression results (both compression factor and time needed for the compression proper) between different schemes
- find that a simple data re-ordering (i.e., no pedestal subtraction) leads to a good compression factor and minimal compression time
- this is for uBooNE MC, to be seen whether coherent noise changes the picture substantially
- notes:
- times do not include the time needed for any data re-ordering, which was discussed last time
- compression factors are not absolute results since not all advanced options are switched on; however, expect the qualitative results to remain valid
- link to re-ordered data format can be found here
The tentative plan is to meet again on November 29.
There are minutes attached to this event.
Show them.