The PIAF parallel commands
The PIAF parallel commands
- PIAF parallel, e.g. NT/PLOT
- The Server receives a Client request to NT/PLOT the nTuple
- Each Slave, based on its position in the group, decides on the portion of the nTuple to read
- Each Slave reads that portion of the nTuple from the PIAF disks into local memory (unless the portion is > 54 MB)
- The maximum total data size cached is thus around 400 MB
- Other nTuples already in the cache are kept, unless cache space runs out
- The disk access is achieved via NFS on each PIAF node
- The Slaves operate on the nTuple and write histogram(s) and output to their respective log files