Optimizing Scalable RIP

This page applies to Harlequin v13.1r0 and later; and to Harlequin Core but not Harlequin MultiRIP

Optimizing performance of the Scalable RIP follows the same principles as optimizing single RIP performance: Utilize as much memory, processor performance, and I/O bandwidth of the machine as possible.

In general, the best performance will be when running one Farm RIP per hardware thread. If other significant processing is being performed on the machine, deduct the threads required to support this processing first.

The Scalable RIP divides the memory given to it between all of the RIP instances. The controlling RIP takes a minimal allocation in addition to the amount specified for dividing between the farm RIPs. The Scalable RIP should be given as much memory as possible without interfering with other processing required on the machine. Memory swapping will severely impact performance.

One of the key areas where the Scalable RIP may differ from a single RIP is that it is quite easy to run out of I/O bandwidth. We recommend that the output from the Scalable RIP is put on a dedicated disk. The output disk should not be shared with the input files, or the disk containing the SW folder. If possible, locate input files, SW folder, and output files all on separate disks. The output disk should have a high write bandwidth.