Harlequin® Parallel Pages

Interpreting and rendering in parallel

Harlequin RIPs have supported various forms of parallel processing for many years, and Harlequin Host Renderer took another big step in multi-threading with the addition of Harlequin Parallel Pages in Version 3.

five CPU cores share processing to reduce total output time
Multiple threads - Multiple threads are used to perform the various steps of RIPping and outputting the pages of a job at once; there are no longer any periods when only a single thread is active.

Harlequin Parallel Pages, decouples interpretation from compositing and rendering, and allows one thread to interpret page two of a job while the other threads are still compositing, rendering and outputting page one. This ensures that processor cores are kept busy, ensuring the highest possible throughput.

This gives:
  • a performance boost in single-RIP integrations, allowing them to be used to drive light production and medium volume devices at higher page rates avoiding the need for implementing a RIP farm for light and medium production, reducing development costs and time to market.
  • it also enables more efficient utilization of CPU cores in configurations using a ‘farm’ of multiple RIPs, ensuring that all cores are kept busy for a larger proportion of the time without risking lost performance caused by clashing requests for CPU time from too many threads at once. High volume devices can be driven with fewer RIPs on fewer computers, reducing the bill of materials for hardware and operating systems for your DFE. It also avoids any need for custom hardware.
two traces, one with Parallel pages, one without
Performance - The top trace shows processor activity for Harlequin Host Renderer using multithreaded compositing and rendering, but not Harlequin Parallel Pages™. The bottom trace shows that the Harlequin Host Renderer using Harlequin Parallel Pages is 1.3 times faster.