Browse Prior Art Database

A System for Performance Benchmarking of Complex Print Jobs

IP.com Disclosure Number: IPCOM000018737D
Original Publication Date: 2003-Aug-04
Included in the Prior Art Database: 2003-Aug-04
Document File: 2 page(s) / 47K

Publishing Venue

IBM

Abstract

This article describes how a virtual printer system, based on an actual print controller, can be used to estimate performance of complex print jobs. This system offers a way to determine if the job will likely print at rated speed on the target printer without having to run it on an actual printer.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 46% of the total text.

Page 1 of 2

A System for Performance Benchmarking of Complex Print Jobs

One of the main problems in high speed digital printing is maintaining the printer dataflow at rated speed. High capacity printers are almost exclusively based on the continuous forms technology, so data underruns by the print controller require either the print engine to stop abruptly, or that it emit some blank pages, or some combination of both.

These problems exist for all continuous forms printers, but are especially acute for color printers. Color printers commonly require much more data to be processed per page than black and white printers. In addition, the multiple head technology required to print color makes these printers even harder to stop.

No high speed printer can guarantee to print an arbitrary job at rated speed. All printer manufacturers impose restrictions on the print data to ensure that the job is printed without underruns. The restrictions are generally imposed on both the data complexity (e.g., the number of vector elements on a page) and quantity of data (e.g., image coverage on each page). For some of the color printers that rely on special-purpose hardware and carefully built jobs to maintain rated speed, such as IBM Infoprint Color 130 Plus, these restrictions can be exceedingly complex.

Violating the job restrictions even slightly may make the job underperform badly. As an extreme case, shifting an object on the page by a single pel might cause the job that was printing at rated speed print at a fraction of rated speed. The result of this is that it is exceedingly hard for the users and job and application generators to have any confidence that their jobs will print at rated speed.

This invention proposes a system that will allow for benchmarking jobs using a virtual printer and give a good indication to the user if the job will print at rated speed. The system may be either purpose built just for performance benchmarking or, more likely, built as an extension of a system for previewing jobs.

This description will, for clarity, assume that the benchmarking capability is being used in the context of a general preview system.

In the preview system, the job is rasterized using an actual printer rasterizer. Instead of the data being sent to printheads, a raster file output is produced. Benchmarking of the job involves two separate tasks - the rasterizer performance measurement and the bandwidth analysis.

To measure the rasterizer performance, the parts of the rasterizer that do not include the input I/O and the mechanism simulator (i.e., producing the output raster images) are timed. This allows for a time measurement for each job. Obtaining this measurement is in itself not very meaningful, since the printer controller will probably be running on a machine with a different model of processor, different number of processor for parallel page processing, different disk configuration, different memory configuration, and so on.

The question next is ne...