Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!sun-barr!ames!attctc!vector!egsner!csccat!larry
From: larry@csccat.UUCP (Larry Spence)
Newsgroups: comp.lang.postscript
Subject: Re: PostScript Benchmarks
Keywords: benchmarks performance
Message-ID: <3330@csccat.UUCP>
Date: 10 Oct 89 19:14:56 GMT
References: <7757@cg-atla.UUCP>
Reply-To: larry@csccat.UUCP (Larry Spence)
Distribution: comp
Organization: Computer Support Corporation. Dallas,Texas
Lines: 24

In article <7757@cg-atla.UUCP> bradlee@cg-atla.UUCP (Rob Bradlee) writes:
>I would like to obtain or create a reasonable set of PostScript pages
>for benchmarking output performance.  It seems to me that the items on
>the page to vary would be number of curves, number of different fonts,
>number of resident vs. non-resident fonts, image size and format (binary
>vs. hex).  Any observations or suggestions you can offer on this subject
>would be most welcome.

A recent issue of the Seybold Report benchmarked a bunch of high-res image-
setters (Lino, etc.), using a set of test pages.  Some had mainly text at
different sizes, others had heavy graphics, etc.  I'll bet you could get
them to send you the PS files for your own tests.  This would have the 
added benefit of allowing you to compare your figures with other machines'.
I don't recall the exact issue;  it was within the last 2-3 months.  Let me
know if you can't find it.

A lot of interesting things turned up in Seybold's testing, like interpreter
memory management barfing on large grayscale images, etc.  The tests are
tortuous enough to serve as a (very informal) validation suite.

-- 
Larry Spence
larry@csccat
...{texbell,texsun,attctc}!csccat!larry