Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!think.com!paperboy!hsdndev!husc6!genrad!stardent!larryg From: larryg@Stardent.COM (Larry Gelberg) Newsgroups: comp.graphics.visualization Subject: Re: Building AVS modules Message-ID: <1991May23.152627.2969@Stardent.COM> Date: 23 May 91 15:26:27 GMT References: Distribution: comp.graphics.visualization Organization: Stardent Computer, Concord MA Lines: 25 In-reply-to: somick@nmsu.edu's message of 22 May 91 23:37:53 GMT One succesful technique for dealing with time-based data within AVS is to build up a 4D field where the first three dimensions are space and the last dimension is time. I have a couple of modules which then extract specific time slices (a 4D to 3D 'orthoslicer') or linearly interpolate between adjacent time steps. The biggest disadvantage to this approach is that at least one module needs to store all the time steps in memory at once. This is especially cumbersome in AVS2 which does not use shared memory to elminate data redundancy. AVS3 is available to Stardent customers now - contact your sales-unit. A more elegant approach to dealing with large amounts of time-based data is to have a module (co-routine) either generate the data on the fly (import your simulation) or only read those time slices off disk which you are currently using. I hope this helps! -- _______________________________________________________________ /\ Larry Gelberg larryg@stardent.com \____ /\\\ Visualization Software Group \ \ \\\\\ Stardent Computer Inc. _____________ \ \