Legal notice   Contact   Internals   Search 
System
Status
News /
Events
Support /
Documentation
Accounts /
Projects
Organisation Public
Relations
 

Tachyon

a Parallel / Multiprocessor Ray Tracing System.

Bernd Kallies(1)
Revision History:
Revision 1.1, Published 2006/10/27 11:17:01 (UTC) by Bernd Kallies

Table of Contents

1. What is Tachyon?
2. Version
3. Usage
4. Online Documentation

List of Tables

1. Tachyon Binaries and Libraries

1. What is Tachyon?

Tachyon is a multiprocessor ray tracing system which can be used as a standalone program, or as a library within other programs. It supports almost any Unix-like system, Windows, and MacOS X, as well as several embedded systems and high end supercomputers.

Tachyon is written in C, uses POSIX threads and/or MPI for utilization of multiple processors, and can also run on uniprocessor machines. Tachyon is easy to port to new machines, and is fast, using ray tracing acceleration schemes to achieve good performance on very complex scenes. (It will handle scenes containing more than 1,000,000 objects on a fairly average computer)

We use Tachyon as default raytracer from within VMD.

2. Version

The current version installed at HLRN is Tachyon 0.97.

The installation path is /aws/viz/tachyon-0.97.

We provide threaded as well as mpi-parallel executables and libraries in 32bit and 64bit object mode, see Table 1.

Support for jpeg and png output was added.

Table 1. Tachyon Binaries and Libraries

Directory Content
bin/aix-thr32/ Tachyon executable, threaded, 32bit
bin/aix-thr64/ Tachyon executable, threaded, 64bit
bin/aix-mpi32/ Tachyon executable, MPI, 32bit
bin/aix-mpi64/ Tachyon executable, MPI, 64bit
bin/ symbolic links to files in bin/aix-thr64
lib/aix-thr/ Tachyon library, threaded, 32+64bit
lib/aix-mpi/ Tachyon library, MPI, 32+64bit
lib/ symbolic links to files in lib/aix-thr
include/ Tachyon include files
scenes/ Example scenes for testing

3. Usage

3.1. Quickstart

Invoke tachyon on the command line to see the options.

Use the scenes provided in the installation directory for playing

3.2. Multithreaded Version

The threaded versions use all processors on the current machine by default.

To specify the number of parallel threads, use the -numthreads option.

For running in batch, set up a serial job, and specify ConsumableCpus matching the number of threads specified by -numthreads. See Running a serial program in The HLRN Quickstart Guide for a general example.

3.3. MPI Version

See Running programs using poe in The HLRN Quickstart Guide for running MPI applications interactively.

See Running an MPI program in The HLRN Quickstart Guide for setting up batch jobs.

3.4. Linking with libtachyon

Select the appropriate library depending on your parallelization scheme, see Table 1.

The library libtachyon.a contains objects designed for static linking.

The library libtachyon.so contains shared objects designed for run-time linking. Specify the -brtl linker flag when you want to access shared libraries instead of static ones.

You may have to specify the location of libjpeg, libpng, and libz, too. Use the versions installed below /class/lib. See http://docserv.hlrn.de:8080/sw/www/offer/berni/Cppc.html for descriptions.

4. Online Documentation




2003-2008 © Norddeutscher Verbund für Hoch- und Höchstleistungsrechnen (HLRN)