Re: Parallel Construction of NavMeshes
Vincent Knecht <[email protected]>
| Newsgroups | gmane.comp.graphics.crystalspace.devel |
|---|---|
| Message-ID | <CAGsPR3hzO3KCrwonQrh22=V30oy9QmX3Y9BxSoqzBEXpEPFCpg@mail.gmail.com> |
2012/7/30 Sam Devlin <[email protected]> > Hello All, > I believe I have parallel navmesh constuction working as of revision 4936 > of my GSoC branch at > https://cel.svn.sourceforge.net/svnroot/cel/cel/branches/soc2012/ai/ > > Unfortunately, I only have a single core machine. On which this code is > functional but slower than the previous method. > > If anyone with a multi-core machine could check out revisions 4936 (with > parallel construction) and 4933 (without) and compare apppathfindingtest > both for functionality and speed, it would be greatly appreciated. > > To build a navmesh and so test the new method, the user must press 'b'. > After a brief pause the ground of the map should be covered with a yellow > mesh divided into tiles which is then used by R&D to find a path for Cally > to any location the user left clicks on with the mouse. She will then > follow the path. > > Please let me know if anyone gets the time to try this out. > Hello Sam, tried with CS trunk r38578 built in debug mode on a Core2-Duo T6400 2GHz 64bits Linux. With CEL r4933 : construction takes ~4s With CEL r4936 : construction takes ~3s FPS were about the same in each case. I didn't notice notice functionality differences. I'll retry with optimized builds later... ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Crystal-develop mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/crystal-develop