Great, works perfect! I thought there is only a Linux SMP version. But since you mention you run SGI SMP I assume it is possible to compile a SUN SMP version, too. Since OpenDX seems to do parallel mode using Posix threads it should run on any plattform that has a pthread library installed. I will give it a try! Would just be nice of others using an Sun SMP version of OpenDX.
What parts are really parallelized in dxexec - I know I can just read the source, but it would be nice if there are any infos on that, or if somebody could roughly describe it. How is the scaling on certain functions? Is there any accessible information on the scaling of the transformation tools, eg. I suspect FFT to scale lineary but others being worse... Raphael > I don't know about the Linux SMP version. But if it's like the SGI, the > parallelism is at the process level, where you have multiple dxexecs > running: one master, and one per processor as slaves, communicating through > > IPC. Check your process list: > | \-+- 1163153 dxui ... > | \-+- 1155549 dxworker ... > | \-+- 1163133 dxexec ... > | > | |--- 1163161 dxexec ... > | > | \--- 1163211 dxexec ... > > Try dx -processors 2 and run some of the samples. If you see two dxexecs > cranking up there at near 100% sometimes, your SMP is probably working. > VolumeRenderingSimple.net is one where you can see this easily -- go into > rotate mode (Ctrl-R), and start dragging the mouse around like mad. You > should peg both your CPUs while its rendering. > > Randy
