First of all would like to thank all the people for this forum. It is a really enjoyable place to learn jitter and a nice place to exchange ideas.
My question is related with the best method to render a mesh into two shared context in two computers. The specs of the computers are: MCP 2.53 GHz + 4 GB Ram and a PPC with 1.67 GHz + 2 GB. As you can see there is a big difference between both computers so my idea is to do most of the calculations on the MCP and use the PPC just to render the data, which would be sent using an ethernet cable with a jit.net.send object.
My source patch is taken from a thread where Kyred and Nesa worked out a method to render a mesh into two shared contexts. But my question is where in the ‘pipeline’ of model->mesh->render I need to send the data to the PPC computer. Even more, I am curious if the render contexts, even though they are in two different computers, still can be shared. My point is that I want to optimize as much as possible the work so the PPC can be used only to render the data that has been calculated on the MCP(I am planning to apply some op and expr to the model).
I am also wondering that, in the case i want to use a texture for the mesh, I would need to send another matrix with that information. I hope I am not so baroque in the way I explain the situation!
jitter does not support shared contexts across computers (or across video cards for that matter).
you want to have your rendering context on the computer with the fastest graphics card.
you can send matrices with jit.net.send/receive, so perhaps you can split up your processing before you reach the graphics card. processing matrices on one computer and rendering opengl on another might be an optimal setup. however it also might be more trouble than it’s worth. as always, test out before committing.