My Pc has got a ATI Radeon HD-3450. I wonder if its possible to just put another graphic card to it and one could use it as a second render station for i.e. inbetween renderings and use the primary for the final rendering.
I’ve read about the Crossfire-Technology of ATI, which actually is doing this. But unfortunatley, I have the wrong motherboard.
Would it be possible to use a second graphic card for renderings (not the final) by placing the render window onto a second screen (run by second card). Or isn’t it so easy.
no it is not so easy.
vvvv renders on each graphics card all what is necessary to be able to show the content of the renderers associated with their monitors.
so all inbetween renderings will still be made by your primary card.
the result of putting an inbetween renderer onto the second monitor will just be that the second card will also render everything what is necessary to be able to show that inbetween rendering.
maybe a resource copy node would be nice to be able to explicitly copy the content changes of meshes or textures computed on one card to another. this is not done by default because it is probably a very slow thing to do.
That sounds still interesting for me.
-Am I right assuming, I could channel a “closed loop” like a collision shader(GPU)->pipet(CPU) and a UDP grey value(CPU)->dynamTexture(GPU)->contour freeframe(CPU) could be done upon a second graphic card?
And if so, how can I set destination engine. Is it by device node? I tried some settings, but couldnt see a difference, with my single card only, though.
As far as I understand device node, I can set a graphic card (?!?!) to my desired device number (i.e. 0 and 1) and then set every output renderer to i.e. 0 and my “closed loop” renderings to 1.
-Is this correct?
yes this should be correct…
we’ve hidden device pins in all “render sink” nodes; think of all the nodes which are at the bottom of ex9-subgraphs:
several writer nodes, the renderer, the pipet.
there you can choose a manual device with a positive number from 0 on, or -1 for automatic device selection. this automatic device selection is done
- depending on which monitor a renderer is, or
- in case of a pipet or writer, what device the other renderers already use (in the hope that some ex9 resources on the device can be reused)
note that sometimes more than one device needs to created on a single graphics card (dx slang = adapter). if monitors in different resolutions are attached and if you are not in spanmode than devices are created for each monitor. resources between them are not shared, even when they are controlled by the same card. vvvv only creates devices automatically if needed - e.g. if a renderer is visible on the second monitor.
with device (auto) you can inspect the automatically created devices.
when you assign devices manually, then you need to create devices first: with the device (manual) node. here you have the possibility to choose the physical adapter (=graphcs card).
Thanks a lot gregsn, I’m going to try it.