Peter Pajchel

I'm currently working on a Mac Pro testing out how VBLsync works with multiple jit.windows on multiple GPUs

Current test setup is ( 2009 model 4 x Nvidia GT120 / OS X 10.6.2 / max 5.1.1 Beta)

GPU2 = two monitors / running one jit.window on each monitor output

GPU3 = two monitors / running one jit.window on each monitor output

jit.windows Attributes are = ( name @sync 0 @border 0 @depthbuffer 1 )

When I enable VBL-sync on two jit.windows running at the same GPU

the fps in jitter fluctuates between 60-30 fps.

However, with just one VBL-sync enabled jit.window per GPU I get a nice "steady" 60 fps.

Running two or three jit.windows on the same monitor output with VBLsync enabled also works.

Is it possible to run two VBL-sync enabled jit.windows on one GPU using both monitor outputs at 60 fps?

GPU2 =  two monitors / running one jit.window on each monitor output

GPU3 =  two monitors / running one jit.window on each monitor output

When I enable VBL-sync on two jit.windows running at the same GPU
the fps in jitter fluctuates between 60-30 fps.

However, with just one VBL-sync enabled jit.window per GPU I get a nice "steady" 60 fps.
Running two or three jit.windows on the same monitor output with VBLsync enabled also works.

jit-window-vbl-sync-on-multiple-gpus

> Is it possible to run two VBL-sync enabled jit.windows on one GPU using both monitor outputs at 60 fps?

The best way to get multiple outputs is to use a single large window (put two videoplanes next to each other in the same window) and a matrix triplehead2go to split the window.

The best way to get multiple outputs is to use a single large window (put two videoplanes next to each other in the same window) and a matrix  triplehead2go to split the window.

i believe joshua's recommendation is specifically for output to multiple windows from a single card (or a single card with multiple ports, such as your gt120, or a matrox).

if you want to make use of multiple cards, you must create a separate rendering context (jit.(p)window/jit.gl.render combination), and make sure that your jit.window @rect attribute places the window in the proper display coordinates for that card's display.

sharing one gl context across multiple graphics cards is a recipe for disaster (or at least poor performance).

you might find it necessary to re-instantiate your render context after the patch loads. this can be done easily by toggling the @floating attribute of jit.window.

jit.window VBL sync on multiple GPUs