hi everyone,
still trying to find out what is the exact way to get the best performance from a multichannel system, taking up the guidelines dictated by loockheed martin:
Synchronization
For best performance, the client machines should utilize genlock-synchronized graphics cards to ensure an exact frame-for-frame match between adjoining display devices. Failure to use graphics cards with synchronized frame generation locks may result in undesirable display characteristics, as one display may fall out-of-sync with another adjoining display. This is especially probable if the complexity of the visual scene varies across the complete field-of view. Without genlock synchronization, a client rendering a simple portion of the scene may finish and present its display earlier than another client rendering a more complex region of the same scene. The software synchronization ensures that all clients have submitted all GPU work before the host sends a message to present the frame. If the GPU is not done with all this work by the time the present message is received, the frame flip may be delayed. If hardware synchronization is not being used, this delay may lead to displays being noticeably out of sync. The best way to avoid this scenario is run with settings that keep all systems CPU bound, or to frame-rate-limit the Host to provide enough time for all clients to finish GPU work.
Note: While client displays are software synchronized, the host simulates and renders as fast as possible to provide the highest quality simulation. Rendering to displays from the host is allowed, but host displays are not synchronized with client displays (even using hardware synchronization).
from all the tests carried out by the various users it would seem that no one has ever found true perfection, or due to inadequate hardware or incorrect settings between vsync, hertz between the various PCs. In fact, nobody has ever clarified what hardware is needed to get a perfect setup. The lockheed martin guideline speaks explicitly about particular graphics cards but none so far I have seen uses something like this.
From my personal experience I tried some nvidia quadro cards but the result was very bad in terms of graphics and syncronization (they are not in fact suitable for gaming)
So I ask the lockheed martin team, to be able to give a more comprehensive guide on the type of tests performed and the appropriate hardware. among other things, having to pay $ 2300 for a software that does not have any kind of detailed guideline becomes difficult and you opt for that point for ridiculous cost systems...
thankyou
simone
