For several releases we have provided the option to use GPU technology from compatible GPU cards to enhance the solver speed. There were several restrictions to this option:
- the use of multiple GPUs was not supported
- the available memory on the GPU restricted the maximum models size
- the computer might have stopped responding as the card diverted memory from other tasks, while running computation-intensive tasks
- Windows Remote Desktop did not support direct-access to GPU resources, so you could not utilize the GPU card on the remote machine.
- On Windows 7 or later, a job launched locally would fail if you logged off the server.
In this release, we have removed this option for all solvers. If you used this option in the past, you may see a slight increase in your analysis time; in some instances, you may see a slight decrease. In general, the impact on your analysis time should be negligible.