stev wrote:@Codysluder
Thanks for the heads up on the 2 or 3 issues.
My issue is that the GPU client takes up so much processing power, that I need to pause the F@H. If I run only the CPU SMP, there is no lag. I would like to run both the SMP and GPU combined all the time without the GPU lagging the system. There is no GPU slider (yet?) for F@H. Is there a way to trim back the GPU F@H to allow the lag or 3-5 second freeze and unfreeze cycles with the computer to go away?
Development plans to reintroduce the slider though there is no estimate for when that might happen. Please don't expect miracles, however.
The FahCore divides up the processing into compute blocks (mini-jobs) for the GPU to process. The duration of those blocks depends on several factors including which GPU you happen to have. Unlike your Operating System, GPUs have no inherent method of interrupting a block nor a priority scheme which can designate work from one source to take priority over work from another source. Capabilities like that may someday be added by AMD/NV/Intel/etc. but they're not available today so FAH can't use them.
To be efficient, the FahCore must plan ahead by transferring data for one block while working on other blocks rather than waiting for the (SLOW) pci-e to transfer the data while the GPU is idle. In other words, an efficient design makes inserting work from another source rather difficult. Adding the GPU slider will help, of course, but you'll still have to wait for the appropriate opportunity to insert your processing into the First-Come-First-Served sequence.
There has been talk of reintroducing a screen-saver. This is probably the only real alternative because the inherent pauses won't happen when you're trying to do something else by automating the FAH to pause when you're watching a video (etc.). Nevertheless, I'd like to see the ability to automatically pause GPU folding while certain programs are running.