It's a little complicated - and it depends on whether GPU-accelerated video decoding is being used or not. If it's not being used, the cache size corresponds to the regular RAM that is used. If GPU-accalerated video decoding is used, it can use a lot more of the GPU's RAM to buffer video frames. In order to optimise things in this case, MixEmergency keeps as much as it can on the RAM of the GPU. Most computers don't have a huge amount of RAM available to the GPU - so if MixEmergency tries to allocate 1 GB of GPU RAM to buffering, and the GPU only has 1 GB available, then it is going to have to shift a lot of that back to the main RAM in order to free up the GPU's RAM for other things (e.g. other Mac OS X applications graphics, etc). Shifting the RAM back and forth between the GPU and the CPU is an expensive operation - so we try to avoid it. So, it's better to be conservative with the level of RAM allocated to the cache, to try to avoid a situation where MixEmergency starts to use 'too much'.
