True. But it would still help meter updates/plug-in automation, etc. I usually think of 3 basic system resources that we try to maximize: CPU usage, PCI usage, and disk bandwidth. The number of RTAS plug-ins, along with the HTDM/DirectConnect host processing, and the general complexity of your session (automation, number of meter displays, amount of MIDI data) contributes to the CPU usage, and using higher buffer settings for RTAS or StreamManager can help if you are running into CPU limits. With PCI, the number of audio streams is what pushes the limits (to -6042 errors). Spreading the voices across more DSPs (using fewer voices per DSP) can help, and spreading some tracks on IDE or FireWire can help if the SCSI card is using up PCI bandwidth in a "bad" way. With disk bandwidth, it isn't as much of a problem as it used to be (-9073 errors), but spreading voices across more disks can help, as well as increasing the SCSI card bandwidth (this is where you have a tradeoff between disk performance and PCI usage - -9073 errors vs. -6042 errors). And don't forget the DAE disk buffer setting - higher settings decrease the disk performance needed, with the cost of slower startup time since more data must be pre-buffered. As far as RAM, you're right, adding RAM probably won't help these kinds of problems. But it does allow you to use bigger DAE buffers with more tracks if you are having -9073 disk too slow problems.