Input and Output Latency using 48kHz and 96kHz

[Chri]

Audio latency is an important issue for monitoring with active plug-ins, e.g. Guitar Rig.

One reason for the input and output latency is the hardware buffer, which can be adjusted in size. (By the way, does anyone know where to find a list of possible buffer sizes for the LE and MPowered interfaces ?).

Now the latency caused by the buffer is calculated by dividing the amount of samples by the sampling rate. Reducing buffersizes will reduce latency. Increasing sampling rate will also reduce latency.

As far as I know, the minumum buffer size you can use without getting clicky audio is limited by the speed of your machine (CPU, Ram, FSB, etc.). Lowering the buffer size will require more processing power. Now the question is: Does increasing the sampling rate (e.g. from 48kHz to 96kHz) increase the demands for processing power in the same way ? If not, one could use the higher sampling rate to reduce the latency, right ?

Is it possible to set the input and output sampling rate to different values ? If so, one could use 96kHz as output rate to reduce output latency, but still record at 48kHz.

Regards,

Chri