So far, the sample block has been used to specify the output of the implemented algorithms. There is one more block, the perform block, that can be used around the sample block. The perform block does not operate at the sample level, but at the block level (determined by the vector size of your system). This allows for optimizations in variable declarations and input unpacking (given that, however, the in1, in2 etc... constructs, when used in perform, will not retrieve values at the sample level, but at the vector level). The out1, out2, etc.. constructs are not available in the perform block.
Here the in1 and in2 are unpacked in the sample block. This means that their values are calculated once per sample. in3, which is unpacked in the perform section, on the other hand, is calculated once per audio vector.
perform:
mix = in3
sample:
out1 = (in1 * (1.0 - mix)) + (in2 * mix)
One thing that you might have noticed from this example and previous ones is that there was no need of explicitly declaring the ins and outs block. In fact, if they are not declared, they will automatically match the highest in.. and out.. count of the perform / sample blocks. In this example, this would be ins 3 and outs 1.
NOTE:
The perform block is optional. sample can be defined on its own. However, if perform is defined, the sample block must be contained in it.