We can use simultaneous randomized benchmarking (SRB) to quantify the crosstalk impact of simultaneous CNOTs. The crosstalk error is caused by many reasons, for example, the always-on ZZ interaction, the crosstalk between microwave lines.
My question is, is the number given by SRB experiment an average of all these reasons? Which one is the most important reason to cause crosstalk error between simultaneous cnots? Why is the error rate of cnot increased a lot during simultaneous executions?