We need for test purposes a Super Speed Device Loopback.
The need of additianal components should be as low as possible.
Which Device should we select, FX3, CX3?
Or is there an other Super Speed Device we can use for our purposes?
FX3 meets your requirement.
You can refer to FX3 Application Note AN65974 ( Design Example 1 Section 10) . Download link is given below. In this example code FX3 is connected to Xilinx via FMC connector board. Fx3 receives data via one EP send it to FPGA via over GPIF interface. FPGA then loopback data to host via another EP of FX3.