If you want real world performance test it head to head on the same system.
Just make sure the entire disk has finished being encrypted before testing, as it is not fair to test while FileVault is still encrypting the disk.
i5 and i7 intel CPUs include dedicated encryption/decryption logic that makes the FileVault encryption less of a load on the system. Older chips do not, so that would be done in software.
Macs with rotational disks may not notice, as the disk transfer times may totally overshadow the encryption/decryption times.
Macs with SSDs may not notice if they just came from a rotational drive, as everything is so much faster.
Macs with SSDs where the user has had the system for awhile, may notice a tiny change, especially if they are measuring before and after due to the fast SSD I/O such that encryption/decryption may be a slightly larger portion of the total I/O time.
Macs with SSDs should really be using FileVault, as SSD are very difficult to securely erase due to the nature of the SSD internal operations. With FileVault, once you reformat the drive, the decryption keys are destroyed and from that point forward the rest of the data is just a bunch of random bits. To this end, Apple has removed the Secure erase option for SSD storage (secure erase would also unnecessarily shorten the life of an SSD, as SSDs have a limited number of write cycles).
Personally, I do not really notice it, but I have an SSD and an i7 CPU.
