What would cause some drives to certify much faster than others?

@astrostl 

This is a part of the hardware design apparently. I tested this here and see what you are referencing, so it is not a problem with your blades.

It appears the first two drives (PCI wise) get priority for read requests. (similar to my earlier description, but I am sure I did not describe it accurately from an engineering perspective)

There seems to be no difference on write requests when I tested.

This is a complex design to get 8 blades into one enclosure. You are likely measuring an artifact of that. It has no impact in the "real world", so lets ignore this.

I would technically surmise that it still has at least some real-world impact, as data that happens to land on the early blades will seemingly perform better in ops under duress as evidenced by certification and verification. I don't think it has one that matters to me, though, and my main concern was whether or not I had a weird/glitchy device. I'm satisfied that I don't, so thanks a bunch for co-testing it!

@astrostl 

Thanks, it would have minimal impact. Don't forget that a sustained read across all drives (the scenario) will flood the Thunderbolt bus, so this scenario of 2 drives going faster would have no real world impact, as with sequential reads on a volume, all drives would be requested to deliver data in order. which it will do at 2800MB/s.

At least we know this is a hardware design, not a faulty enclosure or drives.