It is almost free on modern CPUs that have hardware acceleration, yea

Space-faring electronics aren't exactly cost-sensitive - the cost of a cluster of crypto-accelerated CPUs or rad-hardened FPGAs is peanuts compared to the human and launch costs that go into these satellites.

Issue is the satellite was launched 10 years ago with 20-year-old tech. So, calculations of today may not be applicable on them.

Wireguard uses ChaCha20, which to my knowledge neither has nor requires HW acceleration to be fast.

It's faster on CPUs without dedicated hardware than AES, but that doesn't mean that it's faster than fixed-function AES hardware.

> However, the software performance [of wireguard] is far below the speed of wire.

https://github.com/chili-chips-ba/wireguard-fpga

Here is their excuse:

> Panasonic told us that enabling encryption could incur a 20–30% capacity loss. In addition, when using IPsec, ESP and IP headers can introduce 20–30 bytes of overhead, which is nontrivial for small-packet applications like VoIP and video calls

> Panasonic told us that enabling encryption could incur a 20–30% capacity loss.

Wow, I guess they're still betting on customers sending tons of redundant data up/down that they can shave off via compression? That's such a 90s modem thing to do. ("Faster than 56 kbit/s!!")