[Cryptech Tech] Tor consensus use case

[Linus Nordberg]

Joachim Strömbergson <joachim at secworks.se> wrote
Thu, 29 Jan 2015 14:58:14 +0100:

| Aloha!
| 
| Linus Nordberg wrote:
| > Depending on the interface to this thing, the requirements might be
| > a bit tighter than that. Just to be overly clear, blocking for "a
| > long time" on sign and verify operations will induce sadness.
| 
| So what would be happy response times?

I would have to experiment a bit in order to answer that.


| Based on what I can see from core vendors for FPGAs, I think that we
| should be able to do a RSA-2048 operation in something like 300 ms, or
| even faster than that. A guesstimate for RSA-4096 is around a second.
| There seems to be few cores that supports keys beyond 2048 bits.

I think 2048 is fine. Those who worry about that could use keys with a
shorter life time.