[Cryptech Tech] HWRNG specs

[Fredrik Thulin]

Hi Markus

On Monday, December 08, 2014 01:12:11 PM Olli Ottela wrote:
...
> about how open source the hardware is,

It will be very, very open. We haven't really gotten to the point of 
publication yet though - it is work in progress.

> at what speed can entropy be sampled,

There is a difference between entropy and random data. I'm going to assume you 
are familiar with this difference and just wasn't that precise in asking your 
questions =).

The board you saw a picture of is one of the two entropy sources. We're still 
in the testing phase of the avalanche noise extraction, but it seems to be 
possible - although probably not advisable - to extract entropy of a sufficient 
level to actually pass as good random data at about 20-30 kB/s.

> how Ent and Dieharder have evaluated the output.

We are using the entropy to feed a proper CSPRNG, so I would not expect any 
issues with the output random data.

> The cost is something to consider aswell: TFC already requires users invest 
> in hardware.

Yeah, well - I wouldn't expect what we produce to be the cheapest HWRNG 
around.

Maybe the best (read: most elaborate) and maybe the fastest, but not likely 
anywhere near the price range of various OSHW random number generators you can 
find online today, such as https://www.tindie.com/products/WaywardGeek/infinite-noise/ for example.

/Fredrik