« first day (3362 days earlier)      last day (1610 days later) » 

Trey
Trey
07:47
I'm a little confused about TNRGs and CSPRNG, regarding this question crypto.stackexchange.com/questions/38053/… it says that TNRGs are rarely uniform, so why not use them as a CSPRNG rather than a seed for a CSPRNG?
 
1 hour later…
SEJPM
SEJPM
08:47
@Trey they should be uniform so they would make for bad CSPRNGs
also they're usually slow
(e.g. RDSEED is like 10-100x slower than RDRAND on Intel platforms)
Woodstock
Woodstock
hello, in most modern stream ciphers, is it fair to say that the key stream is generated by running the key and nonce through a symmetric algo?
@SEJPM that's fair sorry :)
SEJPM
SEJPM
@Woodstock "symmetric algo" has no precise definition
but yes, you'd find similar kinds of operations in modern stream and block ciphers and hash functions
Woodstock
Woodstock
@SEJPM, could you give me a one-liner on how key stream is made?
ah, so it is a block cipher (type construction or equiv), that uses the key and nonce
to make the keystream? (high-level here)
SEJPM
SEJPM
technically the underlying constructions aren't usually block ciphers because you can't invert them given the key
Woodstock
Woodstock
You're nearly encrypting "nothing" with the key and nonce, to create a stream of waffle, that will be arithmetically combined with the plaintext to for cipher text
what types of constructions are used? compression functions maybe?
SEJPM
SEJPM
08:53
Usually (e.g. in ChaCha / Salsa type ciphers) keystreams are generated by combining the key with a nonce and a stream-counter (for multiple fixed width "blocks") in such a way that not knowing the key the output looks random even when the nonce and counter are known.
technically PRFs in counter-mode is what is being done
(at least for chacha / salsa I don't know about others)
(and you can use block ciphers as PRFs for up to 2^(n/2) blocks for n-bit block size)
 
14 hours later…
Trey
Trey
22:35
@SEJPM so they are bad because they are slow basically? I thought they being uniform was good since that would me randomness

« first day (3362 days earlier)      last day (1610 days later) »