Factorization of RSA-250 (opens in new tab)

(lists.gforge.inria.fr)

30 pointskissickas6y ago5 comments

5 comments

4 comments · 1 top-level

zakk6y ago· 3 in thread

Estimating a cost of 1 cent per core-hour, this cost more than 200k€. Since INRIA is a national research institute, I am guessing at least part of this has been paid with taxpayer money, either directly from the Institute or from grants.

I am wondering how such an expense can be justified.

It's an honest question: I find all of this very exciting and brilliant, however having written grant proposals myself, I am wondering how could they 'sell' factoring larger and larger numbers as worth funding.

KCUOJJQJ6y ago

Noob here. Could this work as a benchmark and tell us how safe RSA-1024, RSA-2048, RSA-16384 etc. are?

BTW, why shouldn't everyone switch to RSA-16384? According to https://wiki.gnupg.org/LargeKeys "elder versions supported creating of keys up to 16 KiB." When it was possible to create 16384 long keys, it must still be possible to use them with new versions of GnuPG, right? And https://www.keylength.com/en/compare/ tells me that 16384 bits are way better than 4096 bits. According to "Lenstra Updated" your data is protected until the year 2153 with 16384 bit keys and until the year 2060 with 4096 bit keys.

dippersauce6y ago

I would think a primary reason for that is performance. A key that large would require a lot of entropy for initial generation, a large(r) amount of memory, as well as making encryption much more computationally expensive. I'd also be worried that with such large keys there might be greater potential for side-channel attacks.

1 more reply

elliottback6y ago

From the description it doesn't sound like there is any new technique being used, just an application of existing software to a new large prime.

There are a very few justifications:

* They get new bounds on factorization performance, which might be important to keep tabs on over time, in case no one else does it, and suddenly we find all our primes factored

* For some reason, INRIA really really hates unfactorized primes

I agree with you that, unless their upcoming research paper introduces new information/experimental procedure, it was a waste of money.

j / k navigate · click thread line to collapse