phc-discussions - Re: [PHC] A review per day

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <540A28DF.1040704@ciphershed.org>
Date: Fri, 05 Sep 2014 17:19:27 -0400
From: Bill Cox <waywardgeek@...hershed.org>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] A review per day - Economics of Makwa

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I wanted to say a bit more about the economics of building a Makwa
password hashing box full of ASICs.

The ultra-high end ASIC built in 22nm Intel process using the craziest
fastest densest technology ever created is really only an option the
MiB can afford (and maybe a few ultra-high-volume ASICs).  You need
proof you are willing to write a $10M check just to get in the door!
They want to see a credible plan to buy a total of $100M worth of
chips from then over the lifetime of your ASIC.  BitCoin is headed in
this direction, but they aren't there yet!

In reality, your VC is going to offer you $1M to develop the ASIC, and
$1M to build the boxes and commercialize it (or 2X that... whatever).
 For $1M you can talk to about a dozen ASIC vendors who will want a
credible story that you will buy $10M worth of ASICs over this
design's lifetime.

I'm not sure what technology you can build masks for now days for
under $200.  Maybe 90nm or 65nm?  Something mature for sure.  You need
to budget that twice in case there are mistakes!  Designing the chip
is maybe a couple of guys for a year, so another $400.  Packaging,
G&A, etc... maybe you get silicon working on a real board for $1M.
Then you crank up production.

These boxes, per Makwa core, which your low-end ASIC will have one
each of, will run closer to 1GHz, not 3.4GHz.  It also generates a
*lot* more heat!  You'll likely need that $100 flip-chip packaging
just for your 1 core, and that ASIC might cost you $150 to $200 each.

It would hash about 1000 Makwa hashes in parallel, with up to a cost
of 1,000,000 squarings in 1 second, with 1,000 in parallel.

Let's assume you get paid $0.0001 per password hash.  Then your Makwa
box generates a cent every 10 seconds.  That's over $8,500/day!  For 1
ASIC!

On the other hand, if an attacker wants to buy Makwa time to crack a
password with 30 bits of entropy (by which I mean it will take us 2^30
guesses), then an attacker will have to pay $100,000 to crack your
password!

Fun, huh?

Bill
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=HVIw
-----END PGP SIGNATURE-----