Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20131203223928.GB24951@openwall.com>
Date: Wed, 4 Dec 2013 02:39:28 +0400
From: Solar Designer <solar@...nwall.com>
To: john-users@...ts.openwall.com
Subject: Re: Energy-efficient bcrypt cracking (Passwords^13 slides)

On Tue, Dec 03, 2013 at 10:30:13PM +0100, Jeroen wrote:
> It might be worth considering testing the Intel Xeon E3-1265L v2/3 (4c/8t).
> About -10% CPU power compared to a Intel Core i7-4770K (according to
> 'normal' benchmarks) but using only 45 Watt TDP(!). Might be a winner in the
> x86 arena.

We couldn't actually measure power consumed by the system with i7-4770K
for this presentation, so we had to rely on 84 W TDP for our estimates.
That system was at a remote location and didn't contain builtin power
measurement.  Additionally, estimating CPU power consumption is tricky
even with physical access.  I suspect that this CPU was actually drawing
somewhat less than 84 W during our bcrypt benchmark, despite of it
running at 3.7 GHz (due to turbo), which we did verify.

Yes, it'd be curious to test both E3-1265L v3 and i7-4770K, and measure
their actual power usage.  Also, with low-TDP x86 chips longer test runs
are needed for throttling to possibly kick in as the CPU heats up.  45 W
TDP may mean that this will become the power limit when the CPU reaches
a certain temperature, but may be exceeded until then.  I am unsure if
those "normal" benchmarks you're referring to were run for long enough
periods of time.

Alexander

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.