phc-discussions - Re: [PHC] Another PHC candidates "mechanical" tests

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: <54610738.4000100@gmail.com>
Date: Mon, 10 Nov 2014 19:43:04 +0100
From: Milan Broz <gmazyland@...il.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] Another PHC candidates "mechanical" tests

Hi,

On 11/10/2014 04:13 PM, Sascha Schmidt wrote:
>> - I tried to visualize real used memory (measured by rusage()) and
>> run time with variable mcost/tcost (and generate some fancy graphs:-)
> 
> Judging from our graphs you might have confused the instances of
> Catena. Catena-DBG should require about 1.5 times the size of
> Catena-BRG and it does so in our tests. Also Catena and Catena-BRG
> should be absolutely equal in all categories.

I do not think so, you can check it in hash_libs dir. The makefile compiles only
required sources - for dbg catena-DBG.o for brg catena-BRG.o.

But I think the problem is that for submitted Catena I used minimal mcost value 18.
(Lower values without patch fails.)
(IOW direct comparison of Catena and Catena-BRG on graph is not possible now,
there are different baseline parameters.)

And there is logaritmic scale (so better to check raw output), but I really do not
see 1.5 times the size between Catena-DBG and Catena-BRG. How did you measure memory use?
I use getrusage(RUSAGE_SELF) on forked process.

But it is possible I did something wrong of course...

Thanks,
Milan