john-users - Re: Is JTR MPIrun can be optimized for more cores ?

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 10 Mar 2010 09:26:40 +0100
From: "websiteaccess@...il.com" <websiteaccess@...il.com>
To: john-users@...ts.openwall.com
Subject: Re: Is JTR MPIrun can be optimized for more cores ?

On Wed, 10 Mar 2010 03:05:07 +0100, Magnum, P.I. wrote:
>> I tested JTR MPI 10 with -incremental mode (compiled SSE2).
>> 
>>  I ran JTR with a 2 millions hashes list, with my own .chr (can 
>> crack up to 10 letters/digits/some symbols)
>> 
>> Here it is 3 tests (same hashes list, all identical except cores 
>> numbers, duration for each test 5 minutes): 
> 
> This would be a more relevant test:
> 
> 1. Run your first test (2 cores) for 20 minutes.
> 2. Run your second test (4 cores) for 10 minutes.
> 3. Run your third test (8 cores) for 5 minutes.
> 
> and you could also throw in:
> 
> 4. Run a non-mpi john for 40 minutes.
> 
> Now compare the results (as in number of passwords cracked). This 
> will be a better indicator of how john-mpi scales.

 Here it is a new test with Magnum's advice :

------------------------------------------------------
1  core,  duration 40 mn : found 8676
------------------------------------------------------
2 cores, duration  20 mn : found 8207
------------------------------------------------------
4 cores, duration  10 mn : found 7886
------------------------------------------------------
8 cores, duration   5  mn : found 7189
------------------------------------------------------