![]() |
|
Date: Tue, 9 Sep 2014 08:13:42 -0500 From: Richard Miles <richard.k.miles@...glemail.com> To: john-users@...ts.openwall.com Subject: Re: Questions about statistics output Hi all, I understand the issues pointed by magnum, I just want to say that I would like to see some feature as the one requested by Len Rose, a total SUM of all processors would be very helpful. Maybe not for guys that are really hardcore such as magnum and solar d., but for sure it will be very useful for users and novices as I'm. :) Thanks. On Tue, Sep 9, 2014 at 1:10 AM, magnum <john.magnum@...hmail.com> wrote: > On 2014-09-09 03:13, Len Rose wrote: > >> I am using version 1.8.0.2-bleeding-jumbo_mpi on a small 32-node cluster. >> >> I have been using a much older version for the last few years >> (john-1.7.9-jumbo-7) and decided to try the latest code. >> >> I noticed we no longer see the SUM and averages on the status output >> so was wondering if there is some new way to get that without writing >> some external script? >> >> i.e. " SUM: guesses: 0 time: 1:10:03:10 0.00% (3) c/s: 388 avg12.14 " >> >> Now it's simply: >> >> 1 2g 0:21:07:53 3/3 0.000026g/s 18.93p/s 50.50c/s 50.50C/s >> 2 0g 0:21:07:21 3/3 0g/s 15.99p/s 50.36c/s 50.36C/s >> 3 0g 0:21:07:37 4.00% 2/3 (ETA: 2014-09-29 22:10) 0g/s 3.217p/s 10.13c/s >> 10.13C/s >> (...) >> > > If I recall correctly I dropped the MPI summary line when merging 1.8 > core, as I made the MPI code as similar to --fork code path as at all > possible. I might look into re-implementing that summary if you really need > it. > > That's fine but the "ETA" field keeps disappearing and I was >> wondering if we could still have the code print a similarly formatted >> date for every node so that I can produce a script that can handle >> summing and averaging on the above output? >> > > In your example you are running batch mode. Node 3 and others are still in > stage 2/3 and their ETA show when /that stage/ will finish. Nodes 1, 2 and > 8 has already entered stage 3 and show no ETA because it's beyond year 2029 > or something like that. There wouldn't be much point in showing an ETA for > them. > > A summary line would not be able show a reasonable ETA at all in a similar > case. I can't remember if the older versions tried to or not but as you can > imagine, trying to average "in two weeks" and "in 15+ years" will not > produce anything usable. > > You should be able to handle the difference with a trivial regexp though. > > magnum > >
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.