Hi Jonathon I remember when I tried to make the "phrase table loading" messages fit better to what was happening in the decoder, some of the regression tests broke. I'm not sure if they rely on the timing messages being in a particular format. But yes, it would be great if the timing messages reflected reality better,
cheers - Barry On 10/08/12 15:47, Jonathan Clark wrote: > Also, there is currently no way of relating these stats back to the actual > sentence they came from when using multiple threads as far as I can tell. > Shall I also prefix each of these stats with the line number that the > source sentence came from? This should be useful for timing analysis such > as Lane's sentence splitting. > > Will modifying the format of these output lines with a prefix cause trouble > for anyone? > > > > On Fri, Aug 10, 2012 at 10:41 AM, Marcin Junczys-Dowmunt<[email protected] >> wrote: >> Reminds me of http://xkcd.com/552/ >> >> W dniu 10.08.2012 16:37, Lane Schwartz pisze: >>> Well, it may be mostly bogus, but it's not *totally* bogus. :) >>> >>> I use this number when I perform more advanced corpus splitting (see >>> my upcoming MT Marathon paper!), and while I can't claim to know that >>> it's accurate, it does at least seem to be well-proportioned. That is, >>> very short sentences that are processed very quickly report a small >>> number here, and very long sentences that are processed very slowly >>> report a large number here. >>> >>> On Fri, Aug 10, 2012 at 10:29 AM, Marcin Junczys-Dowmunt >>> <[email protected]> wrote: >>>> I on the other hand always had the impression that the time reported >>>> there is a total bogus, especially for multi-threaded decoding. >>>> >>>> W dniu 10.08.2012 16:26, Lane Schwartz pisze: >>>>> Not sure what it's supposed to be, but I like having some result that >>>>> reports the total per-sentence processing time, including both >>>>> collecting options and search. >>>>> >>>>> I'd just always assumed that the search time reported was that number, >>>>> I figured that to get just the search time you could subtract the >>>>> Collecting Options time. >>>>> >>>>> On Fri, Aug 10, 2012 at 10:13 AM, Jonathan Clark<[email protected]> >> wrote: >>>>>> Hi all, >>>>>> >>>>>> I just noticed that the moses time reporting is rather misleading. >>>>>> >>>>>> We see lines: >>>>>> Collecting options took 13.390 seconds >>>>>> Search took 13.390 seconds >>>>>> Translation took 13.390 seconds >>>>>> >>>>>> >>>>>> However, the "Search took X seconds" count also includes collecting >> options, >>>>>> which seems wrong. I have a patch for this I can push, but I just >> want to >>>>>> make sure I'm not missing something. This is broken, right? >>>>>> >>>>>> Jon >> _______________________________________________ >> Moses-support mailing list >> [email protected] >> http://mailman.mit.edu/mailman/listinfo/moses-support >> > > > _______________________________________________ > Moses-support mailing list > [email protected] > http://mailman.mit.edu/mailman/listinfo/moses-support -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
