MathGroup Archive 2011

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: ProgressIndicator and ParallelTable problem :(

  • To: mathgroup at smc.vnet.net
  • Subject: [mg117987] Re: ProgressIndicator and ParallelTable problem :(
  • From: psycho_dad <s.nesseris at gmail.com>
  • Date: Thu, 7 Apr 2011 08:06:07 -0400 (EDT)
  • References: <inhasb$2ad$1@smc.vnet.net>

On Apr 6, 11:13 am, Eric Brown <eric.c.br... at mac.com> wrote:
> I used to be bothered by this, until I reasoned that it's not a well-defined task.
>
> Say there are ten tasks that need to be processed in a ParallelTable, and I have a quad-processor.  Also, consider that each task might take a different amount of time.
>
> So, things start off fine, Task 1 is on Processor 1, Task 2 is on Processor 2, etc.
>
> 1 -> 1
> 2 -> 2
> 3 -> 3
> 4 -> 4
>
> So what is the index for ProgressIndicator?  I would say 4, but 1 would be just as valid.
>
> Now let 2,3, and 4 finish sooner than 1.  Now,
>
> 1->1
> 5->2
> 6->3
> 7->4
>
> So what is the index for ProgressIndicator?  7 or 1?
>
> I totally see what you want to do.  I ended up cutting the problem into batches and monitoring progress on those instead.
>
> Eric

Hi Eric,
I see what you mean. However, I am not interested on the index of each
job, but I want the ProgressIndicator to update after every job is
finished. Using your notation:
Suppose you have 5 Tasks (A, B, C, D, E) that you want to run on a
quad-processor, ie 4 parallel kernels. Also, as in your example,
suppose that each task might take a different amount of time.

What I want to get is something like (assuming that B finished 1st,
then A, then D, then C and finally E):
Task      % on ProgressIndicator
B           20%
A           40%
D           60%
C           80%
E         100%

Furthermore, consider that you have 100 tasks that you want to run on
the 4 parallel kernels. So the question is, how can I force the
ProgressIndicator update every after completed task?

Cheers


  • Prev by Date: Re: Shadow problem of a variable defined in a function context inside different packages
  • Next by Date: Re: why extending numbers by zeros instead of dropping precision is a good idea
  • Previous by thread: Re: ProgressIndicator and ParallelTable problem :(
  • Next by thread: Re: ProgressIndicator and ParallelTable problem :(