thanks for your quick response.
so, the main reason is the amount of memory, right?
the process image is consist of heap, stack, envirnment and so on, it will not be very large.
Josh Hursey <jjhursey_at_[hidden]> wrote:
This is to be expected. BLCR saves the entire process [individual
rank] image to disk, so if your process is large, and allocates a
significant amount of memory then the context files will be large.
Unfortunately there is not way to make the BLCR context files smaller
at the moment.
-- Josh
On Apr 20, 2006, at 1:53 AM, Mars Lenjoy wrote:
> i'm using lamcheckpoint.
> for the parallel mesh program, checkpoint context file is really
> big, terrible, displayed followed, and take it so long time. does
> it save all the date in the memory?
>
> -r-------- 419411063 Apr 19 23:08 context.21494-n0-21727
> -r-------- 465662631 Apr 19 23:08 context.21494-n0-21728
> -r-------- 456671359 Apr 19 23:06 context.21494-n1-21921
> -r-------- 480885959 Apr 19 23:06 context.21494-n2-14486
> -r-------- 131587 Apr 19 23:08 context.mpirun.21494
>
>
>
> Love cheap thrills? Enjoy PC-to-Phone calls to 30+ countries for
> just 2?min with Yahoo! Messenger with Voice.
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
----
Josh Hursey
jjhursey_at_[hidden]
http://www.lam-mpi.org/
_______________________________________________
This list is archived at http://www.lam-mpi.org/MailArchives/lam/
---------------------------------
Yahoo! Messenger with Voice. PC-to-Phone calls for ridiculously low rates.
|
