search

radical-collaboration / HT-BAC_SC17_paper

0 stars 0 forks source link

Clarify text #20

Open sjzasada opened 7 years ago

sjzasada commented 7 years ago

https://github.com/radical-collaboration/HT-BAC_SC17_paper/blob/a44aae9724a183d552dc7bba16e55535189f5330/rct.tex#L34-L37

The latter clause should be telling me why something is a problem, but I'm not sure it is. Please clarify.

mturilli commented 7 years ago

Any better?

The scalable execution of applications with large ensembles of tasks is
challenging. Traditionally, two methods are used to execute multiple tasks on
a resource: each task is scheduled as an individual job, or MPI capabilities
are used to execute multiple tasks as part of a single job. The former method
suffers from unpredictable queue time: each task independently awaits in the
resource's queue to be scheduled. The latter method relies on the fault
tolerance of MPI, and is suitable to execute tasks that are homogeneous and
have no interdependencies.
shantenujha commented 7 years ago

yes.

sjzasada commented 7 years ago

Looks good but change 'awaits' to 'waits'