ࡱ> 0/` ( / 0LDArial\ 0\0ذ| 0|@ .  @n?" dd@  @@`` p0 $    0AA@3ʚ;ʚ;g4?d?d| 04ppp@ <4ddddw 0 80___PPT10 U  Day3 NotesBatch Queue PredictionRefer to Dan N. slides There are two methods possible to get slots using the current foundation (inverse methods) There are interesting possibilities if we consider probabilities/confidence in a scheduled makespan and weights to see if deadline can be met Modeling makespan as Markov models Similar to checkpoint scheduling Scheduler is agnostic to the probability we set some arbitrary cut-off\#P!P)PP#!),4zScheduler discussions2Question of if scheduling is iterative How does the initial performance model answers get mapped to vgDL Scheduler steps Scheduler gets a conservative estimation for each task of workflow from performance model vgDL  No of resources, type, slot, structure Task to slot mapping *zZZz,dmA  8What is the optimal way to reuse slots How do we bootstrap the asking vgDL take average case of resources the ones we get maybe better or worse the slots will need to be moved vgES collective answer is found or not found we really need some feedback back from vgES The iterative runs of scheduler need to be close together and need to ask for all each time since things will keep changing Maybe strategy should be Ask coarse grained approximations Fine grained mapping and then ask We need to provide a simple algorithm for slot count construction. Walk through DAG KPePPTPPEPDPPKeTED  PFfP/p     0` 33` Sf3f` 33g` f` www3PP` ZXdbmo` \ғ3y`Ӣ` 3f3ff` 3f3FKf` hk]wwwfܹ` ff>>\`Y{ff` R>&- {p_/̴>?" dd@,|?" dd@   " @ ` n?" dd@   @@``PR    @ ` ` p>> f(    6,  `}  T Click to edit Master title style! !  0   `  RClick to edit Master text styles Second level Third level Fourth level Fifth level!     S  08 ^ `  >*  0, ^   @*  0 ^ `  @*H  0޽h ? 3380___PPT10.dpK* Default Design 0 zr@ (    0v P    P*    0z     R*  d  c $ ?    0@~  0  RClick to edit Master text styles Second level Third level Fourth level Fifth level!     S  6( _P   P*    6$ _   R*  H  0޽h ? 3380___PPT10.dn$ 0  $(  r  S @G>  r  S HI `    H  0޽h ? 3380___PPT10.d*$  0 0 $(   r  S ̬ `}   r  S  `  H  0޽h ? 3380___PPT10.dP~/$  0 p$(  r  S  `}   r  S  `  H  0޽h ? 3380___PPT10.dP/  0 $(  $r $ S P `  H $ 0޽h ? 3380___PPT10.dP^  0 P(  X  C      S T 0    H  0޽h ? 3380___PPT10.dn  0 `(  X  C      S @ 0    H  0޽h ? 3380___PPT10.dn  0  (   X  C      S  0    H  0޽h ? 3380___PPT10.d@2  0 ((  (X ( C     ( S ` 0    H ( 0޽h ? 3380___PPT10.d`r0h BL%`'n!t)#+- 1Oh+'0 `h  Day3AM NotesUNCUNC46Microsoft Office PowerPoint@P@7ed@ ,XdGg    y--$xx--'@Arial-. 2 67 Day3 Notes."System-՜.+,0    On-screen ShowUNC-G ArialDefault Design Day3 NotesBatch Queue PredictionScheduler discussionsSlide 4  Fonts UsedDesign Template Slide Titles_-0UNCUNC  !"#$%&()*+,-.1Root EntrydO)Current User'SummaryInformation(PowerPoint Document(-DocumentSummaryInformation8