User Tools

Site Tools


projects:yieldidiom

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
projects:yieldidiom [2009/06/11 13:21]
cormac
projects:yieldidiom [2009/10/06 13:56] (current)
jaeheon
Line 1: Line 1:
 ====Stuff to do==== ====Stuff to do====
  
-== Type System ​== +== Memory Models and Scheduling (Oct 6, 2009) == 
-  * type system for Java or for IMP +  * orthogonal issues, but closely related 
-  * similar ​to Types for Atomicity paper (PLDI03) or TLDI03 +  * Need to tie together 10tldi and 09concurrency 
-  * deadlocking issue+  * Hypothesis: if ->pr then ->cr, where ->pr is preemptive/​relaxed_mm execution and ->cr is cooperative/​relaxed_mm execution 
-  * soundness proof+  * Ideally, if ->pr then ->c : a well-formed program, even w/ races, still executes according to s.c. and cooperative annotations.  
-  * implementation(5KLOC)+  * Some races are impossible due to scheduling;​ 
 +        {{projects:​impossible-race.png?100}} 
 +  * Some races are related to scheduling. 
 +        {{projects:​scheduling-race.png?200}} 
 +  * Relationship between various semantics  
 +        {{projects:​semantics-diamond.png?400}} 
 + 
 +-----
  
 == Yield checking in RR == == Yield checking in RR ==
Line 34: Line 41:
   * SC(P) ⊆ HB(P): trivial   * SC(P) ⊆ HB(P): trivial
   * HB(P) ⊆ SC(P): ?   * HB(P) ⊆ SC(P): ?
 +
 +== Type System ==
 +  * <​del>​type system for Java or for IMP</​del>​
 +  * <​del>​similar to Types for Atomicity paper (PLDI03) or TLDI03</​del>​
 +  * <​del>​deadlocking issue?</​del>​
 +  * <​del>​soundness proof?</​del>​
 +  * <​del>​implementation?​ (5KLOC)</​del>​
 +
 +
projects/yieldidiom.txt · Last modified: 2009/10/06 13:56 by jaeheon