 Marc Sergent
|
157dd97956
src/sched_policies/component_worker.c : Fix a false-positioned assert
|
12 éve |
|
Marc Sergent
|
913b6e242b
Several renamings for Hierarchical Schedulers to adapt the naming
|
12 éve |
|
Marc Sergent
|
638f30cba2
Merge from trunk @11789:11910
|
12 éve |
 Andra Hugo
|
1622267740
remove forgot printf in prev commit
|
12 éve |
|
Andra Hugo
|
4bd482e7b9
move hack out of the include files
|
12 éve |
|
Andra Hugo
|
e28f7852a3
hypervisor: steal work from other ctxs when the worker is idle
|
12 éve |
 Samuel Thibault
|
2fc5035219
Submit MPI data migration asynchronously, so it's all overlapped
|
12 éve |
|
Samuel Thibault
|
57720a2a82
Fix new features which actually appeared in 1.1
|
12 éve |
|
Samuel Thibault
|
c437b85f8b
Document how to perform MPI data migration
|
12 éve |
|
Samuel Thibault
|
c5bef53cf6
port r11793 from 1.1: Flush output at each line, to be able to check out progression
|
12 éve |
|
Samuel Thibault
|
96e6e58776
port r11793 from 1.1: Flush output at each line, to be able to check out progression
|
12 éve |
|
Samuel Thibault
|
804e6f8233
properly test for task termination
|
12 éve |
|
Samuel Thibault
|
e7d7ce19d4
typo
|
12 éve |
|
Marc Sergent
|
0026768555
Adding the handling of multiple fathers in hierarchical schedulers,
|
12 éve |
|
Marc Sergent
|
e998e008c6
Merge from trunk @11714:11788
|
12 éve |
|
Andra Hugo
|
2769aaa8cb
hypervisor: correct computation of idle
|
12 éve |
|
Samuel Thibault
|
4d81325ef7
fix missing initialization
|
12 éve |
|
Samuel Thibault
|
b8062a9fba
keep pushed back tasks in proper order
|
12 éve |
|
Andra Hugo
|
20906d8d77
hyp only: optimise resize (use idle and speed for the triggering policy)
|
12 éve |
|
Andra Hugo
|
6266721e89
do not take the mutex if the current worker is calling the code
|
12 éve |
|
Andra Hugo
|
39b06ba608
do not take changing_ctx_mutex when finishing executing the tasks (in decrement_nsubmitted_tasks), just check if the counter is 1, do what you have to do, and then decrement it (decrement the counter while verifying it wakes up the the thread calling wait_for_all)
|
12 éve |
|
Marc Sergent
|
fe685563e7
Adding several features which needs each other to compute correctly :
|
12 éve |
|
Samuel Thibault
|
b70f84e342
Properly find the task which optimizes computation time benefit
|
12 éve |
|
Samuel Thibault
|
a62e0dba32
better comment
|
12 éve |
|
Samuel Thibault
|
a06ea29938
Use separate expected ends, to avoid systematically preferring small tasks, possibly leading to starvation
|
12 éve |
|
Samuel Thibault
|
4f82d35ce9
Avoid keeping the lock held while scheduling tasks, to let scheduling happen in parallel. This means several processors might take 5 tasks each, thus emptying the fifo, but at least we don't have a huge contention point
|
12 éve |
|
Samuel Thibault
|
23b0ea64b9
Add a heft variant which draws tasks from its own fifo
|
12 éve |
|
Samuel Thibault
|
30c9f2732c
Plug perfmodel_select children normally from the scheduling strategy, instead of internally within perfmodel_select. This makes the room call go through the MCT node
|
12 éve |
|
Samuel Thibault
|
f7684ba8e2
keep the original data transfer time for the beta penalty, otherwise it seems we do not take communications into account enough to achieve overlapping. Proper communication scheduling would allow to replace this heuristic
|
12 éve |
|
Samuel Thibault
|
b693f565e0
Handle case when there is no worker any more for a given scheduling node
|
12 éve |
