 Samuel Pitoiset
|
a82def595d
hypervisor: fix double free in sc_hypervisor_get_ctxs_on_level()
|
10 年 前 |
|
Samuel Pitoiset
|
6f5db044a9
hypervisor: fix a deadlock when using lazy resize and perf counters
|
10 年 前 |
|
Samuel Pitoiset
|
bdae1fb9cb
hypervisor: fix use of unused variables
|
10 年 前 |
 Andra Hugo
|
39ccc67d88
add hypervisor strategy monitoring perf counter (not portable, the counter reading the flops dependent on the architecture)
|
10 年 前 |
|
Andra Hugo
|
ba3e4d7403
small fixes done while writing my thesis + task window at context level (before pushing tasks to the scheduler) expressed w.r.t. to time and not number of tasks
|
11 年 前 |
 Nathalie Furmento
|
cc9aecb792
merge trunk
|
11 年 前 |
 Samuel Thibault
|
717fbf7ab1
Always call init_iterator, it does not make sense otherwise
|
11 年 前 |
|
Andra Hugo
|
d4e55d69c9
fixing mem corruption
|
11 年 前 |
|
Andra Hugo
|
5f388d2a95
hierarchical contexts - bug fixing and DEBUG macro
|
12 年 前 |
|
Andra Hugo
|
491a6c6124
fix hierarchical resizing of contexts
|
12 年 前 |
|
Andra Hugo
|
60d44eb737
resize hierarchic ctxs hierarchically :)
|
12 年 前 |
|
Nathalie Furmento
|
e97679a558
fxt: rename function starpu_trace_user_event() as starpu_fxt_trace_user_event()
|
12 年 前 |
|
Nathalie Furmento
|
56986e683e
rename starpu_get_nready_tasks_of_sched_ctx to starpu_sched_ctx_get_nready_tasks
|
12 年 前 |
|
Andra Hugo
|
bfb858fe1d
Add hierarchical contexts
|
12 年 前 |
|
Andra Hugo
|
e28f7852a3
hypervisor: steal work from other ctxs when the worker is idle
|
12 年 前 |
|
Andra Hugo
|
2769aaa8cb
hypervisor: correct computation of idle
|
12 年 前 |
|
Andra Hugo
|
20906d8d77
hyp only: optimise resize (use idle and speed for the triggering policy)
|
12 年 前 |
|
Andra Hugo
|
12dec4c5cc
hypervisor optimisation (avoid unnecessary notifications/locks/trylocks) + add the hyp the traces
|
12 年 前 |
|
Andra Hugo
|
9bb7bdb07f
Increment/decrement submitted/ready tasks per context (instead of doing it per context & globally)
|
12 年 前 |
|
Andra Hugo
|
5a22d83b5c
hypervisor: feft_lp: compute max by using the execution of the last sample instead of the idle time
|
12 年 前 |
|
Andra Hugo
|
c4873d4813
hypervisor lp: max is computed using the time of executing tasks instead of idle time
|
12 年 前 |
|
Andra Hugo
|
994728b26b
hypervisor lp: max computation for the contexts sharing resources ( do not consider shared workers as full workers)
|
12 年 前 |
|
Andra Hugo
|
1289532233
whenever there are a few resources the lp cannot assign, all contexts share these resources
|
12 年 前 |
|
Andra Hugo
|
082dc8501c
hypervisor: adapt lp solution when max has an important impact on the solution (exec lp twice for now, next co share resources)
|
12 年 前 |
|
Andra Hugo
|
52459c8508
*eager: wake up workers when we add them to a context
|
12 年 前 |
|
Andra Hugo
|
646c61a828
improve idle computations in order to better compute max_workers for the resizing lp
|
12 年 前 |
|
Andra Hugo
|
9502f53f25
introduce min max as a constraint for the linear program
|
12 年 前 |
|
Andra Hugo
|
9f20f53ce9
fix size_ctxs (do not consider ready tasks for size, only for resize), speed is computed on fixed value of the time
|
12 年 前 |
|
Andra Hugo
|
f108a67609
add a prolog callback to a task (+ example)
|
12 年 前 |
|
Andra Hugo
|
39cfbbcb7f
consider the flops of ready tasks when resizing and compute the speed for a sample of time too
|
12 年 前 |
