 Olivier Aumage
|
52ea41d70b
- add the possibility to associate some user data to sched contexts (needed for some INTERTWinE work)
|
9 years ago |
 Samuel Thibault
|
160b44ac45
inline the trivial functions accessing starpu_config
|
9 years ago |
 Terry Cojean
|
0248b236d2
Fix the trouble I created in r15580. Also add a default mode for _get_next_sched_ctx_to_pop_into()
|
10 years ago |
 Andra Hugo
|
43f256f1c4
deal with shared workers between contexts having no starpu scheduling policy (allow the strategy to schedule tasks on not blocked shared workers)
|
10 years ago |
|
Andra Hugo
|
305100c749
add Beranger's ordo
|
10 years ago |
|
Olivier Aumage
|
029217b712
- fix a race condition between a pthread_cond wake-up and a resource destruction
|
10 years ago |
|
Samuel Thibault
|
51d9fdc477
|
10 years ago |
|
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
|
10 years ago |
|
Andra Hugo
|
afade57efa
allow ctx to execute parallel codes by reusing starpu workers (keep the workers alive instead of getting them to sleep) -> TODO try to factorize with parallel tasks on combined workers (doing practically the same, duplicate tasks and barrier_wait before & after exec)
|
11 years ago |
 Nathalie Furmento
|
c506a5901d
src/core/sched_ctx: new internal function _starpu_sched_ctx_get_sched_ctx_for_worker_and_job to return a context if the worker is the master of a parallel context, and the job is meant to be executed on this parallel context
|
11 years ago |
|
Andra Hugo
|
5593439277
changing perf_model structure: arch now contains several devices, so we can have for eg one device STARPU_CPU with 2 cores one device STARPU_CUDA_WORKER with 1 core
|
11 years ago |
|
Andra Hugo
|
ab7bd6810c
nested scheduling contexts + worker collection keeping track of blocked workers
|
11 years ago |
|
Andra Hugo
|
c8def5adac
sharing workers btw ctxs with no sched_policy
|
11 years ago |
|
Andra Hugo
|
3ca3790ef1
* sched_ctxs without sched_policy
|
11 years ago |
|
Andra Hugo
|
5a219c8a8f
* patch Terry: fix book workers (book workers that have already been booked and wake up eventually workers that we don't want anymore in the group)
|
11 years ago |
|
Andra Hugo
|
6fd68c0e07
execute parallel code (openmp for eg) inside contexts ------ temporary solution
|
11 years ago |
|
Andra Hugo
|
bfb858fe1d
Add hierarchical contexts
|
11 years ago |
|
Andra Hugo
|
12dec4c5cc
hypervisor optimisation (avoid unnecessary notifications/locks/trylocks) + add the hyp the traces
|
12 years ago |
|
Nathalie Furmento
|
9bb5a09263
src: minor fixes
|
12 years ago |
|
Andra Hugo
|
9bb7bdb07f
Increment/decrement submitted/ready tasks per context (instead of doing it per context & globally)
|
12 years ago |
|
Nathalie Furmento
|
e9b6c7eb1a
src: core/task.c: add missign include - core/sched_ctx.h: define function outside ifdef
|
12 years ago |
|
Nathalie Furmento
|
c1bcca575d
The functions starpu_sched_set/get_min/max_priority set/get the
|
12 years ago |
|
Nathalie Furmento
|
43838444ce
Function starpu_sched_ctx_create() now takes a variable argument list
|
12 years ago |
|
Samuel Thibault
|
1ad22db615
port r11138 from 1.1: Replace global mutex by global rwlock
|
12 years ago |
|
Andra Hugo
|
c4873d4813
hypervisor lp: max is computed using the time of executing tasks instead of idle time
|
12 years ago |
|
Andra Hugo
|
52459c8508
*eager: wake up workers when we add them to a context
|
12 years ago |
|
Andra Hugo
|
f31697b342
regiseter callback for the context, to be executed when the last task finished executing
|
12 years ago |
|
Andra Hugo
|
93ed656614
add missing hyp functions to the doc + some code cleaning
|
12 years ago |
|
Andra Hugo
|
c6ae0527ab
workers keep a chained list of contexts (instead of a sparse list) and remove as many loops as possible at pop time
|
12 years ago |
|
Andra Hugo
|
5fa6ed1bb7
allow passing the pointer of a scheduler to a context
|
12 years ago |
