Samuel Thibault
|
91fd2cfc15
Fix mem_reclaim: on write request, if some requests are pending (notably reclaiming), create an additional write request which will wait for them before invalidating all copies
|
10 years ago |
Nathalie Furmento
|
bb61f3808d
update cnrs copyright
|
11 years ago |
Samuel Thibault
|
7b2d1621c4
* Add starpu_data_wont_use to advise that a piece of data will not be used
|
11 years ago |
Nathalie Furmento
|
7732f6f7a1
src/datawizard/user_interactions.c: rename internal tasks name to be clearer
|
11 years ago |
Samuel Thibault
|
4ee5550b33
introduce starpu_data_fetch_on_node for fetches which can forcibly flush data out
|
11 years ago |
Samuel Thibault
|
4c6de74837
Add starpu_data_idle_prefetch_on_node
|
11 years ago |
Samuel Thibault
|
4b1a16edde
Fix University name after the fusion..
|
11 years ago |
Samuel Thibault
|
f43bf86292
Fix complexity of implicit task/data dependency, from quadratic to linear
|
11 years ago |
Samuel Thibault
|
9d36c055e7
Make sure not to add the post_sync_task when none was define because sequential consistency was not enabled
|
11 years ago |
Samuel Thibault
|
4a9e46d095
Reduce the memory footprint of the "requested" bitfield
|
11 years ago |
Samuel Thibault
|
24f00257b5
Fix acquisition of data on node -1, re-enable using it in data_invalidate
|
12 years ago |
Samuel Thibault
|
c8c8eeeb31
More fixes for acquiring node -1 case. Still not working apparently, so not using it for data_invalidate for now.
|
12 years ago |
Samuel Thibault
|
515b80ef3f
Avoid making starpu_data_invalidate actually allocate RAM memory, by passing node -1 to acquire_on_node, and making the latter not allocate in that case. This fixes overzealous memory consumption by MPI layer
|
12 years ago |
Samuel Thibault
|
2f62b5cd10
Use big fat abortions when one tries to make a task or callback sleep, instead of just returning EDEADLCK which few people will test
|
12 years ago |
Nathalie Furmento
|
31f6ba2030
use STARPU_MAIN_RAM when appropriate
|
12 years ago |
Samuel Thibault
|
6dcf911b93
use STARPU_MAIN_RAM macro
|
12 years ago |
Nathalie Furmento
|
9928249727
jobs: remove field struct _starpu_job::model_name to avoid unnecessary
|
12 years ago |
Samuel Thibault
|
631d6e3185
port r 11514 from 1.1: Add unpartition state in trace
|
12 years ago |
Samuel Thibault
|
01aa31bde8
port r11297 from 1.1: Avoid looking through all data for not_important data if the application did not set any data as not_important.
|
12 years ago |
Samuel Thibault
|
2580597242
On data_acquire in codelet or callback (which is forbidden), raise an assertion failure instead of returning an error code that people don't usually check
|
12 years ago |
Nathalie Furmento
|
0d8d65bebb
data management: New functions
|
12 years ago |
Samuel Thibault
|
f2fffd0544
do not add the post_sync task if there is none. Fixes acquiring a data whose consistency flag is cleared
|
12 years ago |
Samuel Thibault
|
912cad9cce
make sure pre/post_sync_task are set in the non-sequential-consistency mode
|
12 years ago |
Samuel Thibault
|
27040a7d73
make assertions more verbose
|
12 years ago |
Nathalie Furmento
|
30860cac02
Renaming of diverse functions and datatypes.
|
12 years ago |
Nathalie Furmento
|
d61d7d6d9a
merge trunk -r 9571:9579
|
12 years ago |
Marc Sergent
|
ff83ffc582
Modifications of internal handling of tags in StarPU-MPI.
|
13 years ago |
Nathalie Furmento
|
7f792e3082
Define new public API starpu_pthread which is similar to the pthread API.
|
13 years ago |
Marc Sergent
|
df02c8376e
src/datawizard/user_interactions.c: in function starpu_data_release_on_node, change the order for functions _starpu_unlock_post_sync_tasks and _starpu_release_data_on_node as _starpu_unlock_post_sync_tasks needs to access handle which can be freed in _starpu_release_data_on_node
|
13 years ago |
Samuel Thibault
|
38e722860a
Initialize cond and mutex dynamically, so they can be destroyed dynamically by helgrind
|
13 years ago |