 Samuel Thibault
|
f313079113
Factorize code to prepare for OpenCL pipelining
|
11 years ago |
|
Samuel Thibault
|
62e0ff89af
Fix missing include for windows port of locking
|
11 years ago |
|
Samuel Thibault
|
23636fb4f4
Fix simgrid build
|
11 years ago |
|
Samuel Thibault
|
77e107e4af
Warn if the user uses concurrent execution on a device which doesn't support it
|
11 years ago |
|
Samuel Thibault
|
62194c5c86
Fix filling the pipeline
|
11 years ago |
|
Samuel Thibault
|
f99ef13c88
Add CUDA kernel submission pipelining, to overlap costs and allow concurrent
|
11 years ago |
|
Samuel Thibault
|
647ccc35a7
Fix estimated allocated memory
|
11 years ago |
|
Samuel Thibault
|
41cfdc75cd
free data outside critical section
|
11 years ago |
|
Samuel Thibault
|
3b766361ce
free data outside critical section
|
11 years ago |
|
Samuel Thibault
|
206b3adf49
port r13269 from 1.1: free data outside critical section
|
11 years ago |
|
Samuel Thibault
|
aa5c3e5c50
Let interfaces declare which transfers they allow with the can_copy methode.
|
11 years ago |
|
Samuel Thibault
|
1ec6838c01
Implement perfmodel locking on windows
|
11 years ago |
|
Samuel Thibault
|
e5cf638f86
Pass codelet model when computing footprints for the bound computation
|
11 years ago |
|
Samuel Thibault
|
b294c2f942
Fix hardcoded limitation to 64 MPI nodes in traces
|
11 years ago |
|
Samuel Thibault
|
f8ff9c7d1c
Fix build without memcpy_peer
|
11 years ago |
|
Samuel Thibault
|
89ece6b14f
changelog
|
11 years ago |
|
Samuel Thibault
|
2c6e5a75b0
enable gpu-gpu transfers for matrices
|
11 years ago |
|
Samuel Thibault
|
d76f89457d
Separate bitfields set by starpu from bitfields set by user, since the latter are accessed unsafely by starpu, but the former shouldn't
|
11 years ago |
|
Samuel Thibault
|
2a1bb7bf49
Update 1.1.2 svn revision
|
11 years ago |
|
Samuel Thibault
|
14b8f531a6
Lock performance model files while writing and reading them to avoid issues on parallel launches, MPI runs notably.
|
11 years ago |
 Nathalie Furmento
|
034274135d
tests/perfmodels: add testcase to test memory is properly cleaned on
|
11 years ago |
|
Nathalie Furmento
|
5ad43ea8fd
src/core/perfmodel/perfmodel_history.c: improve memory management
|
11 years ago |
|
Nathalie Furmento
|
f4600eb1ba
merge trunk
|
11 years ago |
|
Nathalie Furmento
|
6d015be440
tests/main/{subgraph_repeat_tag.c,subgraph_repeat_regenerate_tag.c}: stop regenerating tasks when all loops are done, this permits to call starpu_data_unregister(), otherwise it would wait for the tasks
|
11 years ago |
|
Nathalie Furmento
|
f28a0d1765
src/core/perfmodel/perfmodel_history.c: remove wrongly committed fprintf
|
11 years ago |
|
Nathalie Furmento
|
3a634d01c2
src/core/perfmodel/perfmodel_history.c: do not allocate per_arch object if already allocated
|
11 years ago |
|
Nathalie Furmento
|
3fcbeee67e
examples/sched_ctx/nested_sched_ctxs.c: specify the task is parallel
|
11 years ago |
|
Nathalie Furmento
|
d47bd756f6
src/core/perfmodel/perfmodel_history.c: fix malloc call
|
11 years ago |
|
Nathalie Furmento
|
95991aa822
src/core/sched_ctx.c: fix #13209, we are looping on the masters, not the workers
|
11 years ago |
|
Nathalie Furmento
|
d87787c423
src/core/workers.c: revert #13221 to go back to #13209
|
11 years ago |
