 Andra Hugo
|
71cc7e4997
don't indicate number of impls they will be computed when executed
|
11 years ago |
|
Andra Hugo
|
10fe309eac
compil issues
|
11 years ago |
 Nathalie Furmento
|
436a56e106
configure.ac: --enable-blas-lib now also accepts the value mkl (to be consistent with the other blas libraries)
|
11 years ago |
|
Andra Hugo
|
c873b63e88
small fix
|
11 years ago |
|
Andra Hugo
|
a9b35bd4a7
fix malloc & memset perf model
|
11 years ago |
|
Andra Hugo
|
a0f55874a4
fixes for valid model
|
11 years ago |
|
Andra Hugo
|
002b53889f
fix bugs with the ids of the arch combinations
|
11 years ago |
 Samuel Thibault
|
02ce5c2683
Drop spurious changes (again...)
|
11 years ago |
|
Andra Hugo
|
fca611cf90
small compil issue
|
11 years ago |
|
Samuel Thibault
|
fe9ca6b56f
port r12993 from 1.1: Fix crash with a lot of MPI nodes
|
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
|
2eaf836156
merge with trunk
|
11 years ago |
|
Samuel Thibault
|
f14e68dc2c
Separate worker and thread state, to see streamed kernel executions
|
11 years ago |
|
Samuel Thibault
|
02acdca9b3
drop spurious changes
|
11 years ago |
|
Samuel Thibault
|
e22c0d9891
Fix function name: the existing function deals with a thread, not a worker
|
11 years ago |
|
Samuel Thibault
|
3c1e8b4b7f
Do not register threads several times
|
11 years ago |
|
Samuel Thibault
|
9bd9a1d234
Add missing pre_exec hook for dm
|
11 years ago |
|
Samuel Thibault
|
35d4d64183
Do not take the first measurement into account, it is very often quite bogus
|
11 years ago |
|
Samuel Thibault
|
8d8e0903d0
port r12948 from 1.1: Add changelog entry
|
11 years ago |
|
Samuel Thibault
|
b98077fba7
Add STARPU_TRACE_BUFFER_SIZE environment variable to easily set the FxT buffer size
|
11 years ago |
|
Samuel Thibault
|
9218801246
Do not expose a known-to-be-racy-but-we-re-fine optimization to helgrind
|
11 years ago |
|
Samuel Thibault
|
f3423acafe
Set the worker status within the scheduling mutex section, for coherency with starpu_wakeup_worker()
|
11 years ago |
|
Samuel Thibault
|
ee0e188163
Separate bitfields protected by the sync_mutex from bitfields which are not
|
11 years ago |
|
Samuel Thibault
|
1bcd390134
Explain the assertion a bit
|
11 years ago |
|
Samuel Thibault
|
7dd6b42215
Document that blocking spinlocks are only available on Linux
|
11 years ago |
|
Samuel Thibault
|
039524c7cf
Spinlocks now block after a hundred tries. This avoids typical 10ms pauses
|
11 years ago |
|
Andra Hugo
|
3b1f62f519
create branch to modify the perf_model management in order to consider scheduling contexts too
|
11 years ago |
|
Samuel Thibault
|
58c32f0bc6
Note that r12914 from 1.1 was already in the trunk actually (Fix subtraction)
|
11 years ago |
|
Nathalie Furmento
|
3dfa294771
doc/tutorial/vector_scal_task_insert.c: wait for task completion
|
11 years ago |
|
Nathalie Furmento
|
022b2af020
tutorial: add a task_insert version of vector_scal
|
11 years ago |
