 Cédric Augonnet
|
02119279fc
We now check whether the following blocking calls are called from a callback or
|
16 years ago |
|
Cédric Augonnet
|
a9394e1af3
Keep the status of each worker so that we can detect whether a blocking call is
|
16 years ago |
|
Cédric Augonnet
|
4ac2e92f42
Add another classical synchronization problem to stress data management.
|
16 years ago |
|
Cédric Augonnet
|
152b6a7329
Add a the classical dining philosopher example: we have to explicitely order
|
16 years ago |
|
Cédric Augonnet
|
9e1bd3857d
Make it possible to select the maximum number of buffers per task at configure
|
16 years ago |
|
Cédric Augonnet
|
da38bcd033
Implement async data transfers for the "block" data interface.
|
16 years ago |
|
Cédric Augonnet
|
d3cfd654a8
Cleanup FxT tool, there is no svg output anymore (Vite is much better in every
|
16 years ago |
|
Cédric Augonnet
|
d8132d45cf
Make it possible to select the colour of the tasks in the DOT output depending
|
16 years ago |
|
Cédric Augonnet
|
82828a010d
On some systems (eg. MacOS), it's just not possible to bind threads at all, so
|
16 years ago |
|
Cédric Augonnet
|
98540ae1f7
don't compile CUDA functions if CUDA is not available
|
16 years ago |
|
Cédric Augonnet
|
5c984e9279
make sure that FFTW is available for the starpufft lib
|
16 years ago |
|
Cédric Augonnet
|
1773849014
Properly handle the case of write-only buffers when prefetching (in this case,
|
16 years ago |
|
Cédric Augonnet
|
277365f5c4
if a piece of data is accessed in a write only mode, there is no transfer
|
16 years ago |
|
Cédric Augonnet
|
272db710fc
We now store what is the mask associated to each worker, so that we can test if
|
16 years ago |
|
Cédric Augonnet
|
dc04c50247
Make the test more ""challenging"" (test the case where only a subset of the
|
16 years ago |
|
Cédric Augonnet
|
2e5a4054ec
cleanup tests
|
16 years ago |
|
Cédric Augonnet
|
e3defda082
Properly handle the case of 'write-only' buffers: there is no need for having a
|
16 years ago |
|
Cédric Augonnet
|
4ac9f355f0
In case we have multiple processes using StarPU (eg. with MPI), we need to
|
16 years ago |
|
Cédric Augonnet
|
e3ca8477e7
C++ (and therefore nvcc) requires that we don't use anonymous enum types.
|
16 years ago |
|
Cédric Augonnet
|
86f0556a11
- Add the starpu_execute_on_each_worker function. This function makes it
|
16 years ago |
|
Cédric Augonnet
|
7c31645307
The examples should not assume that StarPU was installed, we directly use the
|
16 years ago |
|
Cédric Augonnet
|
f33fc377f9
Get rid of the starpu-async branch where all recent changes have been made.
|
16 years ago |
|
Cédric Augonnet
|
18cac2a36a
Fix starpu_sync_data_with_mem by properly releasing the data once it has been
|
16 years ago |
|
Cédric Augonnet
|
62979fee84
add a sanity check that currently fails ... :/
|
16 years ago |
|
Cédric Augonnet
|
97c3bc1a57
we now assume that _fetch_data is called with the header_lock taken so that we
|
16 years ago |
|
Cédric Augonnet
|
adce7323af
properly update the state of the data even when it's just to invalidate a
|
16 years ago |
|
Cédric Augonnet
|
d28f2fc606
yet another patch to cleanup _fetch_data
|
16 years ago |
|
Cédric Augonnet
|
326d62a34f
get rid of the useless allocate_per_node_buffer function
|
16 years ago |
|
Cédric Augonnet
|
b110f4e08a
- add some sanity checks
|
16 years ago |
|
Cédric Augonnet
|
056e59082a
Factorize the data transfer code, it should be a little easier to define a new
|
16 years ago |
