 Samuel Thibault
|
7832518f4d
port r12742 from 1.1: Fix OpenCL device number detection when MAXOPENCLDEVS is not big enough
|
11 gadi atpakaļ |
 Nathalie Furmento
|
7891f63369
src/drivers/opencl/driver_opencl.c: revert #12566 and #12578, and explain why, the function availability would need to be checked at runtime, we do not want to do that
|
11 gadi atpakaļ |
|
Samuel Thibault
|
d4aeabce96
Also make sure that the header has the declaration
|
11 gadi atpakaļ |
|
Samuel Thibault
|
d0eab8ef8c
Do not rely on the .h file for functions actually available at link time, do check them explicitly
|
11 gadi atpakaļ |
|
Samuel Thibault
|
83a38dfa85
clEnqueueMarker is deprecated by clEnqueueMarkerWithWaitList. Be able to use both
|
11 gadi atpakaļ |
|
Samuel Thibault
|
092f322b1c
Add CUDA concurrent kernel execution support through the STARPU_NWORKER_PER_CUDA environment variable.
|
11 gadi atpakaļ |
|
Samuel Thibault
|
5a9ffce947
backport r12506: fix disabling out-of-order for data transfers
|
11 gadi atpakaļ |
|
Samuel Thibault
|
bf8aa6338d
Let the OpenCL driver progress while the GPU is computing
|
11 gadi atpakaļ |
|
Samuel Thibault
|
8f26d26ad7
Introduce the STARPU_CUDA_ASYNC and STARPU_OPENCL_ASYNC flags. No optimization is done yet for now.
|
11 gadi atpakaļ |
|
Samuel Thibault
|
8cc1f6c0db
Use separate OpenCL command queues for CPU->GPU, GPU->CPU, and GPU->GPU transfers
|
11 gadi atpakaļ |
|
Samuel Thibault
|
e4c9c8774a
Fix GPU memory size in the simgrid case
|
11 gadi atpakaļ |
|
Samuel Thibault
|
5ec6f8b8a7
Drop unused mask parameter of _starpu_fetch_task_input and _starpu_push_task_output
|
11 gadi atpakaļ |
|
Samuel Thibault
|
b1dfd30389
Store GPU size in platform file
|
11 gadi atpakaļ |
|
Samuel Thibault
|
e9e4a6361a
Use the same cuda/opencl exclusion mechanism on simgrid, instead of having a bogus assertion
|
11 gadi atpakaļ |
|
Samuel Thibault
|
6325a34ab3
Move device id lookup into run-driver functions, to avoid paying the cost at each task execution
|
11 gadi atpakaļ |
|
Samuel Thibault
|
d448281fc3
New STARPU_DISABLE_KERNELS environment variable to disable actual kernel execution.
|
11 gadi atpakaļ |
|
Samuel Thibault
|
89ac7e59ac
Provide hint for the user about missing func pointer
|
11 gadi atpakaļ |
|
Samuel Thibault
|
b801a66994
port r 11483 from 1.1: Default to using only 90% of the available GPU memory, to avoid seeing cudaMemset run out of memory..
|
11 gadi atpakaļ |
|
Samuel Thibault
|
4ce5fe9027
port r11357 from 1.1: Fix memleak
|
11 gadi atpakaļ |
|
Nathalie Furmento
|
483f8e2979
merge trunk
|
11 gadi atpakaļ |
|
Samuel Thibault
|
5bd1d28ab6
Fix worker initialization order: initialize fields before creating the first context.
|
11 gadi atpakaļ |
 Thibaud Lambert
|
e2dd19c39c
mic (perfmodels): merge trunk
|
11 gadi atpakaļ |
|
Nathalie Furmento
|
e0ea9be964
src/drivers/opencl/driver_opencl.c: properly set global size for the driver, and add it in worker name
|
11 gadi atpakaļ |
|
Thibaud Lambert
|
cb72812881
mic (perfmodels): merge trunk
|
11 gadi atpakaļ |
|
Thibaud Lambert
|
1e6badef4a
mic (perfmodel): Correction compilation error
|
11 gadi atpakaļ |
|
Samuel Thibault
|
a3f27f08ad
Make some warnings clearer. Avoid a very odd warning when the codelet has more parameters than allowed
|
12 gadi atpakaļ |
|
Nathalie Furmento
|
30860cac02
Renaming of diverse functions and datatypes.
|
12 gadi atpakaļ |
|
Nathalie Furmento
|
786b8fb4b7
move _STARPU_PTHREAD_XXX macros to public API starpu_thread_util and rename them to STARPU_PTHREAD_XXX
|
12 gadi atpakaļ |
|
Nathalie Furmento
|
097a832e3c
Allow to have a dynamically allocated number of buffers per task, and
|
12 gadi atpakaļ |
|
Nathalie Furmento
|
2cb9fcd148
use starpu_ssize_t instead of ssize_t to be windows-able
|
12 gadi atpakaļ |
