 Nathalie Furmento
|
5f35f4c737
src/datawizard/interfaces/data_interface.c: rename internal function
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
59dc8c395f
src/datawizard: add missing include starpu_scheduler.h
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
ddf6a00144
src/datawizard/malloc: define prototypes
|
%!s(int64=13) %!d(string=hai) anos |
 Samuel Thibault
|
01764c5ddc
ignore race about the watchdog state, we are ok with it
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
15acb4be0f
doc/Makefile.am: add showcheck target
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
78f4820844
mpi/src/starpu_mpi.c: initialisation can only be done once mpi itself has been initialised
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
a31d46c22b
mpi/src/starpu_mpi.c: make sure all internal structures are initialised before starting mpi communication thread
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
7667e0b20a
tests/perfmodels/value_nan.c: delete temporary file
|
%!s(int64=13) %!d(string=hai) anos |
|
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..
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
5d3da08d0e
src/datawizard/interfaces/matrix_interface.c: fix cast
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
a9cd1dc130
port r11478 from 1.1: Fix comment
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
5c6c5c188f
src/datawizard/interfaces/matrix_interface.c: fix tab format
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
2437d1e11b
src/datawizard/interfaces/matrix_interface.c: when cudaMemcpy2D() fails, tries to copy by calling copy_any_to_any()
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
1e620a1aef
port r11472 from 1.1: Document that allocation cache is enabled by default
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
8b83bc72aa
port r11470 from 1.1: pin memory for matrix_as_vector
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
bb88fe0e68
port r11466 from 1.1: Even without allocation cache, data release is done lazily, we can not count on precision of value returned by starpu_memory_get_available
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
c0cb3f849f
port r 11459 from 1.1: Do not force calibration systematically when a history-based model is used. Let the scheduler trigger it when it actually needs an expected performance
|
%!s(int64=13) %!d(string=hai) anos |
 Andra Hugo
|
9bb7bdb07f
Increment/decrement submitted/ready tasks per context (instead of doing it per context & globally)
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
daaa65f811
tests/main/multithreaded.c: add error message
|
%!s(int64=13) %!d(string=hai) anos |
|
Andra Hugo
|
5a22d83b5c
hypervisor: feft_lp: compute max by using the execution of the last sample instead of the idle time
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
a4aece4c5d
port r11446 from 1.1: fix memleak
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
40c0bdba54
tools/mvsc: turn batch files silent again (the problem was not there)
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
8f69d75bdd
move implementation of starpu_task_list functions in public .h file
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
afe2f1f956
port r11442 from 1.1: Continue the execute_on_each_worker example with FFT instead of cublas, for completeness
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
dfd8911aa3
src/core/topology.c: set userdata to NULL to make sure it is going to be properly initialised in case we call starpu_init() again in the same application
|
%!s(int64=13) %!d(string=hai) anos |
|
Nathalie Furmento
|
0cf9cf4000
src/datawizard/malloc.c: fix misplaced }
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
3830a76d5b
port r11435 from 1.1: make sure we do not overflow
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
0fec9f8acd
port r11433 from 1.1: Fix simgrid simulation with batched allocation: we really need to return different allocated values
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
4eb6989308
Do not let the prio scheduler decrement task stats several times when popping a task with multiple available implementations
|
%!s(int64=13) %!d(string=hai) anos |
|
Samuel Thibault
|
020bb9e52a
port r11429 from 1.1: Enable gpu-gpu direct transfer. This does not seem to harm performance any more
|
%!s(int64=13) %!d(string=hai) anos |
