 Samuel Thibault
|
0e385974a9
port r11312 from 1.1: Small CUDA allocations (<= 4MiB) are now batched to avoid the huge cudaMalloc overhead.
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
970c99c558
port r11310 from 1.1: Fix looking through a list while removing elements
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
e1501b5592
port r11308 from 1.1: record allocation size in trace
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
f850d2f26a
port r11305 from 1.1: the handle interfaces are already freed at this point, use the footprint pre-computed in the mc
|
%!s(int64=12) %!d(string=hai) anos |
 Nathalie Furmento
|
4bdac6b139
configure.ac: add option --coverage to CXXFLAGS (to avoid error error 'Hidden symbol atexit is referenced by DSO' when using libtool with gcov)
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
c4dc808530
port r11299 from 1.1: Use a hash table to speed up allocation cache reuse by hashing through the data footprint
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
01aa31bde8
port r11297 from 1.1: Avoid looking through all data for not_important data if the application did not set any data as not_important.
|
%!s(int64=12) %!d(string=hai) anos |
 Andra Hugo
|
c4873d4813
hypervisor lp: max is computed using the time of executing tasks instead of idle time
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
0bb629fed8
Avoid reverting the ordering of cache items
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
689a54fffe
fix doc and function prototypes
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
8d57290fd9
tests/main/starpu_init.c: use setenv instead of putenv, this is more valgrind friendly
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
e8a8200fc9
doc/doxygen/chapters/code/disk_copy.c: use setenv instead of putenv, this is more valgrind friendly
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
a2dd3ee669
tests/disk/disk_copy.c: use setenv instead of putenv, this is more valgrind friendly
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
a7383bc086
mpi/tests/insert_task_cache.c: use setenv instead of putenv, this is more valgrind friendly
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
e9fbf7551e
Add more CUDA path candidates
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
4cae2beddb
port r11283 from 1.1: Prefetch data for tasks with execute_on_a_specific_worker too
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
6f8c409637
tests/Makefile.am: have make clean delete core file (which is produced by expected to fail errorcheck/invalid_blocking_calls)
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
da74baa0b0
doc/doxygen/refman.tex: remove commands which are no longer valid with recent versions of doxygen (thanks again to Cyril Bordage for the BR)
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
af61053fe3
configure.ac: do not build documentation when option --disable-build-doc is set (Thanks to Cyril Bordage for the BR)
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
2d158cc65e
Allow to estimate task duration even if the task is not submitted yet (and thus the model is not loaded yet)
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
b217a5399a
data_requests[i] may change, it is still fine for _starpu_handle_node_data_requests to read it unprotected
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
ea8dc70b68
FIXME done
|
%!s(int64=12) %!d(string=hai) anos |
|
Samuel Thibault
|
054daa2f26
Fix memory leak during reduction
|
%!s(int64=12) %!d(string=hai) anos |
|
Andra Hugo
|
994728b26b
hypervisor lp: max computation for the contexts sharing resources ( do not consider shared workers as full workers)
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
30c460fb4d
mpi: forgot the implementation of starpu_mpi_insert_task
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
e030f1a6df
starpu_mpi_insert_task had been renamed in starpu_mpi_task_insert and
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
1b82b9cc99
starpu_insert_task had been renamed in starpu_task_insert and kept as
|
%!s(int64=12) %!d(string=hai) anos |
|
Nathalie Furmento
|
67ad578ac7
insert_task: new internal function _starpu_task_insert_v which takes a
|
%!s(int64=12) %!d(string=hai) anos |
|
Andra Hugo
|
1289532233
whenever there are a few resources the lp cannot assign, all contexts share these resources
|
%!s(int64=12) %!d(string=hai) anos |
|
Andra Hugo
|
082dc8501c
hypervisor: adapt lp solution when max has an important impact on the solution (exec lp twice for now, next co share resources)
|
%!s(int64=12) %!d(string=hai) anos |
