| .. |
|
perfmodel.c
|
2d158cc65e
Allow to estimate task duration even if the task is not submitted yet (and thus the model is not loaded yet)
|
лет назад: 13 |
|
perfmodel.h
|
497ce06798
merge branches/mic: update of perfmodel storage + update _STARPU_PERFMODEL_VERSION
|
лет назад: 13 |
|
perfmodel_bus.c
|
020bb9e52a
port r11429 from 1.1: Enable gpu-gpu direct transfer. This does not seem to harm performance any more
|
лет назад: 13 |
|
perfmodel_history.c
|
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
|
лет назад: 13 |
|
perfmodel_nan.c
|
8fced1a509
src/core/perfmodel/perfmodel_nan.c: ungetc cannot be called several times, it is not safe
|
лет назад: 13 |
|
perfmodel_print.c
|
3afcf39929
mic (perfmodels): merge trunk
|
лет назад: 13 |
|
regression.c
|
74889b4885
perfmodel: rename struct starpu_history_list to struct starpu_perfmodel_history_list
|
лет назад: 14 |
|
regression.h
|
74889b4885
perfmodel: rename struct starpu_history_list to struct starpu_perfmodel_history_list
|
лет назад: 14 |
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