You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
All code should be performance portable from the start rather than retrofitting.
No need for MPI parallelization since the size of the individual instances should be relatively small. With a device target, there's not much need to split the domain.
Test every combination of hardware and software libraries within the automated system. More robust nightly testing should be included.
Constructing the code in the kernel sense. Even if it doesn't perform well on the GPU, its worth using the GPU model for future considerations.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Target user is ExaWind
Performance portability
All code should be performance portable from the start rather than retrofitting.
No need for MPI parallelization since the size of the individual instances should be relatively small. With a device target, there's not much need to split the domain.
Test every combination of hardware and software libraries within the automated system. More robust nightly testing should be included.
Constructing the code in the kernel sense. Even if it doesn't perform well on the GPU, its worth using the GPU model for future considerations.
Maintain our own Spack package.
Beta Was this translation helpful? Give feedback.
All reactions