remove the dependency of gdrcopy #201
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Right now we use either nvlink or ibgda, gdrcopy is not needed anymore.
I find that modifying the
nvshmemt_ibgda_progress
function can make nvshmem compile without gdrcopy. This can greatly simplify the process of installing DeepEP. We just need to compile nvshmem, with one additional driver flagoptions nvidia NVreg_EnableStreamMemOPs=1 NVreg_RegistryDwords="PeerMappingOverride=1;"
.NOTE: theoretically ibgda has a cpu-assisted mode that needs gdrcopy, however, the default mode is "auto", and if we compile without gdrcopy, the cpu-assisted mode will be turned off automatically. therefore, we don't need to do anything special in deepep.