Using DMA engines for collectives (ConCCL) lifts concurrent compute-communication speedup on MI300X from 21% to 72% of ideal.
{ARK}:{GPU-driven} code execution for distributed deep learning,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AR 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Optimizing ML Concurrent Computation and Communication with GPU DMA Engines
Using DMA engines for collectives (ConCCL) lifts concurrent compute-communication speedup on MI300X from 21% to 72% of ideal.