On the energy efficiency of sparse matrix computations on multi-GPU clusters

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.