Processing-in-memory for genomics workloads

May 31, 2025 by W. Simon, Leonid Yavits, Konstantina Koliogeorgi, Yann Falevoz, Yoshihiro Shibuya, D. Lavenier, I. Boybat, Klea Zambaku, Berkan Sahin, Mohammad Sadrosadati, Onur Mutlu, A. Sebastian, R. Chikhi, The BioPIM Consortium, Can Alkan (IEEE Micro)

DOI 10.48550/arXiv.2506.00597

We built BioPIM to shove genomics compute into memory and kill the needless data shuttling that turns sequencing into a datacenter problem; by co-designing classic bioinformatics algorithms and data structures with PIM architectures we make high-throughput DNA/RNA analysis fast, cheap, and energy-sipping enough to move genomics out of the cloud and into the clinic.

source S2, openalex