NMP-PaK: Near-Memory Processing Acceleration of Scalable De Novo Genome Assembly DOI: 10.1145/3695053.3731056 Conference: The 52nd IEEE/ACM International Symposium on Computer Architecture; June 21–25, 2025; Tokyo, Japan ISBN: 979-8-4007-1261-6/2025/06

Heewoo Kim†, Sanjay Sri Vallabh Singapuram*, Haojie Ye*, Joseph Izraelevitz†, Trevor Mudge*, Ronald Dreslinski*, Nishil Talati* †University of Colorado, Boulder, CO, USA; *University of Michigan, Ann Arbor, MI, USA Email: [email protected] · 2025

Read NMP-PaK: Near-Memory Processing Acceleration of Scalable De Novo Genome Assembly DOI: 10.1145/3695053.3731056 Conference: The 52nd IEEE/ACM International Symposium on Computer Architecture; June 21–25, 2025; Tokyo, Japan ISBN: 979-8-4007-1261-6/2025/06 by Heewoo Kim†, Sanjay Sri Vallabh Singapuram*, Haojie Ye*, Joseph Izraelevitz†, Trevor Mudge*, Ronald Dreslinski*, Nishil Talati* †University of Colorado, Boulder, CO, USA; *University of Michigan, Ann Arbor, MI, USA Email: [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-08071.txt.

Description

Heewoo Kim and six co-authors designed a hardware accelerator for genome assembly in a 2025 paper. Assembling a genome from scratch, without a reference, opens unknown genomes to study but devours computing power and memory. The data structures are huge, irregular, and tangled together, which suits ordinary processors poorly. Their system, NMP-PaK, puts processing inside the memory itself.

Processing elements sit at the memory channels with scratchpad room for the assembly's shifting data structures. Software tricks include batch processing to shrink memory use and splitting work between the main processor and the memory units. The result: the same assembly with a fourteen-fold smaller memory footprint, sixteen times the baseline speed, and over eight times the throughput of the best distributed assemblers at equal resources.

Catalog record

CREATORHeewoo Kim†, Sanjay Sri Vallabh Singapuram*, Haojie Ye*, Joseph Izraelevitz†, Trevor Mudge*, Ronald Dreslinski*, Nishil Talati* †University of Colorado, Boulder, CO, USA; *University of Michigan, Ann Arbor, MI, USA Email: [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-08071

Browse the full library · About the Tower of Babel