
Exascale computing designates high-performance computing systems capable of executing at least one exaflop — 10¹⁸ floating-point operations per second — a threshold formally crossed with the deployment of Frontier at Oak Ridge National Laboratory in 2022, followed by Aurora at Argonne and El Capitan at Lawrence Livermore.
The scientific applications that justify this infrastructure — full-resolution climate modelling, nuclear stockpile simulation, whole-brain neural circuit mapping, precision cosmology, and pandemic-scale epidemiological modelling — share a common profile: problems whose spatial or temporal resolution was provably capped by petascale arithmetic throughput.
The publications and patents indexed below document the hardware architectures, system software, programming models, and domain-specific applications that define the current exascale research frontier.
This is our latest selection of worldwide publications and patents in english on Exascale Computing, between many scientific online journals, classified and focused on exascale comput, exaflop, Frontier supercomputer, Aurora supercomputer, El Capitan supercomputer, exascale HPC, exascale workload, high-performance computing node, exascale interconnect, Slingshot interconnect, exascale storage hierarchy, parallel file system, liquid cooling HPC, exascale memory bandwidth, high-bandwidth memory HPC, exascale MPI, exascale OpenMP, exascale fault tolerance, checkpoint restart HPC, exascale, exascale accelerator and GPU cluster HPC.
Publication: no recent news on this particular topic. Please try the extensive manual search in the Publication Database linked just above.
Selective inline compression using data processing unit on a parallel file system
Patent published on the 2026-07-16 in US under Ref US20260202962 by DELL PRODUCTS L P [US] (Doddaiah Ramesh [us], Chen Xiangping [us], Krasner Jonathan Ichael [us])
Abstract: Embodiments disclosed herein relate to selective inline compression of data using data processing units (DPUs) of hosts running on the file system prior to storing the data on a remote storage system. To be compressed by the DPUs of the hosts, data should be not already compressed, the data should not subject to frequent read/write cycles off the remote storage system, and the DPU should have sufficient computing resources to compress the data. A file system is connected to the hosts and the rem[...]
Our summary: Selective inline compression is performed using data processing units (DPUs) on hosts before storing data in a remote storage system. The process relies on a file heatmap generated by machine learning to assess data access frequency. Only uncompressed data that meets specific criteria is eligible for compression by the DPUs.
inline compression, data processing unit, file system, machine learning
Patent

















