Acurast, a decentralized compute network, has announced its Milestone Cray research and development upgrade, introducing native support for compute clusters, as disclosed to Finbold on February 5.
Named after Seymour Cray, a modern supercomputing pioneer whose ideas made computation essential to science, engineering, and national research, the update aims to maximize performance through coordination rather than centralization.
Milestone Cray is set to roll out on a testnet first, with a mainnet launch planned after validation and performance testing.
What does the Milestone Cray upgrade bring?
The new model has the ability to link multiple independent devices into a single logical compute unit. Instead of treating each phone or node as an isolated worker, this means that Acurast can link devices into coordinated clusters using local networks, or even via USB-C.
With these connections, users can split, synchronize, and recombine their tasks more efficiently to avoid overwhelming workloads.
Until now, memory and processing limits of individual workers have greatly limited decentralized compute. For instance, on Acurast, the cap was set at what a single phone could handle. Milestone Cray is expected to address this issue by pooling resources across clustered devices.
“Farms” as as unified, high-performance entities
The upgrade also changes how “farms”, i.e., groups of devices operated by a single participant, are utilized and rewarded.
Previously, these were treated as collections of independent processors. Now, however, they can act as unified, high-performance entities.
Overall, this approach helps the network assign demanding tasks to more capable setups, which also implies high-value workloads will be more aptly rewarded.
Featured image via Shutterstock
Source: https://finbold.com/acurast-rolls-out-milestone-cray-to-bring-supercomputing-to-a-global-decentralized-network/
