Crypto
Home›Crypto›Market Structure›Ethereum prototype cuts blob reconstruction computing…
Ethereum prototype cuts blob reconstruction computing by up to 18x
Simulations of an alternate blob-recovery design showed network-wide reconstruction work dropping from 48.6 CPU-seconds under PeerDAS to 2.75 CPU-seconds in one tested setup.
An Ethereum prototype that splits blob-recovery duties across nodes showed an estimated 11 to 18 times reduction in reconstruction computing work in 1,000-node simulations, according to CryptoSlate. The work focuses on blobs, data that layer-2 rollups use.
The reduced design, described by researcher Csaba Kiraly in a Sept. 3 report, assigns recovery duties using existing mechanisms rather than adding the new row-networking channels envisioned in the full RowDAS networking proposal. CryptoSlate said it can reduce duplicated work by prioritizing particular blobs for high-custody nodes first, then sharing recovered data so other nodes can avoid rebuilding immediately.
CryptoSlate reported that in a configuration with four blobs, with 10% supernodes and no columns withheld, estimated network-wide reconstruction cost fell from 48.6 CPU-seconds under the PeerDAS model to 2.75 CPU-seconds under the reduced design. With a 20% supernode share, the figures were 91 CPU-seconds and 6.6 CPU-seconds, with the totals reflecting accumulated computing work rather than elapsed recovery time.
The prototype kept a PeerDAS-style delayed backstop for any still-missing data, but it does not add the extra resilience of full RowDAS. CryptoSlate also noted the measurements were limited to simulated, in-process networks using real cryptography, with a measured 162 millisecond cost per blob recovery on a Ryzen 9 8945HS processor.