Multiple similar but different approaches are considered for Payload Chunking. The goal of this post is to explain differences, describe trade-offs and analyze how they interact with other EIPs. Motivation While there are many designs for payload chunking, they share some of the common properties: Instead of sending the entire block over the network, we split it in smaller chunks and send them separately Ideally, we would also like to be able to execute chunks as they arrive, even if they arrive out of order (aka chunk execution) In this simplified design, this allows us to go from this model of block propagation and execution (each unit is a block):
3/13/2026While working on prototyping Semantic Payload Chunking (EIP-8101) in Prysm, I noticed several things that I would like to highlight in order to help other developers while implementing the EIP. It's worth noting that my work was based on Fulu fork, and a lot of things would change with Gloas and ePBS (EIP-7732). Chunk Count In order for clients to know when validating individual Chunks and CALs (ChunkAccessLists) is finished, they need to have access to chunkCount. This is currently not part of the ExecutionPayload, but I think it should be. Block Body One of the main design questions is where in BlockBody to put Chunks and CALs. There are two most direcy ways of doing it:
3/3/2026While working on prototyping Semantic Payload Chunking (EIP-8101) in Geth, I noticed several things that I would like to highlight in order to help other developers while implementing the EIP. BAL vs CAL BlockAccessList needs very little modification in order to support ChunkAccessList use cases. In addition to already existing logic, we need to add following functionality: Tracing The AccessListTracer should for reads also keep track of the first tx that created the read. The reads shouldn't be removed if write happens in the later tx (but it's safe not to keep reads if write happened during the same or earlier tx). Buliding Chunk Access List (CAL) With the tracing modification and already existing data, we have enough data to build CAL. One just has to make sure that reads are removed from the CAL if there is a write inside the same chunk (or in any earlier chunk).
3/2/2026Summary The basic implementation of EIP-7782 was built in Geth and Prysm on top of Fusaka fork. In order to run devnet, we modified several other projects: Genesis generator Kurtosis package Dora Devnet can be run with following command using kurtosis:
9/7/2025