L1-zkEVM breakout #05, June 10, 2026

Agenda

Project / work-stream updates:

More updates:

Meeting Time: Wednesday, June 10, 2026 at 15:00 UTC (60 minutes)

GitHub Issue

Meeting Summary:

This was the L1 CKVM breakout meeting number five focused on progress updates as of May 2026. The meeting covered multiple project updates including execution witness and guest program specifications, KVM guest API standardization, weak subjectivity sync improvements, proof infrastructure development, formal verification efforts, and EVM ASM implementation. Key updates included the release of version 0.4.0 for guest program inputs with schema ID prefix bytes for future proofing, seven accepted KVM guest API standards with two new drafts for runtime library and ELF loading, recursive proof proposals for reducing sync data requirements, stable performance of the Ethereum Foundation’s proving cluster, and significant progress on formal verification including RISC-V compliance and EVM opcode implementation in Lean. Additional updates covered memory optimization improvements in Etrex, ongoing work on blocking blobs implementation, and progress on EVM ASM development with 110 opcodes currently formally specified and proven.

Click to expand detailed summary

The meeting focused on updates for the L1 CKVM breakout session, with Ignacio leading the discussion. Key updates included the proposal of EIP 1825 for optional proofs in Heiota, progress on execution witness and guest program specs with two releases (0.4.0 and 0.5.0), and the integration of multiple execution environments (ELs) into the Hive dashboard for compliance testing. The team discussed security findings, including bugs in CISC and libSSC, and progress on benchmarking and metrics work. The Ethereum Foundation’s proving cluster was reported as ready, and the conversation ended with a transition to project updates from Marcin.

Marcin provided an update on the KVM guest API standardization project, reporting that seven standards have been accepted and one is in review. He introduced two new drafts: one defining the runtime with static library and linker script, and another addressing ELF file loading standards to ensure consistent memory initialization across different ZKVMs. Marcin also presented a draft for accelerated memory operations and a logging function proposal, noting that three standards remain open for review.

Francesco presented updates on weak subjectivity sync, proposing the introduction of recursive proofs to reduce the data required for syncing from checkpoints to the chain tip. He also mentioned the migration of Lighthouse to upstream and the extraction of proof engine and gossip components. Han provided an update on the proof of info project, noting the stable performance of the EF cluster and ongoing benchmarks. The team discussed the complexity of managing proof generation infrastructure and acknowledged the need for further experimentation.

Cody provided an update on the formal verification effort for SP1, highlighting that while significant progress was made, there are remaining issues with RISC-V compliance and the need to focus on spec compliance. He also mentioned ongoing work on formal verification for ZISC and the development of the CompPoly library for executable specifications. The meeting then transitioned to the next speakers, with Ignacio indicating that each of the remaining speakers would have about 3-4 minutes to present.

Ben presented his work on optimizing guest programs for ZkVM, including fixes for memory issues, improvements in memory reuse, and code refactoring that resulted in speedups and reduced memory usage. Derek provided an update on EVM ASM, an assembly-level implementation in Lean, reporting progress on formal proofs for opcodes and state transition functions, with plans to integrate it into CI and expand code generation. Peter shared his progress on a prototype implementation of blocking blobs, identifying some spec issues and complexities with KZG libraries, while noting that Ivan’s team is also planning to work on this effort.

Next Steps:

  • ignacio: Share the link to the EIP 1825 slides from the ACDC presentation to the group.
  • ignacio: Coordinate a sync with Homer to try integrating a guest program into the Hive dashboard.
  • ignacio: Work with Mark from Protocol Support to clarify the recording/upload automation trigger word.
  • ignacio: Collaborate with Maria on the new benchmarking repository to help produce repricings for CKVMs after benchmark runs.
  • ignacio: Plan a separate dedicated call with Derek to discuss the EVM ASM project in more detail for broader community awareness.
  • ignacio: Eventually compile a summary of Ben’s optimization findings and share hints/lessons with other EL teams.
  • Marcin: Push the static library and linker script standard toward merge (targeting next week).
  • Marcin: Encourage community review of the three open standards: static library/linker script, accelerated memory operations, and the logging function PRs.
  • Francesco: Continue rebasing Lighthouse on upstream Unstable (GLOASS) and extract the proof engine and gossip components for the initial upstream branch.
  • Francesco: Evaluate whether recursive proofs can mitigate the need for request-response protocol complexity in Lighthouse.
  • Francesco: Continue work on the Block IR / execution layer intermediate representation and gather feedback from formal verification contributors.
  • Han: Complete a larger set of mainnet blocks and full EST blocks benchmarks on the EF proving cluster and share concrete numbers at the next breakout call.
  • Han: Continue fine-tuning the DISK distributed prover to keep all GPUs busy.
  • cody: Continue formal verification efforts for ZISC and prepare findings for broader review when ready.
  • cody: Ensure precise communication of remaining work and scope in all formal verification publications/communications.
  • ben: Investigate and compare different allocators for the guest program and continue profiling efrex.
  • ben: Share the write-up link on compiler optimization impacts in the meeting chat.
  • Derek: Finish remaining EVM opcodes (Cancun opcodes) and continue work on the state transition function in EVM ASM.
  • Derek: Integrate EST into CI and grow CodeGen to extract the full EVM.
  • Peter: Continue prototype implementation of blocking blobs in eTrax and Lighthouse, resolve KZG library complexities, and finalize the draft PR to the EIP with spec issue fixes.
  • Peter: Work toward a devnet to have the end-to-end blocking blobs flow running.

Recording Access:

YouTube recording available: https://youtu.be/5X5TsjIgFgU