L1-zkEVM Breakout #009
Transcript
- Manu
GM!
- Francesco Risitano (Tau)
GM
- ignacio
My slides: https://docs.google.com/presentation/d/1-I1WpEpRtXy-Zx_HSHFvSWnSCfQKApIZY3H1o-jBzqA/edit?usp=sharing Cody slides: https://codygunton.github.io/talks-and-writing/2026-10-14-zkevm-breakout
- Iván | ethrex
Reacted to "My slides: https://d..." with 🙏
- ignacio
Yep, trying to finish the review soon :D
- Manu
- Manu
- Manu
Let’s find a good way to implement a GET proof
- Manu
So dora is happy
- ignacio
Reacted to "So dora is happy" with 😄
- Manu
If you can share kurtosis config file for LH I’ll happily test locally together with Prysm @Francesco Risitano (Tau)
- Francesco Risitano (Tau)
Reacted to "If you can share kurtosis config file for LH I’ll happily test locally together with Prysm @Francesco Risitano (Tau)" with 👍
- Francesco Risitano (Tau)
Excellent yes I can share the config with you when the corner cases are ironed out
- Francesco Risitano (Tau)
https://frisitano.github.io/slides/decks/optional-proofs-2026-10-14/index.html?v=20261014133540#2
- ignacio
Reacted to "https://frisitano.github.io/slides/decks/optional-proofs-2026-10-14/index.html?v=20261014133540#2" with 👍
- ignacio
Can share link to the slides? 🙏
- Somnath Banerjee
Will post the slides here soon.
- ignacio
Reacted to "Will post the slides here soon." with 🙏
- Somnath Banerjee
https://docs.google.com/presentation/d/1mf8FHquNqJrnRc4nlWhNY5k8zv2oa3lYNod4bZWLJAo/edit?usp=sharing
- Lion dapplion
Hello I’m here last minute
- Francesco Risitano (Tau)
Reacted to "Hello I’m here last minute" with 🤝
- ignacio
Reacted to "Hello I’m here last minute" with 🤝
- Francesco Risitano (Tau)
@ignacio can we give @Lion dapplion a few minutes to share an update after Zilkworm
- ignacio
Yep, after Q&A from Zilkworm presentation.
- Francesco Risitano (Tau)
Reacted to "Yep, after Q&A from Zilkworm presentation." with 🤝
- Łukasz Rozmej
airbender is riscv32?
- Paweł Bylica
Replying to "airbender is riscv32?" yes
- Francesco Risitano (Tau)
Can you share the blog for the witness design
- Tullio | Erigon
- Francesco Risitano (Tau)
Reacted to "https://zilkworm.hashnode.dev/introducing-mfbd-a-minimally-perfect-hash-based-flat-witness-format-for-zkevms" with 🤝
- Iván | ethrex
Anyone knows if the Airbender team is making any progress on security?
- Paweł Bylica
Replying to "Anyone knows if the Airbender team is making any progress on security?" I think v3 is 100 bit. Whatever this means.
- Iván | ethrex
Replying to "Anyone knows if the ..." That’s good news, AFAIK it had 80
- Iván | ethrex
Replying to "Anyone knows if the ..." Do you know if probable or conjectured?
- Paweł Bylica
Replying to "Anyone knows if the Airbender team is making any progress on security?" Or 100-bit is optional in v3. Probably worth starting with the release notes
- Iván | ethrex
Reacted to "https://zilkworm.has..." with 🤝
- Francesco Risitano (Tau)
Have you considered using the BAL as a means of informing you state structure?
- Somnath Banerjee
Yes, the vast majority of the work is from Node_store, where BALs don't do much
- Francesco Risitano (Tau)
Reacted to "Yes, the vast majority of the work is from Node_store, where BALs don't do much" with 👍
- ignacio
Thanks a lot again Zilkworm for the presentation!
- Francesco Risitano (Tau)
It would be good to push the silkworm optimisations upstream or at least find a means of discussing them in more detail
- Somnath Banerjee
Reacted to "It would be good t..." with 🔥
- Gary Schulte
Reacted to "It would be good to push the silkworm optimisations upstream or at least find a means of discussing them in more detail" with 🔥
- Somnath Banerjee
Reacted to "Thanks a lot again..." with ❤️
Call summary
Targets
- •EF prover deployment on Sepolia — near term, pending Prysm/Lighthouse interop resolution - 00:50:07
- •Zilkworm Alpha 5 — native stateless input bytes support, two test fixes, full EERE integration out of draft - 01:00:20
Decisions
- •EIP-7709 CFI'd at ACDE; removes 'headers' field from execution witness by routing BlockHash through EIP-2935 contract - 00:21:37
- •Proof engine scope in consensus specs narrowed to verification only; proof generation/fetching is unspecified client responsibility - 01:03:20
Highlights
- Testing Progress:
- ·zkVM test release v21 based on Sepolia official tests; public keys field removed from stateless inputs - 00:23:54
- ·R2 bucket live for Sepolia stateless inputs (50k+ blocks); Hoodi and mainnet buckets planned as forks activate - 00:28:38
- ·Hive consumer now supports REST+SSZ variant of Engine API; Nimbus already merged implementation - 00:31:22
- Cl Optional Proofs:
- ·Consensus-spec refactor: proof engine scope reduced to verification only; client teams free to embed verifier or use HTTP sidecar - 00:44:09
- ·New draft EIP for recursive chain proofs: single proof covering weak-subjectivity checkpoint to head, reducing sync bandwidth by orders of magnitude - 00:46:26
- ·EF intends to deploy prover on Sepolia; client teams invited to join and test EIP-8025 on testnet - 00:50:07
- ·Prysm proof-of-concept EIP-8025 implementation complete (gnark verifier embedded in binary); running verifier node on Sepolia; incompatibility with Lighthouse 'proof type 0' under investigation - 00:50:40
- ·Lighthouse merged proof-engine sidecar support into Unstable; can now run Unstable without an EL; verifiers to be integrated into codebase later - 01:27:52
- Zilkworm Guest Client:
- ·Silkworm Alpha 4 registered in EERE benchmark; passes full zkVM test suite (2 known failures matching Reth); ~half the estimated proving cost of Reth on every block - 00:57:55
- ·Alpha 5 target: native stateless input bytes support (no adapter); two failing tests to be fixed - 01:00:20
- ·Alpha 2→Alpha 4: one-third reduction in executed cycles; SP1 Hypercube used as benchmark VM - 01:01:13
- ·Airbender V3 integration live: ~3.8s proof with two 25090 GPUs; JUMPDEST analysis accounts for up to 30% of EVM execution cycles - 01:12:54
- ·Zilkworm now fully open source (SDLC pipeline, CI, PRs all public) - 01:14:33
- ·MFBD flat witness format: minimally-perfect-hash-based, zero-copy cast from input bytes; DirectState ~11% faster than full reconstruction from stateless input bytes - 01:15:04
- Eip Proposals Glamsterdam:
- ·EIP-8025 (zkEVM stateless inputs) targeting CFI at ACDC tomorrow; attendance encouraged - 00:18:57
- ·EIP-7709 (BlockHash via EIP-2935 contract) CFI'd at ACDE; removes 'headers' field from execution witness - 00:21:37
- Zkevm Standards And Security:
- ·Hash-map collision attack described by Lukasz (Nethermind): all provers for a given zkEVM vulnerable simultaneously; new randomness-seeding standard proposed with OpenVM demo - 00:36:33
- ·RISC-V compliance gaps identified across tested zkVM teams; proposal to require tests alongside each new standard - 00:37:50
Action Items
- •EL/CL client teams and zkEVM stakeholders - Join ACDC tomorrow to support CFI vote for EIP-8025; teams with strong opinions should attend - 00:18:57
- •CL client teams (Prysm, Lighthouse, Francesco) - Agree on Beacon API GET method for execution proofs to support Dora explorer - 00:53:10
- •Zilkworm + guest program teams - Discuss upstreaming Zilkworm RISC-V optimizations (memcpy, byte-swap, JUMPDEST) to benefit all guest programs - 01:29:31
Key decisions
Proof engine scope in consensus specs narrowed to verification only; proof generation/fetching is unspecified client responsibility
Consensus-spec refactor narrowing proof engine scope to verification only
EIPs discussed
AI Disclaimer: Some content or metadata on EIPsInsight may be AI-inferred or automatically compiled. If you find any discrepancy, please contact us at dev@avarch.org.