Robin Salen (@robinsalen) 's Twitter Profile
Robin Salen

@robinsalen

ZK Engineering at @0xPolygon. Previously @Toposware. he/him. Opinions are my own.

ID: 4513099229

linkhttps://github.com/Nashtare calendar_today17-12-2015 11:02:40

556 Tweet

360 Followers

192 Following

Irreducible (@irreduciblehw) 's Twitter Profile Photo

šŸš€Big news! Irreducible launches alpha-ready Binius library and its first application, an Ethereum state proving service. Here’s what it means: 🧵

Hugo Fartingale (@hugomartingale) 's Twitter Profile Photo

coinbase currently estimates 4.2 years to process transfers on solana forced diamond hands (based) i look forward to top blasting the 2029 picotop as a multibillionaire thank you coinbase

coinbase currently estimates 4.2 years to process transfers on solana

forced diamond hands (based)

i look forward to top blasting the 2029 picotop as a multibillionaire

thank you coinbase
Srinath Setty (@srinathtv) 's Twitter Profile Photo

Introducing Twist and Shout, new and fast arguments for proving memory ops! These schemes work well with elliptic-curve commitments and hash-based commitments over binary fields. Using Shout, we also obtain SpeedySpartan, a new SNARK where the prover only commits to its witness.

Introducing Twist and Shout, new and fast arguments for proving memory ops! These schemes work well with elliptic-curve commitments and hash-based commitments over binary fields. Using Shout, we also obtain SpeedySpartan, a new SNARK where the prover only commits to its witness.
henry 🌘 (@hdevalence) 's Twitter Profile Photo

Excited to share that Penumbra's decentralized proving network is already live! Other ZK proving networks start from the assumption that proving workloads are slow, requiring users to outsource compute to GPU clusters. But what if we made proofs too cheap to meter? What if we

Fede’s intern 🄊 (@fede_intern) 's Twitter Profile Photo

This update from Succinct is impressive in terms of performance, but I’m surprised Uma Roy and John Guibas didn’t mention the soundness bug we found in SP1 with 3MI Labs, Aligned, and LambdaClass. Bugs like this don’t come up often—there aren’t many zkVMs in

Radisav Cojbasic (@radi_cojbasic) 's Twitter Profile Photo

Hats down for Fede’s intern 🄊 and CrypĀ·tomer (@ETHDenver) | 3MI Labs and everyone else who participated in this effort. I can’t believe that you got only $50k for this. Funds were at risk and it does not matter what Succinct says. SP1 is an open source zkVM which could be deployed by anyone, not only at

Robin Salen (@robinsalen) 's Twitter Profile Photo

« [saying] user funds were not at risk because the prover is permissioned shows me that we live in two very different worldsĀ Ā» This should *never* be a satisfying answer. Wherever there are bugs, there’s someone to exploit them. 100% aligned with Fede here. Fix bugs. Period.

Grigore Rosu (@rosugrigore) 's Twitter Profile Photo

This could be the most comprehensive benchmarking of the state-of-the-art zkVMs! Seven zkVMs were tested against more than 1225 input files! SP1 (Succinct) and RiscZero (RISC Zero) performed the best overall, but some did better on some inputs. Special thanks to the zkVM

This could be the most comprehensive benchmarking of the state-of-the-art zkVMs!  Seven zkVMs were tested against more than 1225 input files!  SP1 (<a href="/SuccinctLabs/">Succinct</a>) and RiscZero (<a href="/RiscZero/">RISC Zero</a>) performed the best overall, but some did better on some inputs.

Special thanks to the zkVM
Srinath Setty (@srinathtv) 's Twitter Profile Photo

Introducing Neo—not the One from the Matrix, but a new folding-based proof system, with: āœ… Efficient recursion (like Nova) āœ… Small fields: M61/Goldilocks (like STARKs) āœ… Pay-per-bit commitment costs (like Binius) āœ… Plausibly post-quantum secure šŸ”— eprint.iacr.org/2025/294 🧵

Ingonyama (@ingo_zk) 's Twitter Profile Photo

Barrett-Montgomery modular reduction, reimagined. A novel multi-precision algorithm that reduces computational complexity from n² + n to n² + 1 digit multiplications—significantly boosting performance in key primitives like NTT. Groundbreaking work by Yuval Domb. Read the

Barrett-Montgomery modular reduction, reimagined. 

A novel multi-precision algorithm that reduces computational complexity from n² + n to n² + 1 digit multiplications—significantly boosting performance in key primitives like NTT.  

Groundbreaking work by <a href="/yuval_domb/">Yuval Domb</a>.

Read the
Jim Posen (@jimpo_potamus) 's Twitter Profile Photo

This year, Irreducible will ship Binius into production. We're collaborating with industry-leading teams like RISC Zero and Polygon to develop groundbreaking zkVMs. Powered by our FPGA clusters, they will make verifiable computing practical for tons of new use cases.

Fede’s intern 🄊 (@fede_intern) 's Twitter Profile Photo

My partners at Aligned have created a webpage called ZK Benchmarks. A few days ago, I shared it, but now we've added several more results, and the results are fully reproducible with a Jupyter Notebook. We are adding every RISC-V zkVM, proving different examples, and

My partners at <a href="/alignedlayer/">Aligned</a> have created a webpage called ZK Benchmarks. A few days ago, I shared it, but now we've added several more results, and the results are fully reproducible with a Jupyter Notebook.

We are adding every RISC-V zkVM, proving different examples, and
BRIAN SĪžONG (@brianseong99) 's Twitter Profile Photo

#zkVMs need a benchmark challenge like #ARC_AGI is for #AI. Today, we're launching exactly that, this will serve as one of the core task for zkVM benchmarks in the future: Pessimistic Proof — a real-world zk workload used in Agglayer, now benchmarked across zkVMs, featuring

#zkVMs need a benchmark challenge like #ARC_AGI is for #AI.

Today, we're launching exactly that, this will serve as one of the core task for zkVM benchmarks in the future:

Pessimistic Proof — a real-world zk workload used in <a href="/Agglayer/">Agglayer</a>, now benchmarked across zkVMs, featuring
RISC Zero (@risczero) 's Twitter Profile Photo

3/ Since R0VM 1.0, we’ve cut proving time for Ethereum blocks from 35m to just 44 seconds (including execution) and we’re not done. We’re on track to reach real-time proving (<12s) by July.

3/ Since R0VM 1.0, we’ve cut proving time for Ethereum blocks from 35m to just 44 seconds (including execution) and we’re not done. 

We’re on track to reach real-time proving (&lt;12s) by July.