# ZK tldr ## 2024 Oct. ### zkVM testing https://vac.dev/rlog/zkVM-testing/ ## 2024 Sep ### ye zhang 訪談 https://x.com/btceden/status/1806633579159675065 ### PSE change of focus on adoption with Barry leaving ### ZK 硬體加速的展望和未來 https://hackmd.io/@Cysic/BJQcpVbXn ### Stop learning ZK https://medium.com/@laurippeltonen/stop-learning-zk-mathematics-ba1064d26fe0 ### ZK VVM Diara :::success Daira-Emma Hopwood, [Aug 31, 2024 at 19:10:53]: That is true, and I think VPVM is still one letter too long. The word "proof" in that acronym is redundant. @alanszepieniec is right: the focus is not on the proofs, it's on the verifiability. "Verifiable virtual machine" is the correct term scientifically. We want to verify every aspect of each execution of a virtual machine relative to its semantics. It's also catchy. VVM. You can pronounce it "vavoom" in your head. It's fun. Does it clash with any other commonly used three-letter acronym? No, not in this field. Does it bring in any unwanted baggage or is it unnecessarily restrictive? No. (Don't take my word for this, think about it. In particular, note that there's no inaccuracy in viewing satisfiability of an arithmetic circuit as an instance of a VM execution; "virtual machine" encompasses the whole range of machine models, not just ones with sequential steps of an instruction set. The more general usage is scientifically correct and historically justified.) Ok, ship it. They're called VVMs now. Daira-Emma Hopwood, [Aug 31, 2024 at 19:23:39 (Aug 31, 2024 at 20:52:54)]: Also proposed: "zk-VVM" for VVMs designed to have efficient zero-knowledge execution proofs. Technically, this doesn't mean that every execution proof has to be zk. But it does mean there has to be a concretely efficient way of making zk execution proofs. Ideally it should also mean that execution proofs that pass *the standard verification procedure* for that VVM *can* feasibly be made in zk. I don't think that's too restrictive in practice, so let's insist on it as part of the definition. Daira-Emma Hopwood, [Aug 31, 2024 at 20:38:38 (Aug 31, 2024 at 20:51:02)]: This by the way *solves* the problem that arguably caused the terminology misuse. People wanted a way to refer to, not the proofs of execution in isolation (for which "SNARK" or "zk-SNARK" will do fine, and is more than catchy enough), but the system and machine model for which an execution is being verified. There was no term for that, and "zk-VM" was the nearest one to reach for. But scientifically the term "zk-VM" makes no sense. "zk" what? Not the VM itself; a VM by itself is not characterized by having proofs of execution that could potentially leak information about that execution. A VVM, on the other hand, by definition has an execution proof system that may or may not have the zk property. Also, "zk-VVM" is short enough and easy enough to say relative to "zk-VM" that I think it will in practice be able to displace the latter, especially because it has the concrete advantage of resolving the ambiguity: if you use this term then you're obviously speaking precisely and *do not* mean a VVM with non-zk execution proofs. This is useful and I'm going to write a short eprint about it, because there's more to say than will fit in the margins of this group. (I'll make sure to acknowledge the group.) ::: ## 2024 Aug ### ~~Vessel~~ - [website](https://vessel.finance/) - [twitter](https://x.com/VesselFinance) - ZK order book - verifiable AMM-embedded limit order book (VAELOB) - It runs a ZKP protocol to produce proofs based on the sequencer’s execution trace, ensuring that operations adhere to specific rules, preventing unauthorized creation of funds or transfers. https://x.com/VesselFinance/status/1805899522364883006 - 可以想像成是中心化的 CEX，但這個 sequencer 執行的計算軌跡會被 prove #### 內容： 主要有三個 component: the sequencer, the prover, and the smart contract - The sequencer is a transaction execution engine, LIKE CEX - Prover: 負責產生 state machine 計算的證明 目前還沒看到他的 circuit 有公開，所以不知道用什麼，但可能高機率跟 scroll 重疊 隱私部分也不確定，他可能只有用到 succinct 的性質 ---------- ### Verifiable Information Dispersal (By HAO) 可驗證訊息傳播 https://x.com/ittaia/status/1821587801701568626 Context: TODO -------- ### ZK-Rollup Prover Markets (By HAO) ---- ### ~~"sequencing narrative" is doomed~~ Ben Fisch https://x.com/benafisch/status/1823521965476389121 --- ### TACEO (By HAO) https://taceo.io/ https://x.com/TACEO_IO Collaborative Circom is an implementation of collaborative SNARKs ----- ### ~~Cysic~~ ZK proof generation accelerate https://x.com/geeksweb3/status/1826259541719478460?s=12&t=CVNMlt9vq-cn22OKb2aUjg #### 內容 ASIC/FPGA/GPU 加速都做了 目前主要合作方是 Aleo and Scroll ZK 礦池 Cysic的礦池可以為礦工提供“一魚多吃”，就是用戶把自己手上的顯卡/專用設備接入進來，Cysic集中這些算力給Scroll和Aleo這些項目產 proof，換取對方給的Token獎勵，這些Token再被Cyisc礦場分發給礦工 ### Fabric - Fabric: 可驗證處理單元（VPU） - https://www.fabriccryptography.com/ - https://www.techflowpost.com/article/detail_19824.html - 這是一間 - 市場的話 我覺得像是 pqc （post quantum cryptography ）算是有淺在未來市場 - 現在楊博因他們都在做硬體加速，但都是從原本就有的 cpu gpu 去使用 - 如果有新的這種計算單元，我覺得是個蠻大的機會 - Q: 會有這麼多裝置需要特殊硬體嗎? - 但我的想法是可以對標 npu tpu 那種專為 ai 設計的計算單元, 不過目前比不上 AI 那種量級