fikunmi
@fikunmi_ap
Followers
799
Following
5K
Statuses
2K
Research @EclipseFND, prev @2077Research, and some control systems stuff| interested in decentralized governance| I also like chess and working out.
Joined March 2024
I have made some predictions; mostly about ETH L2s. But all the projects I consider top contenders for enabling "dApps at scale" get mentioned. The projects are: (@solana, @EclipseFND, @Aptos, @SuiNetwork, @megaeth_labs, @movementlabsxyz, and @monad_xyz) These are the prediction hashes are: add8e2a43d047ca08e721c64b6747936f291f4d253ab917a81dea1a2232f6e9a 5bc185d7347767956959416d0665859d5e02581ec4752f2788c6afb5e0d1a69c b6c2c8f83b51d55a81cfa2d013cb5a1ce8c0c06a0bc8cbe079a7008e458b81ee f7d5ca1b87e50500caaccd130c83e93f0e3d279202a0e00bd1dad216877a538d We'll see how these predictions age in a year or two. Honorable mentions for projects: @Starknet @zksync
2
2
19
@CantelopePeel Prices are not inherently bad but vanilla FPA are not great mechanisms in a static setting. Doesn't really matter what is being auctioned.
1
0
1
@CantelopePeel Not entirely sure what you mean but if it's what I think it is, it's still poor UX for users because they now have to reason about what their TX is worth to the validator, in addition to what others txs are also worth?
1
0
1
@blendino_ @toghrulmaharram I didn't say HFT, I said financial system; I mean **everything** money.
0
0
0
My apologies, I'm a believer in the one big fat chain thesis. Make no mistake, I also believe there'll be smaller chains used for other stuff but I think the new financial system will be a single chain and the unified liquidity will be it's edge. I'm not really interested in the other usecases because I don't get to push everything to the absolute max. But yeah, you're definitely correct that for other usecases, security, and ease of development will matter infinitely more.
1
0
1
@blendino_ @toghrulmaharram Bro, HFT firms like Jane Street handle 3 M messages per second while handling a tiny fraction of the world's traffic. 1M TPS is laughably small in the grand scheme of things. If there's a point of diminishing, we're so far, we shouldn't be asking if it exists yet.
1
0
0
No, that's an oversimplification. Monad and MegaETH raised because they're building performant EVM stacks, not because they're performant. If it was purely performance then it wouldn't make sense because they're both under the 1M TPS that Firedancer is supposed to do.. Shit, Monad's tangline is 10k TPS and 1s block times. Very performant bro. 👍 No shade to Monad though, I have the utmost respect for Keone, James and Zen as builders.
1
0
0
@toghrulmaharram If that's the context you're using it in then it hasn't been a moat since Solana launched. Let's see how it goes.
1
0
0
@MrPkar @toghrulmaharram Bruh, even just that, boy do I have some news for you 😬. Also what good is a TPU if other components can't keep up. Before we declare the performance wars over, let's actually see 1M TPS in prod 👍
2
0
1
RT @convoluted_code: Rust's surging popularity represents a strategic advantage for Eclipse. Our analysis below explores this in detail, ba…
0
2
0
I read the Aptos Shardines blog post and here's a short summary of what I think. Quick overview of how shardines works: A static analyser identifies dependencies in a list of transactions, splits the transactions across shards such that there are as few conflicts as possible and each shard uses Block-STM to execute. The differences between Shardines and vanilla sharding are: - static analysis to find dependencies. - the advanced sorting and reordering algorithms to minimize cross shard conflicts. I have my reservations about how well Shardines will perform under real workloads because of how (relatively) complex runtime analysis and the order and sort algorithms are, but it is undoubtedly impressive. Very Bullish Aptos. Also, bullish SVM; the fact that Aptos is adding a static analyser to improve performance reinforces my belief that state access lists approach pioneered by Solana is the most viable long term solution from a performance PoV. We already know with confidence that access lists are crucial for good fee markets, now I'm all but convinced now that they're also essential for performance. Chewing glass early remains undefeated. SVM stays winning.
0
0
2
@4lteredBeast @blknoiz06 You're correct but I think Ansem's PoV is that there are no more fundamentally redesigned stacks that will provide superior throughput
1
0
0
Execution is async on Aptos i.e the list of transactions to be executed are broadcasted before the actual execution. Additionally one of the core "findings" of Block-STM is that a strict ordering improves performance. There's some debate as to the veracity of this but it's true then idk how that interfaces with this proposal. Need to think about it more.
1
0
1
@bsturisky I’m not talking about the actual VM itself (although I’d like to hear why you think this), I’m talking about the execution engine and rest of the stack.
1
0
0