"StarkEx measurements break Ethereum's scalability record post-Instanbul"


bstr156 replying to StarkWareLtd
“Hi. ‘Measurements’, its nice to hear, but using what variables, and how? Measured within what context? Could be two nodes talking to each other on a closed network for all I know. Would like to know more so this can hopefully be taken with fewer grains of salt.”

Along with such goodies for comments, like:


Huge fan of the ZKtech Loopring is bringing to production, although I’m a bit biased as they use Chainlink With ETH2, will Rollups scale with the number of shards? Aka theoretically if all shards were flooded with ZKrollups, could we hit and maintain 160k TPS? (2500 x 64 shards)




Zk Rollup Scaling With Sharding? Eth 2.0 Could Hit 160k TPS???(64 Shards*2500 TPS ZKR)=160k. VitalikButerin Says Indeed. Thanks ChainLinkGod


Wow. That is intensely optimistic. Things like network latency do exist. I’d like to hear more perspective from the painfully objective 0xzak of whiteblock*

A lot of grounded folks would like some more valuable perspective on this. My impression here is a lot of optimism is being expressed, not only towards individual parts, also towards the sum of all parts. Have not gotten any responses to my comments thus far. That said, it’s only been ~1hr. However, I doubt anyone will respond with meat.

Has Whiteblock tested any of these solutions in a realistic net environ like how Syscoin Z-DAG was handled?

Hello! I saw some of your conversation on this topic in our Telegram channel, so thanks for sharing your post.

Has Whiteblock tested any of these solutions in a realistic net environ like how Syscoin Z-DAG was handled?

Yes, we have, but not in depth.

We worked with the Starkware team last summer to run some of these initial tests in support of EIP-2028 which reduces the cost of calldata. Their primary motivation was to run these tests against mainnet state, both within a simulated environment on the Genesis platform and on the actual Ethereum mainnet. We ran the tests to completion on the platform, but didn’t continue our research because the EIP was accepted in the Istanbul fork.

Given that this topic is picking up a lot of interest and attention, we’re currently having conversations with various stakeholders and considering revisiting this research to run these tests again under more scrutiny.

We’ll keep you updated in this thread and start a new one for this topic if we decide to put more resources toward these efforts.