bread.mega@bread_
Solana Shreds vs Flashblocks
Not in-depth, and I'll put aggressive caveats on this that it's not comprehensive, but here is a visual comparison of Base touted Flashblocks implementation, Solana blocks and why the entire comparison is bogus.
----------------------------
First, we ask the question of "WTF is 'block times' meant to convey anyways?"
For EVM, it's the time it takes for state updates to propagate throughout the ecosystem (read: how long until apps can ingest new chain data?)
That's why it's an important update for Base - it's a 10x reduction in responsiveness/wait times.
But you cannot extrapolate that to Solana/SVM because it's block propagation pipeline is fundamentally different.
Solana streams block building continuously in what they call "shreds," and on a very, very fast timescale (~15ms batches). You can visually interpet it as the top of a block being sent out before the slots further down being assigned, making those top-of-block tx available to apps before the full 400ms block time is complete.
That's functionally what flashblocks are doing too.
Here is a visual comparing Base (traditional) and Base (flashblocks):
That bottom portion is ~basically how Solana works normally.
• Leader is given 400ms to construct a block
• Leader starts constructing a block
• Leader sends out portions of the block to the ecosystem before the 400ms is completed
• Leader finishes the block, rotates out
So if we understand that, we can visually compare the two in a more-authentic way (Solana is faster btw):
The next question is then "okay, the information is made available but what are the guarantees that applications can ingest the data reliably and it will stay true?"
That's where things get really nuanced, and to-date I haven't been given an answer as to whether or not dominant applications actually ingest Solana shreds as they're created or if they wait 400ms+ to build up confidence.
Base may be able to offer more guarantees because it doesn't have consensus (today) as an L2, but it still has to contend with Ethereum re-orgs/finality.
Anyways, busy day today and have to leave for some ETHDenver talks. Hope you learned something and Base won't be 2x faster than Solana with this implementation.