EOS Development ahead of schedule
That was a nice development update yesterday:
Our original roadmap called for a single threaded implementation to be complete by June 2018 and for multi-threaded development to take place thereafter. We are excited to share that work on the parallel execution engine has begun 8 months ahead of schedule and we believe that it will be ready by June 2018.
For those that wonder what this means: The dev team has already tested up to 10.000 transactions per second on 1 thread. This means that a normal CPU should be able to handle 10K transactions per second. And this is in practise, not in theory because in theory it’s even more:
The final result yielded about 50,000 transfers per second average over many different runs. Keep in mind that these early results have many things that impact performance for better and for worse. It is too early to draw conclusions on the final chain performance, but 50,000 sequential actions per second is a lot closer to the area we want to be — Facebook and Visa and so on.
The EOS Block Producers need to sync all transactions as well and all needs to be read from memory etc. So Dan talks about 10k transactions per second per thread. So what do they mean with multi-threaded? You probably already guessed it: more transactions per second. In the dev group on Telegram Dan talks about adding 100 CPU cores to speeds things up after launch:
The fact that the devs are already implementing this before the first of June 2018 means that EOS will launch with huge capabilities when it comes to transactions per second. If Block Producers indeed run 100 cores then EOS could allow over 1 million transactions per second direct from the the start. Even though the target was “only” 10K to start with.
Being able to run thousands of transactions per second with your dApp is a dream come true for developers. You can create your own “Twitter” or “Facebook” fully protected by private keys. Each “like” and “post” are a blockchain interaction and EOS allows for a lot of them.
A million Ts/second from start is already good news. But there was more in the update:
We have been putting a lot of design work into how two blockchains will communicate with each other. This involves carefully crafting the structure of our merkle trees to make proofs meaningful and efficient. It also means restructuring our block headers and transaction headers.
So even though EOS might already support 1 million transactions per second, there’s still work on using more than just 1 blockchain. Where Proof-Of-Work blockchains create delay due to the search for the right hash, DPOS blockchains have a fixed time-slot when a block is produced. This means a faster settlement and also a faster “hop” from one chain to the other. That’s why adding several extra blockchains on EOS won’t cause a big delay. If they can get this to work properly EOS would be able to scale to even tens of millions transactions per second.
So for all the app-devs out there: Make sure to check EOS and imagine what you can do with thousands of transactions per second for you dApp. Don’t wait for months as we’re getting there quite fast.