Nearly every OS that's been built is based on a 50 year old design choice. Usermode processes have their own virtual address space starting at 0, and OS kernel memory is mapped across all processes to the same location, usually at the top accessible after as system call.
But every kernel driver has access to the entire OS. Any bug in any driver can completely destroy the entire OS. @solana is not designed this way.
.@solana has stateless programs that are pure executable code. Transactions that isolate the state they touch and abort on any failure and execute in a fixed amount of resources. By definition a program can't crash.
These restrictions make it easy to build a high performance OS, but make it really hard for devs to build sophisticated programs. A program that is its own VM could work in theory, in practice, it going to take blood, sweat and tears.
.@neonlabsorg built a VM that runs as a simple program on @solana, but is not limited to its restrictions. Neon can execute EVM txs in parallel, that span multiple solana blocks. Let that sync in for a bit. Neon is a limitless EVM 🤯
This took an enormous amount of blood sweat and tears by the @neonlabsorg engineering team. Huge congrats to them on getting this done!
• • •
Missing some Tweet in this thread? You can try to
force a refresh