Don't have Telegram yet? Try it now!
https://hgpu.org/?p=15974
Decoupled Vector-Fetch Architecture with a Scalarizing Compiler