Vela is a lightweight, scalar-controlled vector coprocessor for VexRiscv, built in SpinalHDL as a foundation for composable tensor acceleration.
Vela is still experimental. It is a working hardware lab, not a finished product platform.
Vela v1already has a documented ISA baseline in vela-isa-spec-v1Vela v1also has a documented microarchitecture baseline in vela-uarch-spec-v1- The current RTL generation entry point is
sbt "runMain vela.GenVela" - Vela is built around a lightweight vector ISA inspired by RVV, but full RVV compatibility is not the goal
This repo is based on Spinal Template Sbt
Generate the current Vela RTL:
sbt "runMain vela.GenVela"vela
├── build.sbt
├── docs
├── hw
│ ├── gen
│ └── spinal
│ ├── vela
│ └── vexriscv
├── README.md
├── scripts
└── verif
MIT. See LICENSE