More similar to SVE than NEON, supports both int and FP vectors. One instruction can process 256 / 512 / 1024 bits of vector at a time if you only need 16 / 8 / 4 vector variables (including masks) in your loop, not 32.

Full FPU and MMU, runs standard Linux e.g. Ubuntu server boots with about 36 MB free RAM.

To answer the previous poster, RV64GC is more similar to Thumb2 (but with 32 registers of 64 bits, not 16 registers of 32 bits) than to the cut-down Thumb1 (technically ARMv6-M) in the Pico. Plus, yeah, DP FPU and vector and MMU.