Collision on large projectiles - multiple raycasts vs collision box?

For larger, medium-speed moving projectiles, I’m wondering which method would be better for collision detection - collision box or multiple raycasts?

Box would work nicely as long as it’s not travelling faster than it’s size, but I’m not sure how optimized that will be.

For the raycast method, having multiple raycasts to cover the separate sides of the projectile would work, but I’m not exactly sure how multiple raycasts vs 1 collision box matches up for better optimization?

Anyone have an idea of how multiple raycasts vs a single collision box measure up?

Try the ___Cast() methods, like SphereCast().