Thanks for the link, that was a great read! I agree that the branch (target) prediction will fail horribly on a naive interpreter that calls different branch t

Thanks for the link, that was a great read! I agree that the branch (target) prediction will fail horribly on a naive interpreter that calls different branch targets from the same static instruction over and over again. That alone explains the low performance.