Abstract This paper proposes as a database machine a system whose main computational engine is an FFP machine. The FFP machine is a small-grain, cellular tree computer with massive parallelism, which directly executes FFP languages. We describe a method to naturally embed relational query languages such as sql into FFP languages. We present FFP programs for some relational query operations (selection, projection, and join), and we show that they are executed efficiently by taking advantage of the massive parallelism accommodated by the FFP machine. Input/output architectures with varying degrees of parallelism are examined, and corresponding algorithms presented. Because the proposed system is also able to perform efficiently numeric and symbolic processing, it is appropriate for computation-intensive database applications such as expert database systems and engineering database systems.