The paper describes the Mark II digital-computing machine now operating in the Computing Machine Laboratory of Manchester University. It has a lOOkc/s parallel cathode-ray-tube store and a 1 Mc/s serial control, B-register and accumulator. Thus, internally, the speed of the machine is inherently greater than that of its predecessor by a factor of ten. Orders which concern the B-register or control are extracted from the store and obeyed in 60microsec.The machine has a floating-point accumulator, numbers being represented in the form xl y , where x and y are respectively 30 and 10 digits in length. Numbers to be added to the accumulator are automatically shifted appropriately before addition; they are then added and the sum is standardized. An order involving such a floatingpoint addition is extracted from the store and obeyed in 180microsec. Similar remarks apply to subtraction. For multiplication, which involves floating-point addition to, or subtraction from, the accumulator, the corresponding time is 360microsec.The number of thermionic cathodes in the Mark II machine is 57 % less than that in the Mark I machine.A comparison is made between the Mark II machine and a possible fixed-point version of it, so that the true importance of the floatingpoint accumulator can be ascertained. The authors arc members of the Electrical Engineering

No abstract

No abstract

No abstract

No abstract