1988 Ed.]
Import and Export (Strategic Commodities) Regulations
[CAP. 60
G 113
[Subsidiary]
"total processing data rate"—
(a) Of a single central processing unit, is its 'processing data rate';
(b) of multiple central processing units which do not share direct access to a common "main storage", is:
The individual 'processing data rate' of each central processing unit, i.e., each unit is separately treated as a single central processing unit as in (a) above; or
(c) of multiple central processing units, which partially or fully share direct access to a common "main storage" at any level, is the sum of:
(1) The highest of the individual 'processing data rates' of all central processing units; and
(2) 0.75 times the 'processing data rate' of each remaining central processing unit, sharing the same "main storage";
assuming the configuration of equipment, which would maximize this sum of rates.
'processing data rate'—
The maximum of either:
(a) the 'floating point processing data rate' (Rf); or
(b) The 'fixed point processing data rate' (Rx).
NB:
The 'processing data rate' of a central processing unit implemented with two or more microprocessor microcircuits, not including any dedicated microprocessor microcircuit used solely for display, keyboard or input/output control, is the sum of the individual 'processing data rates' of all these microprocessor microcircuits.
'floating point processing data rate (Rf)'—
The sum of
(1) 0.85 times the number of bits in a 'fixed point instruction' (nix) or 0.85 times the 'number of bits in a floating point instruction' (nif), if no fixed point instructions are implemented;
(2) 0.15 times the 'number of bits in a floating point instruction' (nif);
(3) 0.40 times the number of bits in a fixed point operand (nox) or 0.40 times the number of bits in a 'floating point operand' (nof), if no fixed point instructions are implemented; and
(4) 0.15 times the 'number of bits in a floating point operand' (nof);
divided by the sum of
(1) 0.85 times the 'execution time' for a fixed point addition (tax) or for a floating point addition (taf), if no fixed point instructions are implemented;
(2) 0.09 times the 'execution time' for a floating point addition (taf); and
(3) 0.06 times the 'execution time' for a floating point multiplication (tmf) or for the fastest available subroutine (sub) to simulate a floating point multiplication instruction, if no floating point multiplication instructions are implemented.
Thus:
Rf = (0.85)nix + (0.15)nif + (0.40)nox + (0.15)nof / (0.85)tax + (0.09)taf + (0.06)tmf
or if no fixed point instructions are implemented then:
Rf = (1.00)nif + (0.55)nof / (0.94)taf + (0.06)tmf
or if no floating point multiplication instructions are implemented (tmsub) then:
Rf = (0.85)nif + (0.15)nif + (0.40)nox + (0.15)nof / (0.85)tax + (0.09)taf + (0.06)tmsub
NB:
If a "digital computer" has neither floating point addition nor floating point multiplication instructions, then its 'floating point processing data rate' is equal to zero.