Method and apparatus for handling overflow and underflow in processing floating-point numbers
DCFirst Claim
1. A method for processing floating-point numbers, each floating-point number having at least a sign portion, an exponent portion and a mantissa portion, comprising the steps of:
- converting a floating-point number memory register representation to a floating-point register representation;
rounding the converted floating-point number;
performing an arithmetic computation upon said rounded number resulting in a new floating-point value; and
converting the resulting new floating-point register value to a floating-point memory register representation.
2 Assignments
Litigations
0 Petitions
Accused Products
Abstract
A method for processing floating-point numbers, each floating-point number having at least sign portion, an exponent portion and a mantissa portion, comprising the steps of converting a floating-point number memory register representation to a floating-point register representation; rounding the converted floating-point number; performing an arithmetic computation upon said rounded number resulting in a new floating-point value; and converting the resulting new floating-point register value to a floating-point memory register representation. An apparatus for processing floating-point numbers comprising means for converting a floating-point value from a random access memory register to a floating-point unit register; means for selecting the type of accuracy required of the converted value; means for selecting the type of rounding required of the converted floating-point value; means for performing an arithmetic operation upon at least one converted floating-point value based upon said selected accuracy and rounding; means for chopping the result of the arithmetic operation to fit a mantissa portion of a floating-point value; and means for converting the resulting floating-point value from the floating-point register representation to the random access memory representation.
128 Citations
28 Claims
-
1. A method for processing floating-point numbers, each floating-point number having at least a sign portion, an exponent portion and a mantissa portion, comprising the steps of:
-
converting a floating-point number memory register representation to a floating-point register representation; rounding the converted floating-point number; performing an arithmetic computation upon said rounded number resulting in a new floating-point value; and converting the resulting new floating-point register value to a floating-point memory register representation. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
-
-
21. A method for processing floating-point numbers from memory register format to arithmetic unit register format and back, each floating-point number having at least a sign portion, an exponent portion and a mantissa portion, comprising the steps of:
-
selecting a primary load control mode for a memory register floating-point number to be converted; testing for a memory value state by accessing bit statuses of the sign portion, exponent portion and mantissa portion of the memory register floating-point number; converting as required by said primary load control mode the result of said testing of the memory value state into the arithmetic unit register; selecting a primary arithmetic control mode for an arithmetic register floating-point number to be computed upon; rounding the arithmetic register floating-point number as required by said primary arithmetic control mode; performing an arithmetic operation upon the rounded arithmetic register floating-point number; selecting a primary store control mode for said arithmetic register floating-point number to be converted; testing the exponent portion of the arithmetic register floating-point number; and converting as required by said primary store control mode the result of said testing of the exponent portion into the memory register.
-
-
22. An apparatus for processing floating-point numbers comprising:
-
means for converting a floating-point value from a random access memory register to a floating-point unit register; means for selecting an accuracy required of the converted value; means for selecting a rounding choice required of the converted floating-point value; means for performing an arithmetic operation upon at least one converted floating-point value based upon said selected accuracy and rounding; means for chopping the result of the arithmetic operation to fit a mantissa portion of a floating-point value; and means for converting the resulting floating-point value from the floating-point register representation to the random access memory representation. - View Dependent Claims (23, 24)
-
-
25. An apparatus for processing one or more floating-point numbers from memory register format to arithmetic unit register format and back comprising:
-
one or more priority decoder circuits for converting said floating-point number to be loaded from a random access memory register to a floating-point unit register; one or more masking circuits for setting the bits of said floating-point numbers from the random access memory register to the floating-point unit register; one or more priority encoder circuits for rounding said floating-point number to be operated upon in said arithmetic logic unit;
an arithmetic logic unit for performing arithmetic operations upon the floating-point register numbers;one or more priority encoder circuit for converting and encoding said floating-point number to be stored from the floating-point unit register to the random access memory register; and one or more masking circuit for setting the bits or said floating-point numbers from the floating-point unit register to the random access memory register. - View Dependent Claims (26, 27, 28)
-
Specification