Instruction to normalize redundantly encoded floating point numbers
AssigneeIntel Corporation
InventorHarshvardhan Sharangpani
Life of this patent
Estimating the term…
Abstract
The present invention is an apparatus to normalize a floating point number. The apparatus has a first storage area comprising the floating point number. The floating point number comprises an exponent field and an explicit bit. The apparatus further comprises a circuit to normalize the floating point number when the explicit bit is not set and the exponent field has a first predetermined value identifying a redundant denormal encoding of the floating point number. Otherwise the encoding of the number is not changed by the circuit.
Claims22
An apparatus to normalize a floating point number in a unified format which is converted from an originating format, the apparatus comprising:
- a first storage area to store a floating point number, the floating point number comprising an exponent field, a significand field and an explicit bit;
- and
- a circuit to normalize the floating point number comprising:
- a detector coupled to the first storage area to detect a condition when the explicit bit is not set and the exponent field has a first predetermined value identifying a redundant denormal encoding of the floating point number in a unified format, the exponent field of the redundant denormal encoding in the unified format having an exponent value in the originating format equal to a sum of a minimum exponent value and an exponent bias in the unified format.
A computer system comprising:
- a processor to execute an instruction to normalize a floating point number in a unified format which is converted from an originating format, the floating point number comprising an exponent field, a significand field and an explicit bit, the execution of the processor causing the processor to at least
- detect a condition when the explicit bit is not set and the exponent field has a first predetermined value identifying a redundant denornal encoding of the floating point number in a unified format, the exponent field of the redundant denormal encoding in the unified format having an exponent value in the originating format equal to a sum of a minimum exponent value and an exponent bias in the unified format;
- and
- a memory to store the instruction, the processor fetching the instruction from memory before executing the instruction.
A computer system comprising:
- processor means to execute an instruction to normalize a floating point number in a unified format which is converted from an originating format, the floating point number comprising an exponent field, a significand field and an explicit bit, the execution of the processor causing the processor to at least
- detect a condition when the explicit is not set and the exponent field has a first predetermined value identifying a redundant denormal encoding of the floating point number in a unified format, the exponent field of the redundant denormal encoding in the unified format having an exponent value in the originating format equal to a sum of a minimum exponent value and an exponent bias in the unified format;
- and
- memory means to store the instruction, the processor means fetching the instruction from the memory means before executing the instruction.
A method of normalizing a floating point number in a unified format which is converted from an originating format in a computer system, the floating point number comprising an exponent field, a significand field and an explicit bit, the method comprising:
- normalizing the floating point number, the normalizing comprising:
- detecting a condition when the explicit bit is not set and the exponent field has
- a first predetermined value identifying a redundant denormal encoding of the floating point number in a unified format, the exponent field of the redundant denormal encoding in the unified format having an exponent value in the originating format equal to a sum of a minimum exponent value and an exponent bias in the unified format.
A method of finding a location index for a set bit at a left-most position in a string of bits, the method comprising the steps of:
- dumping the string of bits into a significand field of a floating point number, the floating point number further comprising an exponent field and an explicit bit;
- setting the exponent field of the floating point number to a first predetermined value identifying a redundant denormal encoding of the floating point number;
- normalizing the floating point number when the explicit bit is not set and the exponent field has a first predetermined value identifying a redundant denormal encoding of the floating point number, the normalizing step converting the exponent field to a normalized exponent value;
- and
- subtracting the normalized exponent value from the first predetermined value to obtain the location index.
Dependencies are inferred from the claim text.