Architecture and instruction set to support interruptible floating point division
AssigneeTexas Instruments Incorporated
InventorsPrasanth VISWANATHAN PILLAI, Venkatesh NATARAJAN, Alexander TESSAROLO
Life of this patent
Estimating the term…
Loading patent…
AssigneeTexas Instruments Incorporated
InventorsPrasanth VISWANATHAN PILLAI, Venkatesh NATARAJAN, Alexander TESSAROLO
Estimating the term…
A processor (and method) includes a core that performs a floating point division through execution of various instructions. The instructions include a sign, exponent, and mantissa (SEM) separation instruction which causes the core to extract the sign, exponent and mantissa values from numerator and denominator floating point numbers. The instructions also include an unsigned mantissa division instruction which cause the core to iteratively perform a conditional subtraction operation to compute a value indicative of a mantissa of the quotient. The instructions further include a merge instruction that causes the core to generate a quotient floating point number using the extracted sign and exponent from the SEM separation instruction and the value indicative of the mantissa of the quotient.
A processor, comprising:
A processor, comprising:
A method implemented on a processor core, comprising:
Dependencies are inferred from the claim text.