Processor for performing multiply-add operations on packed data
A method and apparatus for including in a processor instructions for performing multiply-add operations on packed data. In one embodiment, a processor is coupled to a memory. The memory has stored therein a first packed data and a second packed data. The processor performs operations on data elements in said first packed data and said second packed data to generate a third packed data in response to receiving an instruction. At least two of the data elements in this third packed data storing the result of performing multiply-add operations on data elements in the first and second packed data.
1. A processor comprising:
a memory to store one or more instructions including a multiply-add instruction;
a register file including a plurality of registers to store packed data including a first complex number and a second complex number; and
one or more execution units,
wherein the one or more execution units to multiply the first and second complex number in response to performing the multiply-add instruction,
wherein the one or more execution units to,
generate a result data using elements of the first and second complex number, the result data includes a first result data element (R) and a second result data element (I), the first result data element is equal to [(r1×r2)−(i1×i2)] and the second result data element is equal to [(r1×i2)+(r2×i1)], the first complex number includes a first real value (r1) and a first imaginary value (i1) and the second complex number includes a second real value (r2) and a second imaginary value (i2), and
store the result data in a destination indicated by the multiply-add instruction without adding the first result data element with the second result data element.
2. The processor of claim 1 , said one or more execution units, in response to performing the multiply-add instruction to:
generate a first, second, third, and fourth intermediate data elements, said intermediate data elements being paired into a first and second sets; and
arithmetically combine the intermediate data elements in each of said first and second sets to generate said first result data element and the second result data element respectively representing the addition of paired intermediate data elements, wherein each of said first, second, third, and fourth intermediate data elements provides a higher precision than said first and second real values or said first and second imaginary values.
3. The processor of claim 2 , said one or more execution units, in response to performing the multiply-add instruction to multiply together only corresponding data elements from said registers to store packed data including the first complex number and the second complex number to generate said first, second, third, and fourth intermediate data elements.
4. The processor of claim 2 , wherein said first and second real values and said first and second imaginary values are each stored as one or more packed signed 16-bit integer data elements.
5. The processor of claim 4 , wherein said registers to store packed data are to store the first result data element and the second result data element as packed data elements in the destination indicated by the multiply-add instruction without adding the first result data element with the second result data element.
6. The processor of claim 5 , wherein said first, second, third, and fourth intermediate data elements are each generated as signed 32-bit integer data elements.