Method and structure for pipelining of SIMD conditional moves
A mask is first generated in a general-purpose integer register. The mask is generated by executing a single instruction multiple data (SIMD) instruction on a plurality of operands stored in a plurality of registers and by writing the result to the general-purpose integer register. Next, a conditional-move mask is generated in a register using the mask, and then the conditional-move mask is used in selecting operands from the plurality of operands to generate a result in another register.
1. A computer-based method comprising:
generating a mask wherein said generating said mask includes:
executing a single instruction multiple data instruction on a plurality of operands stored in a first plurality of registers on a processor to generate said mask; and
storing said mask in an available register wherein said available register is one of a second plurality of registers on said processor;
generating a conditional-move mask in another register on said processor using said mask; and
using said conditional-move mask in selecting operands from said plurality of operands to generate a result in yet another register on said processor.
2. The computer-based method of claim 1 wherein said generating a mask further comprises:
comparing at least a first operand and a second operand in a first register in said first plurality of registers with at least a third operand and a fourth operand in a second register in said first plurality of registers wherein
said first operand is compared with said third operand;
said second operand is compared with said fourth operand;
said first operand is stored in a most significant field in said first register;
said second operand is stored in a second most significant field in said first register;
said third operand is stored in a most significant field in said second register; and
said fourth operand is stored in a second most significant field in said second register.
3. The computer-based method of claim 2 wherein said generating a mask further at least comprises
configuring a state of a most significant bit in said mask based upon a result of said comparison of said first and third operands; and
configuring a state of a second most significant bit in said mask based upon a result of said comparison of said second and fourth operands.
4. The computer-based method of claim 3 wherein said generating a conditional-move mask further at least comprises
using said state of said most significant bit in said mask in configuring a most significant field in said conditional-move mask.
5. The computer-based method of claim 4 wherein said generating a conditional-move mask further at least comprises
using said state of said second most significant bit in said mask in configuring a second most significant field in said conditional-move mask.
6. The computer based method of claim 5 wherein said using said conditional-move mask in selecting operands from said plurality of operands to generate a result in yet another register further comprises:
copying an operand from one of said first and second registers to said result register based on a value in said most significant field of said conditional-move mask.
7. The computer-based method of claim 1 wherein said generating a conditional-move mask further at least comprises
using a state of a bit in said mask to determine a state of a bit in a field of said conditional-move mask wherein said field of said conditional-move mask corresponds to said bit in said mask.
8. The computer-based method of claim 7 wherein said state of said bit in said corresponding field of said conditional-move mask is a logical inversion of said state of said bit in said mask.
9. The computer-based method of claim 1 wherein said using said conditional-move mask in selecting operands from said plurality of operands to generate a result in another register further comprises:
copying an operand from one of said plurality of registers to said yet another register based on a value in a sub-field of said conditional-move mask.
10. The computer-based method of claim 1 wherein said first plurality of registers comprises a plurality of floating point registers.
11. The computer-based method of claim 1 wherein said second plurality of registers comprises a plurality of general purpose integer registers.
12. The computer-based method of claim 1 wherein said yet another register is a floating point register.
13. A computer program product comprising a computer readable storage medium having embedded therein computer readable instructions for a method comprising:
generating a mask wherein said generating said mask includes:
executing a single instruction multiple data instruction on a plurality of operands stored in a first plurality of registers on a processor to generate said mask; and
storing said mask in an available register wherein said available register is one of a second plurality of registers on said processor;
generating a conditional-move mask in another register on said processor using said mask; and
using said conditional-move mask in selecting operands from said plurality of operands to generate a result in yet another register on said processor.
14. A system comprising:
a processor; and
a memory, coupled to said processor, storing instructions for a method comprising:
generating a mask wherein said generating said mask includes:
executing a single instruction multiple data instruction on a plurality of operands stored in a first plurality of registers on a processor to generate said mask; and
storing said mask in an available register wherein said available register is one of a second plurality of registers on said processor;
generating a conditional-move mask in another register on said processor using said mask; and
using said conditional-move mask in selecting operands from said plurality of operands to generate a result in yet another register on said processor.