Vector floating-point classification

1. A method to classify source data in a processor in response to a vector floating-point classification instruction, the method comprising:

receiving the vector floating-point classification instruction, wherein the vector floating-point classification instruction includes:

a source register field that specifies a source register containing the source data, wherein the source register comprises a plurality of lanes that each contains a floating-point value;

a destination register field that specifies a destination register to store classification indications for the source data, wherein the destination register comprises a plurality of lanes corresponding to the lanes of the source register; and

a third field that specifies a datapath from among a scalar datapath and a vector datapath to execute the vector floating-point classification instruction; and

executing the vector floating-point classification instruction by the datapath specified by the third field, wherein executing the vector floating-point classification instruction further comprises, for each lane in the source register:

classifying the floating-point value in the lane to identify a type of the floating-point value; and

storing a value indicative of the identified type in the corresponding lane of the destination register.

2. The method of claim 1 , wherein the source data comprises a 512-bit vector.

3. The method of claim 1 , wherein each floating-point value comprises a single precision floating-point value.

4. The method of claim 1 , wherein each floating-point value comprises a double precision floating-point value.

5. The method of claim 1 , wherein the type comprises one selected from the group consisting of: a zero value, a subnormal value, a normal value, an infinite value, a not a number (NaN) value, a quiet NaN (QNaN), and a signaling NaN (SNaN).

6. The method of claim 5 , further comprising:

storing a value of 0 in the corresponding lane of the destination register in response to the identified type being a zero value;

storing a value of 1 in the corresponding lane of the destination register in response to the identified type being a subnormal value;

storing a value of 2 in the corresponding lane of the destination register in response to the identified type being a normal value;

storing a value of 3 in the corresponding lane of the destination register in response to the identified type being an infinite value;

storing a value of 4 in the corresponding lane of the destination register in response to the identified type being a NaN value; and

storing a value of 8 in the corresponding lane of the destination register in response to the identified type being a SNaN value.

7. A data processor, comprising:

a scalar datapath;

a vector datapath coupled in parallel with the scalar datapath;

a source register configured to contain source data; and

a destination register;

wherein the source register comprises a plurality of lanes that each contain a floating-point value and the destination register comprises a plurality of lanes corresponding to the lanes of the source register;

wherein, in response a single vector floating-point classification instruction that specifies a datapath from among the scalar datapath and the vector datapath, the datapath of the data processor is configured to, for each lane in the source register:

classify the floating-point value in the lane to identify a type of the floating-point value; and

store a value indicative of the identified type in the corresponding lane of the destination register.

8. The data processor of claim 7 , wherein the source data comprises a 512-bit vector.

9. The data processor of claim 7 , wherein each floating-point value comprises a single precision floating-point value.

10. The data processor of claim 7 , wherein each floating-point value comprises a double precision floating-point value.

11. The data processor of claim 7 , wherein the type comprises one selected from the group consisting of: a zero value, a subnormal value, a normal value, an infinite value, a not a number (NaN) value, a quiet NaN (QNaN), and a signaling NaN (SNaN).

12. The data processor of claim 11 , wherein, in response to execution of a single vector floating-point classification instruction, the data processor is further configured to, for each lane in the source register:

store a value of 0 in the corresponding lane of the destination register in response to the identified type being a zero value;

store a value of 1 in the corresponding lane of the destination register in response to the identified type being a subnormal value;

store a value of 2 in the corresponding lane of the destination register in response to the identified type being a normal value;

store a value of 3 in the corresponding lane of the destination register in response to the identified type being an infinite value;

store a value of 4 in the corresponding lane of the destination register in response to the identified type being a NaN value; and

store a value of 8 in the corresponding lane of the destination register in response to the identified type being a SNaN value.

13. The method of claim 1 , wherein the vector floating-point classification instruction includes a fourth field that specifies a functional unit of the datapath to perform the vector floating-point classification.

14. The method of claim 1 , wherein the third field of the vector floating-point classification instruction specifies the scalar datapath.

15. The method of claim 1 , wherein:

each of the floating-point values includes a respective exponent and a respective fraction; and

the executing of the vector floating-point classification includes determining the type of each of the floating-point values based on the respective exponent and the respective fraction.

16. The method of claim 15 , wherein:

each of the floating-point values further includes a respective sign; and

the executing of the vector floating-point classification includes determining the type of each of the floating-point values further based on the respective sign.

17. The data processor of claim 7 , wherein:

the single vector floating-point classification instruction further specifies a functional unit of the datapath; and

the functional unit of the datapath is configured to classify the floating-point values and store the values indicative of the identified types in response to the single vector floating-point classification instruction.

18. The data processor of claim 7 , wherein:

each of the floating-point values includes a respective exponent and a respective fraction; and

the datapath is configured to determine the type of each of the floating-point value based on the respective exponent and the respective fraction.

19. The data processor of claim 18 , wherein:

each of the floating-point values further includes a respective sign; and

the datapath is configured to determine the type of each of the floating-point value further based on the respective sign.