Vector floating-point classification
A method to classify source data in a processor in response to a vector floating-point classification instruction includes specifying, in respective fields of the vector floating-point classification instruction, a source register containing the source data and a destination register to store classification indications for the source data. The source register includes a plurality of lanes that each contains a floating-point value and the destination register includes a plurality of lanes corresponding to the lanes of the source register. The method further includes executing the vector floating-point classification instruction by, 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.
1. A processor comprising:
a set of functional units; and
a set of registers coupled to the set of functional units, wherein:
each register of the set of functional units includes a set of lanes; and
the processor is configured to:
receive an instruction that specifies a source register of the set of registers and a destination register of the set of registers; and
based on the instruction, cause a functional unit of the set of functional units to, for each lane of the set of lanes of the source register:
determine whether a respective value stored in the respective lane of the source register matches a type within a set of types; and
store an indicator of whether the respective value matches in a respective lane of the set of lanes of the destination register.
2. The processor of claim 1 , wherein the set of types includes at least one of: a zero value, a subnormal value, a normal value, an infinite value, a not-a-number value, a quiet not-a-number value, and a signaling not-a-number value.
3. The processor of claim 1 , wherein:
the set of functional units includes a plurality of functional units each configured to, for each lane of the set of lanes of the source register:
determine whether the respective value stored in the respective lane of the source register matches a type within the set of types; and
store the indicator of whether the respective value matches in the respective lane of the set of lanes of the destination register; and
the instruction includes a field that specifies the functional unit from among the plurality of functional units.
4. The processor of claim 1 , wherein:
the set of functional units are arranged in a scalar datapath and a vector datapath; and
the instruction includes a field that specifies whether to execute the instruction using the scalar datapath or the vector datapath.
5. The processor of claim 1 , wherein, for each lane of the set of lanes of the source register, the respective value is a floating point value.
6. The processor of claim 1 , wherein the instruction includes a field that specifies whether the respective value stored in each lane of the set of lanes of the source register is a single precision floating point value or a double precision floating point value.
7. The processor of claim 1 , wherein, for each lane of the set of lanes of the source register, the respective value is a single precision floating point value.
8. The processor of claim 1 , wherein, for each lane of the set of lanes of the source register, the respective value is a double precision floating point value.
9. The processor of claim 1 , wherein the instruction includes a field that specifies whether the instruction is permitted to be performed in parallel with another instruction.
10. The processor of claim 1 , wherein the source register has a width of 512 bits.
11. A method comprising:
receiving an instruction that specifies a source register and a destination register, wherein each of the source register and the destination register have a set of lanes; and
based on the instruction, for each lane of the set of lanes of the source register:
determining whether a respective value stored in the respective lane matches a type within a set of types; and
storing an indicator of whether the respective value matches in a respective lane of the set of lanes of the destination register.
12. The method of claim 11 , wherein the set of types includes at least one of: a zero value, a subnormal value, a normal value, an infinite value, a not-a-number value, a quiet not-a-number value, and a signaling not-a-number value.
13. The method of claim 11 , wherein the instruction includes a field that specifies a functional unit from among a set of functional units of a processor to perform the determining of whether the respective value matches a type and the storing of the indicator.
14. The method of claim 11 , wherein the instruction includes a field that specifies a datapath from among a scalar datapath and a vector datapath of a processor to perform the determining of whether the respective value matches a type and the storing of the indicator.
15. The method of claim 11 , wherein, for each lane of the set of lanes of the source register, the respective value is a floating point value.
16. The method of claim 11 , wherein the instruction includes a field that specifies whether the respective value stored in each lane of the set of lanes of the source register is a single precision floating point value or a double precision floating point value.
17. The method of claim 11 , wherein, for each lane of the set of lanes of the source register, the respective value is a single precision floating point value.
18. The method of claim 11 , wherein, for each lane of the set of lanes of the source register, the respective value is a double precision floating point value.
19. The method of claim 11 , wherein the instruction includes a field that specifies whether the instruction is permitted to be performed in parallel with another instruction.
20. The method of claim 11 , wherein the source register has a width of 512 bits.