Method and apparatus for performing a vector bit shuffle

Patent metadata

Field	Value
Publication number	US-10296489-B2
Application number	US-201414583636-A
Country	US
Kind code	B2
Filing date	Dec 27, 2014
Priority date	Dec 27, 2014
Publication date	May 21, 2019
Grant date	May 21, 2019

Abstract

A processor including a first vector register for storing a plurality of source data elements, a second vector register for storing a plurality of control elements, and a vector bit shuffle logic. Each of the control elements in the first vector register corresponds to a different source data element and includes a plurality of bit fields. Each of the bit fields is associated with a single corresponding bit position in a destination mask register and identifies a single bit from the corresponding source data element to be copied to the single corresponding bit position in the destination mask register. The vector bit shuffle logic is to read the bit fields from the second vector register and, for each bit field, to identify a single bit from a single corresponding source data element and copy it to a single corresponding bit position in the destination mask register.

First claim

What is claimed is:

1. A processor comprising:
a decoder to decode a single vector bit shuffle instruction, the vector bit shuffle instruction comprising a first source operand, a second source operand, and a destination operand;
a first vector register identified by the first source operand to store a plurality of source data elements;
a second vector register identified by the second source operand to store a plurality of control elements, each of the control elements corresponding to a different one of the plurality of source data elements in the first vector register and comprising a plurality of bit fields, each bit field corresponding to a single-bit bit position in a destination mask register identified by the destination operand, each bit field further and identifies exactly one bit from the corresponding source data element to be copied to the corresponding single-bit bit position in the destination mask register; and
vector bit shuffle logic to, in response to the single decoded vector bit shuffle instruction, read the bit fields from the second vector register and, for each bit field, to identify exactly one bit from the corresponding source data element and responsively copy only the identified bit from the corresponding source data element to a single-bit bit position corresponding to the bit field in the destination mask register.

2. The processor as in claim 1 , wherein the vector bit shuffle logic comprises one or more multiplexers to select a set of bits from each of the source data elements in accordance with the bit fields in each of the control elements.

3. The processor as in claim 1 , wherein each of the source data elements comprises a 64-bit data element and wherein each bit field comprises at least 6 bits to identify a single bit from each of the 64-bit data elements.

4. The processor as in claim 3 , wherein each of the bit fields comprises a control byte and wherein at least 6 bits are to be selected from each of the control bytes to identify each bit from each of the 64-bit data elements.

5. The processor as in claim 4 , wherein eight bits from each data element are to be selected using eight of the control bytes.

6. The processor as in claim 5 , wherein the bits from each data element are to be concatenated within the destination mask register.

7. The processor as in claim 6 , wherein the first vector register is to store eight of the 64 bit data elements and wherein the destination mask register is to store eight corresponding 8-bit values selected from the eight 64-bit data elements.

8. The processor as in claim 7 , wherein the bits within the mask register are to be used to perform masking operations for one or more subsequent instructions executed by the processor.

9. The processor as in claim 1 , wherein the vector bit shuffle logic is to operate responsive to a vector bit shuffle instruction decoded by decode logic in the processor and executed by execution logic in the processor.

10. A method comprising:
decoding a single vector bit shuffle instruction, the vector bit shuffle instruction comprising a first source operand, a second source operand, and a destination operand;
storing a plurality of source data elements in a first vector register identifies by the first source operand;
storing a plurality of control elements in a second vector register identified by the second source operand, each of the control elements corresponding to a different one of the plurality of source data elements in the first vector register and comprising a plurality of bit fields, each bit field corresponding to a single-bit bit position in a destination mask register identified by the destination operand, each bit field further identifies exactly one bit from the corresponding source data element to be copied to the corresponding single-bit bit position in the destination mask register; and
executing the single decoded vector bit shuffle instruction by reading the bit fields from the second vector register and, for each bit field, to identify exactly one bit from the corresponding source data element and responsively copying only the identified bit from the corresponding source data element to a single-bit bit position corresponding to the bit field in the destination mask register.

11. The method as in claim 10 , further comprising selecting a set of bits from each of the source data elements with one or more multiplexers in accordance with the bit fields in each of the control elements.

12. The method as in claim 10 , wherein each of the source data elements comprises a 64-bit data element and wherein each bit field comprises at least 6 bits to identify a single bit from each of the 64-bit data elements.

13. The method as in claim 12 , wherein each of the bit fields comprises a control byte and wherein at least 6 bits are to be selected from each of the control bytes to identify each bit from each of the 64-bit data elements.

14. The method as in claim 13 wherein eight bits from each data element are to be selected using eight of the control bytes.

15. The method as in claim 14 , wherein the bits from each data element are to be concatenated within the destination mask register.

16. The method as in claim 15 , wherein the first vector register is to store eight of the 64 bit data elements and wherein the destination mask register is to store eight corresponding 8-bit values selected from the eight 64-bit data elements.

17. The method as in claim 16 , wherein the bits within the mask register are to be used to perform masking operations for one or more subsequent instructions executed by the processor.

18. A system comprising:
a memory to store program code and data;
a cache hierarchy comprising multiple cache levels to cache the program code and data in accordance with a specified cache management policy;
an input device to receive input from a user;
a processor to execute the program code and process the data responsive to the input from the user, the processor comprising:
a decoder to decode a single vector bit shuffle instruction, the vector bit shuffle instruction comprising a first source operand, a second source operand, and a destination operand;
a first vector register identified by the first source operand to store a plurality of source data elements;
a second vector register identified by the second source operand to store a plurality of control elements, each of the control elements corresponding to a different one of the plurality of source data elements in the first vector register and comprising a plurality of bit fields, each bit field corresponding to a single-bit bit position in a destination mask register identified by the destination operand, each bit field further identifies exactly one bit from the corresponding source data element to be copied to the corresponding single-bit bit position in the destination mask register; and
vector bit shuffle logic to execute the single decoded vector bit shuffle instruction to read the bit fields from the second vector register and, for each bit field, to identify exactly one bit from the corresponding source data element and responsively copy only the identified bit from the corresponding source data element to a single-bit bit position corresponding to the bit field in the destination mask register.

19. The system as in claim 18 , wherein the vector bit shuffle logic comprises one or more multiplexers to select a set of bits from each of the source data elements in accordance with the bit fields in each of the control elements.

20. The

Assignees

• Intel Corp

Inventors

• Ould-Ahmed-Vall Elmoustapha
• San Adrian Jesus Corbal
• Valentine Robert
• Charney Mark J
• Sole Guillem
• Espasa Roger

Classifications

• G06F15/8084 Primary

  Special arrangements thereof, e.g. mask or switch · CPC title

• G06F15/8076

  Details on data register access · CPC title

• G06F9/30032

  Movement instructions, e.g. MOVE, SHIFT, ROTATE, SHUFFLE · CPC title

• G06F9/3012

  Organisation of register space, e.g. banked or distributed register file · CPC title

• G06F9/30036

  Instructions to perform operations on packed data, e.g. vector, tile or matrix operations · CPC title