INSTRUCTION EXECUTION THAT BROADCASTS AND MASKS DATA VALUES AT DIFFERENT LEVELS OF GRANULARITY

• US 20200134225A1
• Filed: 12/30/2019
• Published: 04/30/2020
• Est. Priority Date: 12/23/2011
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: OULD-AHMED-VALL, ELMOUSTAPHA, VALENTINE, ROBERT, CORBAL, JESUS, TOLL, BRET L., CHARNEY, MARK J.
• Granted Patent: US 11,301,580 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/27 Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
• CPC Class Codes: G06F 21/6227 where protection concerns the structure of data, e.g. records, types, queries
• CPC Class Codes: G06F 21/6254 by anonymising data, e.g. decorrelating personal data from the owner's identification
• CPC Class Codes: G06F 21/70 Protecting specific internal or peripheral components, in which the protection of a component leads to protection of the entire computer
• CPC Class Codes: G06F 9/30018 Bit or string instructions
• CPC Class Codes: G06F 9/30032 Movement instructions, e.g. MOVE, SHIFT, ROTATE, SHUFFLE
• CPC Class Codes: G06F 9/30036 Instructions to perform operations on packed data, e.g. vector, tile or matrix operations
• CPC Class Codes: G06F 9/30038 using a mask
• CPC Class Codes: G06F 9/30101 Special purpose registers
• CPC Class Codes: G06F 9/3802 Instruction prefetching
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

An apparatus is described that includes an execution unit to execute a first instruction and a second instruction. The execution unit includes input register space to store a first data structure to be replicated when executing the first instruction and to store a second data structure to be replicated when executing the second instruction. The first and second data structures are both packed data structures. Data values of the first packed data structure are twice as large as data values of the second packed data structure. The execution unit also includes replication logic circuitry to replicate the first data structure when executing the first instruction to create a first replication data structure, and, to replicate the second data structure when executing the second data instruction to create a second replication data structure. The execution unit also includes masking logic circuitry to mask the first replication data structure at a first granularity and mask the second replication data structure at a second granularity. The second granularity is twice as fine as the first granularity.

Claims

1. A system comprising:

an integrated memory controller unit; and

a processor core coupled with the integrated memory controller unit, the processor core comprising;

multiple levels of cache, including a Level 2 (L2) cache,

a plurality of vector registers,

a plurality of mask registers,

a decode unit circuit to decode a first instruction and a second instruction,

the first instruction having fields to specify a base and an index corresponding to a location in a memory of a first 128-bit packed data structure having two 64-bit elements, having a field to specify a mask register of the plurality of mask registers as a source of a first mask, and having a field to specify a destination register of the plurality of vector registers,

the second instruction having fields to specify a base and an index corresponding to a location in the memory of a second 128-bit packed data structure having four 32-bit elements, having a field to specify a mask register of the plurality of mask registers as a source of a second mask, and having a field to specify a destination register of the plurality of vector registers, and

an execution unit circuit coupled with the decode unit circuit, the plurality of vector registers, and the plurality of mask registers,

the execution unit circuit to perform the first instruction to;

load at least one 64-bit element of the first 128-bit packed data structure,

generate a first masked replication data structure from the first 128-bit packed data structure based on applying the first mask at a 64-bit data element granularity, and with zeroed masking where masked out elements are zeroed, and

store a first result including the first masked replication data structure in the destination register specified by the first instruction, wherein a length of the first masked replication data structure is a multiple of 128-bits and is the same as the destination register specified by the first instruction, and

the execution unit circuit to perform the second instruction to;

load at least one 32-bit element of the second 128-bit packed data structure,

generate a second masked replication data structure from the second 128-bit packed data structure based on applying the second mask at a 32-bit data element granularity, and with the zeroed masking where masked out elements are zeroed, and

store a second result including the second masked replication data structure in the destination register specified by the second instruction, wherein a length of the second masked replication data structure is a multiple of 128-bits and is the same as the destination register specified by the second instruction.

2. The system of claim 1, wherein the processor core also allows using the plurality of mask registers for merged masking in which masked out elements retain initial values they had prior to the merged masking.

3. The system of claim 1, wherein the execution unit circuit when performing the first instruction is not to load a masked out element of the first 128-bit packed data structure.

4. The system of claim 1, wherein the mask registers are in a set of registers with a register that cannot be used as a mask.

5. The system of claim 1, wherein the plurality of mask registers are 64-bit mask registers, and wherein the vector registers are 512-bit vector registers.

6. The system of claim 1, wherein the first result has at least 512-bits and the second result has at least 512-bits.

7. The system of claim 1, wherein the plurality of mask registers are eight mask registers.

8. The system of claim 1, wherein the execution unit circuit includes:

a replication logic circuit to replicate a data structure; and

a masking logic circuit to apply a mask to a data structure.

9. The system of claim 1, wherein the execution unit circuit is included in an out-of-order portion of the processor core.

10. The system of claim 1, wherein the processor core is a reduced instruction set computing (RISC) processor core.

11. The system of claim 1, further comprising a plurality of 64-bit general-purpose registers.

12. The system of claim 1, further comprising a network processor coupled with the processor core.

13. The system of claim 1, further comprising a plurality of coprocessors coupled with the processor core.

14. The system of claim 1, further comprising a ring interconnect unit coupled with the processor core.

15. The system of claim 1, further comprising a high throughput many integrated core (MIC) processor coupled with the processor core.

16. The system of claim 1, further comprising a communication processor coupled with the processor core.

17. The system of claim 1, further comprising an audio processor coupled with the processor core by at least an interconnect.

18. The system of claim 1, further comprising a display unit coupled with the processor core, the display unit to couple to one or more displays.

19. The system of claim 1, further comprising a compression engine coupled with the processor core.

20. A system comprising:

an integrated memory controller unit; and

a processor core coupled with the integrated memory controller unit, the processor core comprising;

multiple levels of cache, including a Level 2 (L2) cache,

a plurality of vector registers,

a plurality of mask registers,

a decode unit circuit to decode an instruction, the instruction having fields to specify a base and an index corresponding to a location in a memory of a 128-bit packed data structure having two 64-bit elements, having a field to specify a mask register of the plurality of mask registers as a source of a mask, and having a field to specify a destination register of the plurality of vector registers, and

an execution unit circuit coupled with the decode unit circuit, the plurality of vector registers, and the plurality of mask registers, the execution unit circuit to perform the instruction to;

load at least one 64-bit element of the 128-bit packed data structure,

generate a masked replication data structure from the 128-bit packed data structure based on applying the mask at a 64-bit data element granularity, and with zeroed masking where masked out elements are zeroed, and

store a result including the masked replication data structure in the destination register, wherein a length of the masked replication data structure is a multiple of 128-bits and is the same as the destination register.

21. The system of claim 20, wherein the processor core also allows using the plurality of mask registers for merged masking in which masked out elements retain initial values they had prior to the merged masking.

22. The system of claim 20, wherein the execution unit circuit when performing the instruction is not to load a masked out element of the 128-bit packed data structure.

23. The system of claim 20, wherein the mask registers are in a set of registers with a register that cannot be used as a mask.

24. The system of claim 20, wherein the plurality of mask registers are 64-bit mask registers, and wherein the vector registers are 512-bit vector registers.

25. The system of claim 20, wherein the execution unit circuit includes:

a replication logic circuit to replicate a data structure; and

a masking logic circuit to apply a mask to a data structure.

26. The system of claim 20, wherein the execution unit circuit is included in an out-of-order portion of the processor core, and wherein the processor core is a reduced instruction set computing (RISC) processor core.

27. The system of claim 20, further comprising a network processor coupled with the processor core.

28. The system of claim 20, further comprising a plurality of coprocessors coupled with the processor core.

29. The system of claim 20, further comprising a general purpose graphics processing unit (GPGPU) coupled with the processor core.

30. The system of claim 20, further comprising an image processor coupled with the processor core by at least an interconnect.

31. A method comprising:

accessing a memory with an integrated memory controller unit;

decoding a first instruction having fields specifying a base and an index corresponding to a location in a memory of a first 128-bit packed data structure having two 64-bit elements, having a field specifying a mask register of a plurality of mask registers as a source of a first mask, and having a field specifying a destination register of a plurality of vector registers;

performing the first instruction, including;

loading at least one 64-bit element of the first 128-bit packed data structure,

generating a first masked replication data structure from the first 128-bit packed data structure based on applying the first mask at a 64-bit data element granularity, and with zeroed masking where masked out elements are zeroed, and

storing a first result including the first masked replication data structure in the destination register specified by the first instruction, wherein a length of the first masked replication data structure is a multiple of 128-bits and is the same as the destination register specified by the first instruction;

decoding a second instruction having fields specifying a base and an index corresponding to a location in the memory of a second 128-bit packed data structure having four 32-bit elements, having a field specifying a mask register of the plurality of mask registers as a source of a second mask, and having a field specifying a destination register of the plurality of vector registers; and

performing the second instruction, including;

loading at least one 32-bit element of the second 128-bit packed data structure,

generating a second masked replication data structure from the second 128-bit packed data structure based on applying the second mask at a 32-bit data element granularity, and with the zeroed masking where masked out elements are zeroed, and

storing a second result including the second masked replication data structure in the destination register specified by the second instruction, wherein a length of the second masked replication data structure is a multiple of 128-bits and is the same as the destination register specified by the second instruction.