Hardware apparatuses and methods relating to elemental register accesses

Patent metadata

Field	Value
Publication number	US-2016188334-A1
Application number	US-201414582784-A
Country	US
Kind code	A1
Filing date	Dec 24, 2014
Priority date	Dec 24, 2014
Publication date	Jun 30, 2016
Grant date	—

Abstract

Methods and apparatuses relating to a vector instruction with a register operand with an elemental offset are described. In one embodiment, a hardware processor includes a decode unit to decode a vector instruction with a register operand with an elemental offset to access a first number of elements in a register specified by the register operand, wherein the first number is a total number of elements in the register minus the elemental offset, access a second number of elements in a next logical register, wherein the second number is the elemental offset, and combine the first number of elements and the second number of elements as a data vector, and an execution unit to execute the vector instruction on the data vector.

First claim

What is claimed is:

1 . A hardware processor comprising:
a decode unit to decode a vector instruction with a register operand with an elemental offset to:
access a first number of elements in a register specified by the register operand, wherein the first number is a total number of elements in the register minus the elemental offset;
access a second number of elements in a next register, wherein the second number is the elemental offset; and
combine the first number of elements and the second number of elements as a data vector; and
an execution unit to execute the vector instruction on the data vector.

2 . The hardware processor of claim 1 , wherein the next register is a next logical register from the register specified by the register operand.

3 . The hardware processor of claim 1 , wherein the next register is specified by a second register operand of the vector instruction.

4 . The hardware processor of claim 1 , comprising a circuit to:
concatenate the elements in the register and the elements in the next register to form a concatenated vector; and shift the concatenated vector by the elemental offset to form the data vector.

5 . The hardware processor of claim 4 , wherein the circuit comprises a multiplier to multiply the elemental offset by a size of each element in the register to determine a number of bits to shift the concatenated vector to form the data vector.

6 . The hardware processor of claim 1 , comprising:
a banked register file that includes the elements in the register and the elements in the next register; and bank logic to combine the first number of elements and the second number of elements from the banked register file as the data vector.

7 . The hardware processor of claim 1 , wherein the register, the next register, and the data vector have a same total number of elements and a same size of each element.

8 . The hardware processor of claim 1 , wherein the decode unit is to output the data vector to the execution unit without an output of the data vector back into the decode unit.

9 . A method comprising:
decoding, with a decode unit, a vector instruction with a register operand with an elemental offset to:
access a first number of elements in a register specified by the register operand, wherein the first number is a total number of elements in the register minus the elemental offset;
access a second number of elements in a next register, wherein the second number is the elemental offset; and
combine the first number of elements and the second number of elements as a data vector; and
executing the vector instruction on the data vector with an execution unit.

10 . The method of claim 9 , wherein the next register is a next logical register from the register specified by the register operand.

11 . The method of claim 9 , wherein the next register is specified by a second register operand of the vector instruction.

12 . The method of claim 9 , comprising:
concatenating the elements in the register and the elements in the next register to form a concatenated vector; and shifting the concatenated vector by the elemental offset to form the data vector.

13 . The method of claim 12 , wherein the shifting comprises multiplying the elemental offset by a size of each element in the register to determine a number of bits to shift the concatenated vector to form the data vector.

14 . The method of claim 9 , comprising:
providing a banked register file that includes the elements in the register and the elements in the next register; and combining the first number of elements and the second number of elements from the banked register file as the data vector.

15 . The method of claim 9 , wherein the register, the next register, and the data vector have a same total number of elements and a same size of each element.

16 . The method of claim 9 , further comprising outputting the data vector to the execution unit from the decode unit without outputting the data vector back into the decode unit.

17 . An apparatus comprising:
a set of one or more processors; and a set of one or more data storage devices that stores code, that when executed by the set of processors causes the set of one or more processors to perform the following: decoding, with a decode unit, a vector instruction with a register operand with an elemental offset to:
access a first number of elements in a register specified by the register operand, wherein the first number is a total number of elements in the register minus the elemental offset;
access a second number of elements in a next register, wherein the second number is the elemental offset; and
combine the first number of elements and the second number of elements as a data vector; and
executing the vector instruction on the data vector with an execution unit.

18 . The apparatus of claim 17 , wherein the next register is a next logical register from the register specified by the register operand.

19 . The apparatus of claim 17 , wherein the next register is specified by a second register operand of the vector instruction.

20 . The apparatus of claim 17 , wherein the set of data storage devices further stores code, that when executed by the set of processors causes the set of processors to perform the following:
concatenating the elements in the register and the elements in the next register to form a concatenated vector; and shifting the concatenated vector by the elemental offset to form the data vector.

21 . The apparatus of claim 20 , wherein the set of data storage devices further stores code, that when executed by the set of processors causes the set of processors to perform the following:
wherein the shifting comprises multiplying the elemental offset by a size of each element in the register to determine a number of bits to shift the concatenated vector to form the data vector.

22 . The apparatus of claim 17 , wherein the set of data storage devices further stores code, that when executed by the set of processors causes the set of processors to perform the following:
providing a banked register file that includes the elements in the register and the elements in the next register; and combining the first number of elements and the second number of elements from the banked register file as the data vector.

23 . The apparatus of claim 17 , wherein the register, the next register, and the data vector have a same total number of elements and a same size of each element.

24 . The apparatus of claim 17 , wherein the set of data storage devices further stores code, that when executed by the set of processors causes the set of processors to perform the following:
further comprising outputting the data vector to the execution unit from the decode unit without outputting the data vector back into the decode unit.

Assignees

• Intel Corp

Inventors

• Lee Victor
• Echeruo Ugonna
• Chrysos George
• Mellempudi Naveen

Classifications

• G06F9/30149

  of variable length instructions · CPC title

• G06F9/30036 Primary

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

• G06F9/30134

  Register stacks; shift registers · CPC title

• G06F9/30038

  using a mask · CPC title