Systems and methods to store a tile register pair to memory
Embodiments detailed herein relate to systems and methods to store a tile register pair to memory. In one example, a processor includes: decode circuitry to decode a store matrix pair instruction having fields for an opcode and source and destination identifiers to identify source and destination matrices, respectively, each matrix having a PAIR parameter equal to TRUE; and execution circuitry to execute the decoded store matrix pair instruction to store every element of left and right tiles of the identified source matrix to corresponding element positions of left and right tiles of the identified destination matrix, respectively, wherein the executing stores a chunk of C elements of one row of the identified source matrix at a time.
1. A processor comprising:
decode circuitry configured to decode a single store matrix pair instruction specifying an opcode, a first source matrix identifier, and a destination identifier; and
the execution circuitry configured to execute the decoded store matrix pair instruction as per the opcode to:
determine that a pair of matrices are to be stored to memory based on a value, the pair of matrices including a first two-dimensional source matrix and a second two-dimensional source matrix; and
store elements from element positions of the first two-dimensional source matrix to corresponding element positions of a first two-dimensional destination matrix in the memory, and store elements from element positions of the second two-dimensional source matrix to corresponding element positions of a second two-dimensional destination matrix in the memory, in response to the determination that the pair of matrices are to be stored to the memory based on the value.
2. The processor of claim 1 , wherein the opcode defines a size of each element of the pair of source matrices and the destination.
3. The processor of claim 2 , wherein the size of each element of is a doubleword.
4. The processor of claim 2 , wherein the size of each element of is a word.
5. The processor of claim 1 , wherein the pair of source matrices are each a plurality of registers to represent a matrix.
6. The processor of claim 1 , wherein the execution circuitry is to fault upon a determination of at least one of:
a number of configured tiles equals zero,
a pair indicator of the identified source matrix is not set to TRUE, and
a VALID parameter of at least one of the identified source matrix is not set to TRUE.
7. A system comprising:
a processor; and
an accelerator coupled to the processor, the accelerator including:
decode circuitry configured to decode a single store matrix pair instruction specifying an opcode, a first source matrix identifier, and a destination identifier; and
the execution circuitry configured to execute the decoded store matrix pair instruction as per the opcode to:
determine that a pair of matrices are to be stored to memory based on a value, the pair of matrices including a first two-dimensional source matrix and a second two-dimensional source matrix; and
store elements from element positions of the first two-dimensional source matrix to corresponding element positions of a first two-dimensional destination matrix in the memory, and store elements from element positions of the second two-dimensional source matrix to corresponding element positions of a second two-dimensional destination matrix in the memory, in response to the determination that the pair of matrices are to be stored to the memory based on the value.
8. The system of claim 7 , wherein the opcode defines a size of each element.
9. An apparatus comprising:
a plurality of memory controllers;
a processor coupled to the plurality of memory controllers, the processor having a plurality of cores, including a core to perform operations corresponding to an instruction, including to:
determine that a pair of matrices are to be stored to memory based on a value, the pair of matrices including a first two-dimensional source matrix and a second two-dimensional source matrix; and
store elements from element positions of the first two-dimensional source matrix to corresponding element positions of a first two-dimensional destination matrix in the memory, and store elements from element positions of the second two-dimensional source matrix to corresponding element positions of a second two-dimensional destination matrix in the memory, in response to the determination that the pair of matrices are to be stored to the memory based on the value.
10. The apparatus of claim 9 , wherein the core is to store every element of the first two-dimensional source matrix to the first two-dimensional destination matrix, and is to store every element of the second two-dimensional source matrix to the second two-dimensional destination matrix.
11. The apparatus of claim 9 , wherein the first and second two-dimensional destination matrices are next to one another in the memory.
12. The apparatus of claim 9 , wherein the first two-dimensional source matrix and the second two-dimensional source matrix are each stored in a plurality of registers of the processor.
13. The apparatus of claim 9 , wherein the first and second two-dimensional source matrices each have eight rows.
14. The apparatus of claim 9 , wherein the instruction has a field to specify a size of the elements of the first two-dimensional source matrix.
15. The apparatus of claim 9 , wherein the elements of the first two-dimensional source matrix are 16-bit elements.
16. The apparatus of claim 9 , wherein the plurality of cores comprise graphics cores.
17. The apparatus of claim 9 , wherein the plurality of cores are heterogeneous.
18. The apparatus of claim 9 , wherein the plurality of cores are heterogeneous graphics cores.
19. The apparatus of claim 9 , further comprising an instruction converter to convert the instruction into one or more instructions of a different instruction set executable by the core.
20. The apparatus of claim 9 , wherein the core is to store the elements from the element positions of each row of the first two-dimensional source matrix to the corresponding element positions of the first two-dimensional destination matrix, and is to store the elements from the element positions of each row of the second two-dimensional source matrix to the corresponding element positions of the second two-dimensional destination matrix.
21. The apparatus of claim 9 , further comprising a bus interface coupled to the processor.
22. The apparatus of claim 9 , further comprising an interconnect interface coupled to the processor.