Systems, methods, and apparatuses for tile broadcast
Embodiments detailed herein relate to matrix operations. In particular, embodiment of broadcasting elements are described. For example, some embodiments describe broadcasting a scalar to all configured data element positons of a destination matrix (tile). For example, some embodiments describe broadcasting a row to all configured data element positons of a destination matrix (tile). For example, some embodiments describe broadcasting a column to all configured data element positons of a destination matrix (tile).
1. A processor comprising:
decode circuitry to decode an instruction having fields for an opcode, an identifier for a source operand, and an identifier for a destination matrix operand; and
execution circuitry to execute the decoded instruction according to the opcode to broadcast a single data element from the identified source operand into each configured data element of the identified destination matrix operand.
2. The processor of claim 1 , wherein the execution circuitry comprises a crossbar switch.
3. The processor of claim 1 , wherein the source operand is stored in memory.
4. The processor of claim 1 , wherein the source operand is stored in a packed data register.
5. The processor of claim 1 , wherein the source operand is stored in a general-purpose data register.
6. The processor of claim 1 , wherein the identified destination matrix operand is a plurality of registers to represent a two-dimensional matrix.
7. The processor of claim 1 , wherein the data elements of the row of data are doubleword in size.
8. The processor of claim 1 , wherein the data elements of the row of data are word in size.
9. The processor of claim 1 , further comprising storage to store a configuration of the identified destination matrix operand.
10. The processor of claim 1 , wherein the configuration is to indicate a number of rows and columns to be used by the identified destination matrix operand.
11. The processor of claim 1 , wherein the execution circuitry is to further zero unconfigured data elements of the identified destination matrix operand.
12. A method comprising:
decoding an instruction having fields for an opcode, an identifier for a source operand, and an identifier for a destination matrix operand; and
executing the decoded instruction according to the opcode to broadcast a single data element from the identified source operand into each configured data element of the identified destination matrix operand.
13. The method of claim 12 , wherein the source operand is stored in memory.
14. The method of claim 12 , wherein the source operand is stored in a register.
15. The method of claim 12 , wherein the identified destination matrix operand is a plurality of registers to represent a two-dimensional matrix.
16. The method of claim 12 , wherein the data elements of the row of data are doubleword in size.
17. The method of claim 12 , wherein the data elements of the row of data are word in size.
18. The method of claim 12 , further comprising:
zeroing unconfigured data elements of the identified destination matrix operand.
19. A non-transitory machine-readable medium storing an instruction which causes a processor to perform a method, the method comprising:
decoding the instruction having fields for an opcode, an identifier for a source operand, and an identifier for a destination matrix operand; and
executing the decoded instruction according to the opcode to broadcast a single data element from the identified source operand into each configured data element of the identified destination matrix operand.
20. A system comprising:
a processor;
an accelerator coupled to the processor, the accelerator including:
decode circuitry to decode an instruction having fields for an opcode, an identifier for a source operand, and an identifier for a destination matrix operand; and
execution circuitry to execute the decoded instruction according to the opcode to broadcast a single data element from the identified source operand into each configured data element of the identified destination matrix operand.