Systems, methods, and apparatuses for tile store
Embodiments detailed herein relate to matrix operations. In particular, the loading of a matrix (tile) from memory. For example, support for a loading instruction is described in at least a form of decode circuitry to decode an instruction having fields for an opcode, a source matrix operand identifier, and destination memory information, and execution circuitry to execute the decoded instruction to store each data element of configured rows of the identified source matrix operand to memory based on the destination memory information.
1 . A processor comprising:
a register to store a configuration value for tiles;
decode circuitry to decode an instruction, the instruction having a field to identify a plurality of non-consecutive vector registers, a field to identify a register to store a base, and a field to identify a register to store an index, wherein the plurality of non-consecutive vector registers are to store 64-bit data elements, and wherein the plurality of non-consecutive vector registers are to have a single-instruction, multiple-data (SIMD) dimension based on the configuration value for the tiles; and
execution circuitry coupled with the decode circuitry, the execution circuitry to perform operations corresponding to the instruction, including to store the 64-bit data elements from the plurality of non-consecutive vector registers to destination memory locations generated using the base and the index.
2 . The processor of claim 1 , wherein the plurality of non-consecutive vector registers is a configured number of non-consecutive vector registers.
3 . The processor of claim 1 , wherein the plurality of non-consecutive vector registers is four non-consecutive vector registers.
4 . The processor of claim 1 , further comprising a register to indicate whether the tiles are configured for use.
5 . The processor of claim 1 , wherein the decode circuitry is to decode a second instruction, and further comprising execution circuitry to perform operations corresponding to the second instruction, including to configure the tiles for use.
6 . The processor of claim 1 , wherein the instruction is to indicate a stride.
7 . The processor of claim 6 , wherein the plurality of non-consecutive vector registers correspond to different rows delineated based on the stride.
8 . The processor of claim 1 , wherein the processor has a reduced instruction set computing (RISC) architecture.
9 . The processor of claim 1 , wherein the processor is a central processing unit (CPU), and wherein the CPU further comprises a reorder buffer and register renaming circuitry.
10 . The processor of claim 1 , further comprising a register to indicate whether the tiles are configured for use, wherein the processor is a central processing unit (CPU), wherein the CPU further comprises a reorder buffer and register renaming circuitry, and wherein the plurality of non-consecutive vector registers is four non-consecutive vector registers.
11 . The processor of claim 10 , wherein the plurality of non-consecutive vector registers is a configured number of non-consecutive vector registers, and wherein the instruction is to indicate a stride.
12 . The processor of claim 11 , wherein the decode circuitry is to decode a second instruction, and further comprising execution circuitry to perform operations corresponding to the second instruction, including to configure the tiles for use, and wherein the plurality of non-consecutive vector registers correspond to different rows delineated based on the stride.
13 . The processor of claim 10 , wherein the decode circuitry is to decode a second instruction, and further comprising execution circuitry to perform operations corresponding to the second instruction, including to configure the tiles for use, and wherein the plurality of non-consecutive vector registers correspond to different rows delineated based on a stride indicated by the instruction.
14 . A method comprising:
storing a configuration value for tiles in a register;
decoding an instruction, the instruction having a field identifying a plurality of non-consecutive vector registers, a field identifying a register to store a base, and a field identifying a register to store an index, wherein the plurality of non-consecutive vector registers store 64-bit data elements, and wherein the plurality of non-consecutive vector registers have a single-instruction, multiple-data (SIMD) dimension based on the configuration value for the tiles; and
performing operations corresponding to the instruction, including storing the 64-bit data elements from the plurality of non-consecutive vector registers to destination memory locations generated using the base and the index.
15 . The method of claim 14 , wherein storing comprises storing the 64-bit data elements from four non-consecutive vector registers to the destination memory locations generated using the base and the index, and further comprising accessing a register to determine whether the tiles are configured for use.
16 . The method of claim 14 , further comprising:
configuring a number of the plurality of non-consecutive vector registers;
decoding a second instruction; and
performing operations corresponding to the second instruction, including configuring the tiles for use.
17 . The method of claim 14 , further comprising determining a stride from the instruction.
18 . A system on a chip (SoC) comprising:
a memory controller; and
a processor coupled with the memory controller, the processor comprising:
a register to store a configuration value for tiles;
decode circuitry to decode an instruction, the instruction having a field to identify a plurality of non-consecutive vector registers, a field to identify a register to store a base, and a field to identify a register to store an index, wherein the plurality of non-consecutive vector registers are to store 64-bit data elements, and wherein the plurality of non-consecutive vector registers are to have a single-instruction, multiple-data (SIMD) dimension based on the configuration value for the tiles; and
execution circuitry coupled with the decode circuitry, the execution circuitry to perform operations corresponding to the instruction, including to store the 64-bit data elements from the plurality of non-consecutive vector registers to destination memory locations generated using the base and the index.
19 . The SoC of claim 18 , wherein the plurality of non-consecutive vector registers is a configured number of non-consecutive vector registers, and further comprising a register to indicate whether the tiles are configured for use.
20 . The SoC of claim 18 , wherein the plurality of non-consecutive vector registers is four non-consecutive vector registers, and wherein the decode circuitry is to decode a second instruction, and further comprising execution circuitry to perform operations corresponding to the second instruction, including to configure the tiles for use.
21 . The SoC of claim 18 , wherein the instruction is to indicate a stride, and wherein the plurality of non-consecutive vector registers correspond to different rows delineated based on the stride.
22 . The SoC of claim 18 , further comprising a register to indicate whether the tiles are configured for use, wherein the processor is a central processing unit (CPU), wherein the CPU further comprises a reorder buffer and register renaming circuitry, and wherein the plurality of non-consecutive vector registers is four non-consecutive vector registers.
23 . A non-transitory machine-readable storage medium storing instructions that, when executed by a machine, are to cause the machine to perform operations, including to:
store a configuration value for tiles in a register;
decode an instruction, the instruction having a field to identify a plurality of non-consecutive vector registers, a field to identify a register to store a base, and a field to identify a register to store an index, wherein the plurality of non-consecutive vector registers are to store 64-bit data elements, and wherein the plurality of non-consecutive vector registers are to have a single-instruction, multiple-data (SIMD) dimension based on the configuration value for the tiles; and
perform operations corresponding to the instruction, including to store the 64-bit data elements from the plurality of non-consecutive vector registers to destination memory locations generated using the base and the index.
24 . The non-transitory machine-readable storage medium of claim 23 , wherein the plurality of non-consecutive vector registers is four non-consecutive vector registers, and wherein the operations further include to access a register to determine whether the tiles are configured for use.
25 . The non-transitory machine-readable storage medium of claim 23 , wherein the operations further include to:
configuring a number of the plurality of non-consecutive vector registers;
decoding a second instruction; and
performing operations corresponding to the second instruction, including configuring the tiles for use.
26 . An apparatus comprising:
a plurality of caches at a plurality of different cache levels;
programmable configuration storage to store configuration information for a matrix, the configuration information to include a first value corresponding to a number of rows for the matrix, a second value corresponding to a number of columns for the matrix, and a start row value corresponding to a row of the matrix at which to restart execution of a matrix store instruction;
a fetch circuitry to fetch the matrix store instruction;
decode circuitry to decode the matrix store instruction, the matrix store instruction having a first operand to reference one or more memory addresses and a second operand to specify one or more registers; and
execution circuitry, coupled to the programmable configuration storage and the decode circuitry, the execution circuitry to perform operations corresponding to the matrix store instruction, including to store the matrix, from the one or more registers specified by the second operand, to the one or more memory addresses referenced by the first operand.
27 . The apparatus of claim 26 , wherein the programmable configuration storage comprises one or more physical registers, and further comprising a tile register including the one or more registers.
28 . The apparatus of claim 27 , wherein the tile register is one of a plurality of tile registers, each of the plurality of tile registers to store one of a plurality of matrices, and the configuration information is to include a plurality of pairs of values, each of the plurality of pairs of values corresponding to a number of rows and a number of columns for one of the plurality of matrices.
29 . The apparatus of claim 28 , wherein the decode circuitry is also to decode a configuration instruction to load the configuration information from one or more memory locations referenced by one or more operands of the configuration instruction, and wherein the configuration information also includes a palette identifier to identify a palette including the plurality of matrices.