Managing data provided to switches in a parallel processing environment
View Patent ↗An integrated circuit comprises a plurality of tiles. Each tile comprises a processor including a storage module, wherein the processor is configured to process multiple streams of instructions, a switch including switching circuitry to forward data received over data paths from other tiles to the processor and to switches of other tiles, and to forward data received from the processor to switches of other tiles, and coupling circuitry configured to couple data resulting from processing an instruction from at least one of the streams of instructions to the storage module and to the switch.
1. An integrated circuit, comprising:
a plurality of tiles, each tile comprising:
a processor, and
a switch including switching circuitry to forward data received over data paths from other tiles to the processor and to switches of other tiles, and to forward data received from the processor to switches of other tiles,
with the processor configured to determine whether to delay executing a first instruction received by the processor until an operation associated with a previous instruction has provided data to the switch, the previous instruction having been received by the processor, with the processor configured to determine whether to delay executing the first instruction based on whether the first instruction is configured to provide data to the switch, and configured to proceed to execute the first instruction while the operation associated with the previous instruction is loading data to the switch if the first instruction is not configured to provide data to the switch.
2. The integrated circuit of claim 1 , wherein the previous instruction loads data from a location outside the tile to be provided to the switch.
3. The integrated circuit of claim 2 , wherein the location outside the tile comprises an external memory coupled to at least one of the tiles.
4. The integrated circuit of claim 2 , wherein the processor comprises at least one pipeline comprising a plurality of stages.
5. The integrated circuit of claim 4 , wherein the processor is configured to couple data from a stage of the pipeline before a final stage to the switch.
6. The integrated circuit of claim 5 , wherein the final stage comprises a write-back stage.
7. The integrated circuit of claim 2 , wherein the previous instruction loads data from the location outside the tile in response to a cache miss.
8. A method for processing instructions in an integrated circuit, the integrated circuit comprising a plurality of tiles, each tile comprising a processor and a switch, the method comprising:
processing instructions in a processor of a tile; and
forwarding data received over data paths from other tiles to the processor and to switches of other tiles, and forwarding data received from the processor to switches of other tiles;
with the processor determining whether to delay executing a first instruction received by the processor until an operation associated with a previous instruction has provided data to the switch, the previous instruction having been received by the processor, with the processor determining whether to delay executing the first instruction based on whether the first instruction is configured to provide data to the switch, and executing the first instruction while the operation associated with the previous instruction is loading data to the switch if the first instruction is not configured to provide data to the switch.
9. The method of claim 8 , wherein the previous instruction loads data from a location outside the tile to be provided to the switch.
10. The method of claim 9 , wherein the location outside the tile comprises an external memory coupled to at least one of the tiles.
11. The method of claim 9 , wherein the processor comprises at least one pipeline comprising a plurality of stages.
12. The method of claim 11 , wherein the processor couples data from a stage of the pipeline before a final stage to the switch.
13. The method of claim 12 , wherein the final stage comprises a write-back stage.
14. The method of claim 9 , wherein the previous instruction loads data from the location outside the tile in response to a cache miss.