Sliced crossbar architecture with no inter-slice communication
View Patent ↗A memory crossbar adapted to transmit a memory transaction to a network resource. The memory crossbar includes a plurality of input ports. Each of the plurality of input ports includes a TAGS generator and a queue adapted to receive a first portion of a first memory transaction and a first portion of a second memory transaction. Each of the plurality of inputs ports also includes an arbiter coupled to the queue and adapted to independently select either the first portion of the first memory transaction or the first portion of the second memory transaction based on a priority associated with the first portion of the first memory transaction and the first portion of the second memory transaction and a multiplexer coupled to the arbiter.
1. A memory crossbar adapted to transmit a memory transaction to a network resource, the memory crossbar comprising a plurality of input ports and a plurality of output ports, the input ports being coupled to the output ports via an internal bus, each of the plurality of input ports comprising:
a queue adapted to receive a first memory transaction and a second memory transaction, wherein the queue comprises a plurality of request stations including a first request station for storing the first memory transaction and a second request station for storing the second memory transaction, the first request station identifying the first memory transaction as being received in the queue before the second memory transaction;
an arbiter coupled to the queue and adapted to select the second request station when the second memory transaction is identified as having a higher priority than the first memory transaction; and
a multiplexer coupled to the arbiter, wherein the second memory transaction is transmitted to one of the output ports via the internal bus and the multiplexer,
wherein the first memory transaction is moved to the second request station and a new memory transaction received in the queue is stored in the first request station.
2. The memory crossbar of claim 1 wherein the network resource comprises a memory resource.
3. The memory crossbar of claim 1 further comprising:
a segment interface coupled to the plurality of input ports; and
a memory controller coupled to the plurality of output ports.
4. The memory crossbar of claim 1 wherein each of the plurality of request stations comprises a forward register, a reverse register, and a delay buffer.
5. The memory crossbar of claim 1 wherein the queue further comprises a queue controller.
6. The memory crossbar of claim 5 wherein the queue controller comprises a forward controller and a reverse controller for each of the plurality of request stations.
7. The memory crossbar of claim 1 wherein the network resource comprises a processor.
8. The memory crossbar of claim 1 wherein the priority associated with the first memory transaction is determined by a position of the first memory transaction in the queue.
9. The memory crossbar of claim 1 wherein the priority associated with the first memory transaction is determined by an age of the first memory transaction.
10. The memory crossbar of claim 1 wherein the priority associated with the second memory transaction is determined by a position of the second memory transaction in the queue.
11. The memory crossbar of claim 1 wherein the priority associated with the second memory transaction is determined by an age of the second memory transaction.
12. A method of transmitting a memory transaction to a network resource coupled to a memory crossbar, the method comprising:
receiving a first memory transaction at an input port of the memory crossbar;
storing the first memory transaction in a first request station in a queue, wherein the queue comprises a plurality of request stations;
receiving a second memory transaction at the input port of the memory crossbar;
storing the second memory transaction in a second request station in the queue, wherein the second request station identifies the second memory transaction as being received in the queue after the first memory transaction;
selecting the second request station when the second memory transaction is identified as having a higher priority than the first memory transaction;
transmitting the second memory transaction to an output port;
moving the first memory transaction to the second request station;
receiving a new memory transaction in the queue; and
storing the new memory transaction in the first request station.
13. The method of claim 12 wherein the network resource comprises a memory resource.
14. The method of claim 12 wherein the network resource comprises a processor.
15. The method of claim 12 wherein each of the plurality of request stations comprises a forward register, a reverse register, and a delay buffer.
16. The method of claim 12 wherein the queue further comprises a queue controller.
17. The method of claim 16 wherein the queue controller comprises a forward controller and a reverse controller for each of the plurality of request stations.
18. The method of claim 12 further comprising determining the priority associated with the first memory transaction by a position of the first memory transaction in the queue.
19. The method of claim 12 further comprising determining the priority associated with the first memory transaction by an age of the first memory transaction.
20. The method of claim 12 further comprising determining the priority associated with the second memory transaction by a position of the second memory transaction in the queue.
21. The method of claim 12 further comprising determining the priority associated with the second memory transaction by an age of the second memory transaction.
22. The method of claim 12 wherein independently the selecting is performed by an arbiter coupled to the queue.