Multiple network interface core apparatus and method
View Patent ↗A network interface controller and network interface control method comprising providing a single integrated circuit as a network interface controller and employing a plurality of network interface cores on the single integrated circuit.
1. A network interface controller apparatus comprising:
a plurality of network interface cores on a single integrated circuit;
a first multiplexer for communications between said plurality of network interface cores and one or more host processors;
a second multiplexer for communications between said plurality of network interface cores and a network router; and
a content addressable memory between the network router and said second multiplexer; and
wherein said content addressable memory comprises slots in which the low order bits of the slot number determine the network interface core associated with such slot.
2. The apparatus of claim 1 additionally comprising a response reordering queue between the one or more host processors and said first multiplexer.
3. The apparatus of claim 1 wherein each of said plurality of network interface cores supports a plurality of parallel programming models.
4. The apparatus of claim 3 wherein each of said plurality of network interface cores supports both Message Passing Interface and Partitioned Global Address Space programming models.
5. The apparatus of claim 4 wherein each of said plurality of network interface cores supports both Message Passing Interface and Partitioned Global Address Space programming models, employing discrete hardware for the two.
6. A network interface control method comprising the steps of providing a single integrated circuit as a network interface controller, employing a plurality of network interface cores on the single integrated circuit, using a first multiplexer for communications between the plurality of network interface cores and one or more host processors, using a second multiplexer for communications between the plurality of network interface cores and a network router, and using a content addressable memory between the network router and the second multiplexer, wherein the content addressable memory comprises slots in which the low order bits of the slot number determine the network interface core associated with such slot.
7. The method of claim 6 additionally comprising using a response reordering queue between the one or more host processors and the first multiplexer.
8. The method of claim 6 wherein each of the plurality of network interface cores supports a plurality of parallel programming models.
9. The method of claim 8 wherein each of the plurality of network interface cores supports both Message Passing Interface and Partitioned Global Address Space programming models.
10. The method of claim 9 wherein each of the plurality of network interface cores supports both Message Passing Interface and Partitioned Global Address Space programming models, employing discrete hardware for the two.