STATISTICS AND FAILURE DETECTION IN A NETWORK ON A CHIP (NoC) NETWORK

• US 20140043962A1
• Filed: 08/07/2012
• Published: 02/13/2014
• Est. Priority Date: 08/07/2012
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Rangan, Venkat, Levin, Jeffrey A., Beraha, Rodolfo G.
• Granted Patent: US 9,647,921 B2
• Time in Patent Office: Days
• US Class Current: 370/228
• CPC Class Codes: G06F 15/7825 Globally asynchronous, locally synchronous, e.g. network on chip
• CPC Class Codes: H04L 41/0659 by isolating or reconfiguring faulty entities
• CPC Class Codes: H04L 41/0661 by reconfiguring faulty entities
• CPC Class Codes: H04L 41/0668 by dynamic selection of recovery network elements, e.g. replacement by the most appropriate element after failure
• CPC Class Codes: H04L 43/04 Processing captured monitoring data, e.g. for logfile generation
• CPC Class Codes: H04L 43/0852 Delays
• CPC Class Codes: H04L 43/10 Active monitoring, e.g. heartbeat, ping or trace-route
• CPC Class Codes: H04L 43/106 using time related information in packets, e.g. by adding timestamps
• CPC Class Codes: H04L 43/50 Testing arrangements

Abstract

Certain aspects of the present disclosure support techniques for collecting system information in a network on a chip (NoC). A dedicated packet may be transmitted from a source node to a destination node. As it traverses through the NoC, the dedicated packet may collect information from various nodes, which may be made available by the destination node. The collected information may be used in an effort to detect failures and collect statistics regarding the NoC.

Claims

1. A method for collecting system information in a network on a chip (NoC) comprising a plurality of interconnected processing nodes, comprising:

transmitting a dedicated packet from a source node to a destination node;

collecting information from various nodes in the NoC as the dedicated packet traverses a route from the source node to the destination node; and

making the collected information available by the destination node.

2. The method of claim 1, wherein the destination node makes the collected information available via polling by other nodes in the NoC.

3. The method of claim 1, wherein the destination node sends the collected information to a central entity for processing.

4. The method of claim 1, wherein the various nodes add their own information to the dedicated packet via at least one of:

pre-pending their own information to a payload of the dedicated packet, appending their own information to the payload of the dedicated packet, or modifying the payload of the dedicated packet.

5. The method of claim 1, wherein the collected information comprises queue statistics for at least one of the various nodes.

6. The method of claim 1, wherein collected information comprises timestamps added by at least one of the source node or one of the various nodes.

7. The method of claim 6, further comprising:

processing information collected in the dedicated packet to determine a traversal time of the dedicated packet from the source node to the destination node.

8. The method of claim 6, further comprising:

determining one or more path delays between nodes based on the timestamps.

9. The method of claim 1, wherein the destination node transmits the dedicated packet back to the source node.

10. The method of claim 9, further comprising:

determining a round-trip time of the dedicated packet from the source node to the destination node and back.

11. The method of claim 1, further comprising:

detecting a failure occurring in the route traversed by the dedicated packet; and

in response, actively probing one or more other routes using dedicated packets.

12. The method of claim 11, further comprising:

re-configuring the route, in an effort to account for the failure, based on information obtained via the probing.

13. The method of claim 1, further comprising:

demarcating one or more traffic flows in the NoC.

14. The method of claim 13, wherein the demarcating comprises, for a multiple input, single output node of the NoC:

blocking incoming traffic on one or more incoming ports of the node which have received dedicated packets; and

transmitting a single dedicated packet on an output port of the node, in an effort to indicate to a receiver node that the incoming ports of the node are synchronized.

15. The method of claim 13, wherein the demarcating comprises, for a single input, multiple output node of the NoC:

receiving the dedicated packet on an incoming port of the node;

replicating the dedicated packet for each output port of the node; and

transmitting the replicated dedicated packet on each of the output ports of the node.

16. The method of claim 13, wherein the demarcating comprises, for a multiple input, multiple output node of the NoC:

receiving dedicated packets on each incoming port of the node;

replicating the received dedicated packets for each output port of the node; and

transmitting the replicated dedicated packets on each of the output ports of the node.

17. An apparatus for collecting system information in a network on a chip (NoC) comprising a plurality of interconnected processing nodes, comprising:

means for transmitting a dedicated packet from a source node to a destination node;

means for collecting information from various nodes in the NoC as the dedicated packet traverses a route from the source node to the destination node; and

means for making the collected information available by the destination node.

18. The apparatus of claim 17, wherein the destination node makes the collected information available via polling by other nodes in the NoC.

19. The apparatus of claim 17, wherein the destination node sends the collected information to a central entity for processing.

20. The apparatus of claim 17, wherein the various nodes add their own information to the dedicated packet via at least one of:

pre-pending their own information to a payload of the dedicated packet, appending their own information to the payload of the dedicated packet, or modifying the payload of the dedicated packet.

21. The apparatus of claim 17, wherein the collected information comprises queue statistics for at least one of the various nodes.

22. The apparatus of claim 17, wherein collected information comprises timestamps added by at least one of the source node or one of the various nodes.

23. The apparatus of claim 22, further comprising:

means for processing information collected in the dedicated packet to determine a traversal time of the dedicated packet from the source node to the destination node.

24. The apparatus of claim 22, further comprising:

means for determining one or more path delays between nodes based on the timestamps.

25. The apparatus of claim 17, wherein the destination node transmits the dedicated packet back to the source node.

26. The apparatus of claim 25, further comprising:

means for determining a round-trip time of the dedicated packet from the source node to the destination node and back.

27. The apparatus of claim 17, further comprising:

means for detecting a failure occurring in the route traversed by the dedicated packet; and

in response, means for actively probing one or more other routes using dedicated packets.

28. The apparatus of claim 27, further comprising:

means for re-configuring the route, in an effort to account for the failure, based on information obtained via the probing.

29. The apparatus of claim 17, further comprising:

means for demarcating one or more traffic flows in the NoC.

30. The apparatus of claim 29, wherein the means for demarcating comprises, for a multiple input, single output node of the NoC:

means for blocking incoming traffic on one or more incoming ports of the node which have received dedicated packets; and

means for transmitting a single dedicated packet on an output port of the node, in an effort to indicate to a receiver node that the incoming ports of the node are synchronized.

31. The apparatus of claim 29, wherein the means for demarcating comprises, for a single input, multiple output node of the NoC:

means for receiving the dedicated packet on an incoming port of the node;

means for replicating the dedicated packet for each output port of the node; and

means for transmitting the replicated dedicated packet on each of the output ports of the node.

32. The apparatus of claim 29, wherein the means for demarcating comprises, for a multiple input, multiple output node of the NoC:

means for receiving dedicated packets on each incoming port of the node;

means for replicating the received dedicated packets for each output port of the node; and

means for transmitting the replicated dedicated packets on each of the output ports of the node.

33. An apparatus for collecting system information in a network on a chip (NoC) comprising a plurality of interconnected processing nodes, comprising:

a processing unit configured to;

transmit a dedicated packet from a source node to a destination node;

collect information from various nodes in the NoC as the dedicated packet traverses a route from the source node to the destination node; and

make the collected information available by the destination node.

34. The apparatus of claim 33, wherein the destination node makes the collected information available via polling by other nodes in the NoC.

35. The apparatus of claim 33, wherein the destination node sends the collected information to a central entity for processing.

36. The apparatus of claim 33, wherein the various nodes add their own information to the dedicated packet via at least one of:

pre-pending their own information to a payload of the dedicated packet, appending their own information to the payload of the dedicated packet, or modifying the payload of the dedicated packet.

37. The apparatus of claim 33, wherein the collected information comprises queue statistics for at least one of the various nodes.

38. The apparatus of claim 33, wherein collected information comprises timestamps added by at least one of the source node or one of the various nodes.

39. The apparatus of claim 38, wherein the processing unit it further configured to:

process information collected in the dedicated packet to determine a traversal time of the dedicated packet from the source node to the destination node.

40. The apparatus of claim 38, wherein the processing unit is further configured to:

determine one or more path delays between nodes based on the timestamps.

41. The apparatus of claim 33, wherein the destination node transmits the dedicated packet back to the source node.

42. The apparatus of claim 41, wherein the processing unit is further configured to:

determine a round-trip time of the dedicated packet from the source node to the destination node and back.

43. The apparatus of claim 33, wherein the processing unit is further configured to:

detect a failure occurring in the route traversed by the dedicated packet; and

in response, actively probe one or more other routes using dedicated packets.

44. The apparatus of claim 43, wherein the processing unit is further configured to:

re-configure the route, in an effort to account for the failure, based on information obtained via the probing.

45. The apparatus of claim 33, wherein the processing unit is further configured to:

demarcate one or more traffic flows in the NoC.

46. The apparatus of claim 45, wherein the processing unit is configured to demarcate one or more traffic flows in the NoC, for a multiple input, single output node of the NoC, by:

blocking incoming traffic on one or more incoming ports of the node which have received dedicated packets; and

transmitting a single dedicated packet on an output port of the node, in an effort to indicate to a receiver node that the incoming ports of the node are synchronized.

47. The apparatus of claim 45, wherein the processing unit is configured to demarcate one or more traffic flows in the NoC, for a single input, multiple output node of the NoC, by:

receiving the dedicated packet on an incoming port of the node;

replicating the dedicated packet for each output port of the node; and

transmitting the replicated dedicated packet on each of the output ports of the node.

48. The apparatus of claim 45, wherein the processing unit is configured to demarcate one or more traffic flows in the NoC, for a multiple input, multiple output node of the NoC, by:

receiving dedicated packets on each incoming port of the node;

replicating the received dedicated packets for each output port of the node; and

transmitting the replicated dedicated packets on each of the output ports of the node.

49. A computer-program product for collecting system information in a network on a chip (NoC) comprising a plurality of interconnected processing nodes, the computer-program product comprising a non-transitory computer-readable medium having instructions execrable to:

transmit a dedicated packet from a source node to a destination node;

collect information from various nodes in the NoC as the dedicated packet traverses a route from the source node to the destination node; and

make the collected information available by the destination node.

50. The computer-program product of claim 49, wherein the destination node makes the collected information available via polling by other nodes in the NoC.

51. The computer-program product of claim 49, wherein the destination node sends the collected information to a central entity for processing.

52. The computer-program product of claim 49, wherein the various nodes add their own information to the dedicated packet via at least one of:

pre-pending their own information to a payload of the dedicated packet, appending their own information to the payload of the dedicated packet, or modifying the payload of the dedicated packet.

53. The computer-program product of claim 49, wherein the collected information comprises queue statistics for at least one of the various nodes.

54. The computer-program product of claim 49, wherein collected information comprises timestamps added by at least one of the source node or one of the various nodes.

55. The computer-program product of claim 54, further comprising instructions executable to:

process information collected in the dedicated packet to determine a traversal time of the dedicated packet from the source node to the destination node.

56. The computer-program product of claim 54, further comprising instructions executable to:

determine one or more path delays between nodes based on the timestamps.

57. The computer-program product of claim 49, wherein the destination node transmits the dedicated packet back to the source node.

58. The computer-program product of claim 57, further comprising instructions executable to:

determine a round-trip time of the dedicated packet from the source node to the destination node and back.

59. The computer-program product of claim 49, further comprising instructions executable to:

detect a failure occurring in the route traversed by the dedicated packet; and

in response, actively probe one or more other routes using dedicated packets.

60. The computer-program product of claim 59, further comprising instructions executable to:

re-configure the route, in an effort to account for the failure, based on information obtained via the probing.

61. The computer-program product of claim 49, further comprising instructions executable to:

demarcate one or more traffic flows in the NoC.

62. The computer-program product of claim 61, wherein the instructions executable to demarcate one or more traffic flows in the NoC, for a multiple input, single output node of the NoC, comprising:

blocking incoming traffic on one or more incoming ports of the node which have received dedicated packets; and

transmitting a single dedicated packet on an output port of the node, in an effort to indicate to a receiver node that the incoming ports of the node are synchronized.

63. The computer-program product of claim 61, wherein the instructions executable to demarcate one or more traffic flows in the NoC, for a single input, multiple output node of the NoC, comprises:

receiving the dedicated packet on an incoming port of the node;

replicating the dedicated packet for each output port of the node; and

transmitting the replicated dedicated packet on each of the output ports of the node.

64. The computer-program product of claim 61, wherein the instructions executable to demarcate one or more traffic flows in the NoC, for a multiple input, multiple output node of the NoC, comprises:

receiving dedicated packets on each incoming port of the node;

replicating the received dedicated packets for each output port of the node; and

transmitting the replicated dedicated packets on each of the output ports of the node.