Performance monitoring mechanism for use in a pattern matching accelerator

• US 8,402,003 B2
• Filed: 02/08/2011
• Issued: 03/19/2013
• Est. Priority Date: 02/08/2011
• Status: Expired due to Fees

• Current Assignee: GlobalFoundries, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: Biran, Giora, Hagleitner, Christoph, Heil, Timothy H., Van Lunteren, Jan
• Primary Examiner(s): LE, DEBBIE M
• Application Number:
  US13/022,904
• Publication Number:
  US 20120203754A1
• Time in Patent Office: 770 Days
• Field of Search: 707/688, 707/706, 706/48, 711/128, 714/728, 714/733, 714/734, 714/738
• US Class Current: 707/688
• CPC Class Codes: G06F 16/90344 by using string matching techniques

Abstract

A pattern matching accelerator (PMA) for assisting software threads to find the presence and location of strings in an input data stream that match a given pattern. The patterns are defined using regular expressions that are compiled into a data structure comprised of rules subsequently processed by the PMA. The patterns to be searched in the input stream are defined by the user as a set of regular expressions. The patterns to be searched are grouped in pattern context sets. The sets of regular expressions which define the pattern context sets are compiled to generate a rules structure used by the PMA hardware. The rules are compiled before search run time and stored in main memory, in rule cache memory within the PMA or a combination thereof. For each input character, the PMA executes the search and returns the search results.

Claims

1. A performance monitoring system for use with a plurality of pattern search engines, comprising:

a collection circuit for collecting sample records from pattern search engines; and

a processing circuit coupled to said collection circuit and operative to generate statistical profiles of activity of said plurality of pattern search engine, wherein sample record statistics are collected at a plurality of resolution levels selected from the group comprising currently loaded contexts that are frequently accessed, tables within particular contexts that are frequently accessed and chunks within particular tables that are frequently accessed.

2. The system according to claim 1, wherein system is configured, initialized and read by an upload manager.

3. The system according to claim 1, wherein a resolution level comprises individual states.

4. The system according to claim 1, wherein a resolution level comprises individual transition rules.

5. The system according to claim 1, wherein a resolution level comprises cache blocks.

6. The system according to claim 1, wherein said collection circuit is operative to collect statistics for sequences of items.

7. The system according to claim 1, wherein statistical profiles of pattern search engine activity are used by an upload manager to determine which rules to be placed in a resident rule cache memory.

8. The system according to claim 1, wherein said processing circuit is operative to process sample record collected based on current system configuration.

9. The system according to claim 1, wherein said processing circuit comprises a plurality of performance counters that are incremented in response to sample record collected.

10. A performance monitoring apparatus for use with a plurality of pattern search engines, comprising

a collection circuit for collecting sample records from pattern search engines;

a processing circuit coupled to said collection circuit and operative to generate statistical profiles of activity of said plurality of pattern search engines; and

a plurality of performance counters coupled to said processing circuit, each performance counter operative to store a statistical profile of activity, wherein sample record statistics are collected at a resolution level selected from the group comprising currently loaded contexts that are frequently accessed, tables within particular contexts that are frequently accessed and chunks within particular tables that are frequently accessed.

11. The apparatus according to claim 10, wherein said collection circuit is operative to receive a sampling of total records processed by each pattern search engine.

12. The apparatus according to claim 10, wherein said collection circuit is operative to collect sample records randomly in time from each pattern search engine.

13. The apparatus according to claim 10, wherein said collection circuit is operative to collect sample records from said pattern search engines on a round robin basis.

14. The apparatus according to claim 10, wherein said collection circuit is operative to mark a sample as invalid if a corresponding pattern search engine is an idle state.

15. The apparatus according to claim 10, wherein said processing circuit is operative to increment a performance counter if a sample record collected matches a corresponding reference sample record.

16. The apparatus according to claim 10, wherein said processing circuit is operative to pass each sample record collected through a programmable hash function to generate an address to a performance counter.

17. A method of performance monitoring for use with a plurality of pattern search engines, said method comprising:

collecting sample records from said plurality of pattern search engines;

providing a plurality of performance counters for storing sample record counts collected from pattern search engines; and

processing said collected sample records to generate statistical profiles of activity of said plurality of pattern search engines and maintained said activity statistical profiles in said performance counters, wherein sample record statistics are collected at a resolution level selected from the group comprising currently loaded contexts that are frequently accessed, tables within particular contexts that are frequently accessed and chunks within particular tables that are frequently accessed.

18. The method according to claim 17, wherein said sample records are collected randomly in time from said pattern search engines.

19. The method according to claim 17, wherein sample records are collected from said pattern search engines on a round robin basis.

20. The method according to claim 17, further comprising marking a sample as invalid if a corresponding pattern search engine is an idle state.

21. The method according to claim 17, further comprising incrementing one of said plurality of performance counters if a sample record collected matches a corresponding reference sample record.

22. The method according to claim 17, further comprising passing each sample record collected through a programmable hash function to generate an address to one or said plurality of performance counters.

23. The method according to claim 17, further comprising determining, by an upload manager, which rules to be placed in a resident rule cache memory based on said statistical profiles of pattern search engine activity.

24. The method according to claim 17, wherein sample records are collected from said plurality of pattern search engines using hardware circuitry.

25. The method according to claim 17, wherein sample records are collected from said plurality of pattern search engines using software processes.

26. The method according to claim 17, wherein said plurality of performance counters are initialized and configured utilizing direct memory access (DMA) in response to a command by an upload manager.

27. The method according to claim 17, wherein said plurality of performance counters are read utilizing direct memory access (DMA) in response to a request by an upload manager.