Electrical computers and digital processing systems: processing – Processing control – Specialized instruction processing in support of testing,...
Reexamination Certificate
2000-07-31
2004-04-20
Coleman, Eric (Department: 2183)
Electrical computers and digital processing systems: processing
Processing control
Specialized instruction processing in support of testing,...
C717S127000
Reexamination Certificate
active
06725363
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to computer systems, and more particularly an apparatus, method and system to filter instructions to achieve more precise event counts.
2. Description of Related Art
Collecting performance data in an operating computer system is a frequent and extremely important task performed by hardware and software engineers. Hardware engineers need performance data to determine how computer hardware operates with operating systems and application programs. Specific designs of hardware structures, such as processor, memory and cache, can have drastically different, and sometimes unpredictable, utilizations for the same set of programs. Performance data can identify how efficiently software uses hardware, and can be helpful in designing improved systems.
It is a problem to accurately monitor the performance of hardware systems without disturbing the operating environment of the computer system, particularly if the performance data is collected over extended period of time such as days, or weeks. In some cases, performance monitoring systems are hand crafted. Costly hardware and software modifications may need to be implemented to ensure that operations of the system are not affected by the monitoring system.
Other profiling systems known in the art collect performance data relating to how frequently instructions are executed. Some prior art profiling systems require that source or object programs be modified to insert instructions which can collect the data when the programs are executing. Modifying the programs means that the programs need to be recompiled and/or relinked.
Additionally, prior art profiling systems generally only determine the frequency of execution of particular instructions, and not where time is spent overall in a program. In a pipelined multi-processor system, the number of cycles required to issue the various instructions is a significant indicator of the performance of the system.
Other known profiling systems can only generate profiles for instructions of an application program. This means that when the application program calls an operating system procedure, no performance data on actual instructions executed by the system procedure are collected. Some profiling systems may measure the amount of time it took to process the system call, and attempt to infer performance data by dividing the time for processing a system call by some “average” execution time of instructions. However, nothing concrete is learned about the actual execution of instructions of system procedures.
One way that the performance of a computer system can be monitored is by using performance counters. The performance counters are able to “count” occurrences of significant events in the system such as cache misses, instructions executed, cycle counts, and so forth. By checking the counters, the performance of the system can be deduced.
The counters could also be used to count the events to determine where time is spent overall in a program. It would be efficient to study the performance of the system if the event counters were on the processing chip itself and use software to extract the performance data from the event counters on the chip.
Thus, there is a need to study the performance of a system without modifying the existing system software. This is also a need for a more efficient method and system to study the performance of the system by applying an event counter onto the hardware itself. There is also a need for a more precise method and system to filter instructions to determine where time or resources are spent in a program rather than utilizing “averages” of counters.
SUMMARY OF THE INVENTION
This invention provides for an apparatus, method, and system for filtering instructions to obtain more precise event counts.
A first aspect of this invention is a method whereby a counter enable bit is registered with a plurality of instructions, executing the instructions thereby causing a plurality of events, filtering the instructions, activating the counter enable bit if the instructions fall within the filter, which then determines whether the event counter coupled to the event is incremented.
A second aspect of this invention is an apparatus to obtain more precise event counts which has a plurality of instructions with corresponding counter enable bits, a filter means to filter the instructions, a plurality of events associated with the plurality of instructions, and each plurality of events having a corresponding event counter incremented in response to the corresponding counter enable bit.
A last aspect of this invention is a system for filtering instructions to obtain more precise event counts having a counter enable bit registered to each of a plurality of instructions, executing the instructions thereby creating a plurality of events having associated event counters, means to filter the instructions, activating the counter enable bit if the instruction is filtered, deactivating the counter enable bit if the instruction is not filtered, and incrementing the associated event counter based upon the counter enable bit.
REFERENCES:
patent: 5107418 (1992-04-01), Cramer et al.
patent: 5659754 (1997-08-01), Grove et al.
patent: 5790859 (1998-08-01), Sarkar
patent: 5822788 (1998-10-01), Kahn et al.
patent: 5956477 (1999-09-01), Ranson et al.
patent: 6000044 (1999-12-01), Chrysos et al.
patent: 6026139 (2000-02-01), Hady et al.
patent: 6175913 (2001-01-01), Chesters et al.
patent: 6295613 (2001-09-01), Bates et al.
patent: 6470492 (2002-10-01), Bala et al.
patent: 6557119 (2003-04-01), Edwards et al.
patent: 6668372 (2003-12-01), Wu
patent: 1 063 585 (2000-12-01), None
Coleman Eric
Sun Microsystems Inc.
Thelen Reid & Priest LLP
Yeung Adrienne
LandOfFree
Method for filtering instructions to get more precise event... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for filtering instructions to get more precise event..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for filtering instructions to get more precise event... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3204833