WO1999057640A1 - Non-intrusive performance monitoring - Google Patents

Non-intrusive performance monitoring Download PDF

Info

Publication number
WO1999057640A1
WO1999057640A1 PCT/US1998/024899 US9824899W WO9957640A1 WO 1999057640 A1 WO1999057640 A1 WO 1999057640A1 US 9824899 W US9824899 W US 9824899W WO 9957640 A1 WO9957640 A1 WO 9957640A1
Authority
WO
WIPO (PCT)
Prior art keywords
circuit
signal
counter
performance parameter
recited
Prior art date
Application number
PCT/US1998/024899
Other languages
French (fr)
Inventor
Daniel Mann
Original Assignee
Advanced Micro Devices, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Advanced Micro Devices, Inc. filed Critical Advanced Micro Devices, Inc.
Priority to JP2000547546A priority Critical patent/JP2002513967A/en
Priority to EP98959548A priority patent/EP1076855B1/en
Priority to DE69803705T priority patent/DE69803705T2/en
Priority to KR1020007012246A priority patent/KR20010043286A/en
Publication of WO1999057640A1 publication Critical patent/WO1999057640A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/30Monitoring
    • G06F11/34Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F7/00Methods or arrangements for processing data by operating upon the order or content of the data handled
    • G06F7/60Methods or arrangements for performing computations using a digital non-denominational number representation, i.e. number representation without radix; Computing devices using combinations of denominational and non-denominational quantity representations, e.g. using difunction pulse trains, STEELE computers, phase computers
    • G06F7/70Methods or arrangements for performing computations using a digital non-denominational number representation, i.e. number representation without radix; Computing devices using combinations of denominational and non-denominational quantity representations, e.g. using difunction pulse trains, STEELE computers, phase computers using stochastic pulse trains, i.e. randomly occurring pulses the average pulse rates of which represent numbers
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/30Monitoring
    • G06F11/34Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
    • G06F11/3409Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/30Monitoring
    • G06F11/34Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
    • G06F11/3466Performance evaluation by tracing or monitoring
    • G06F11/348Circuit details, i.e. tracer hardware
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2201/00Indexing scheme relating to error detection, to error correction, and to monitoring
    • G06F2201/88Monitoring involving counting

Definitions

  • the invention relates to performance monitoring in integrated circuits.
  • two large counters are provided for event counting.
  • the counters can generally be read and written from within register address space.
  • the counters can be configured to measure such parameters as the number of data reads that hit m the cache.
  • the first counter is programmed to record the number of cache hits and the second counter is programmed to record the number of actual data reads performed.
  • the ratio of the two numbers gives the cache hit rate for read operations.
  • Measured performance parameters are a good estimate of future performance. Actual performance at any instant may vary widely from the measured estimate. The typical use of two large counters does not make any attempt to measure this deviation from the average.
  • the overflow signal stops all counting and generates an interrupt.
  • the software interrupt handler then records the counter values, completes post data processing and any other support work necessary.
  • the size of the counters is important. The larger the counter, the less frequently an interrupt is generated. Such interrupts are undesirable because they intrude into normal processor operation. A larger counter also results m greater data averagmg. Therefore, temporary fluctuation in cache hit rate may not be observed. Such temporary fluctuations may, or may not be what is of interest.
  • the invention provides a new technique for gatherings and analyzing performance data with a microprocessor or microcontroller or other integrated circuit.
  • the technique avoids the limitations imposed by fixed size counters which eventually overflow.
  • the method is less intrusive and suitable for monitoring a wide range of performance parameters.
  • an integrated circuit in a first embodiment includes a performance momtonng circuit which includes an adaptive adder circuit coupled to receive a first input signal indicative of a performance parameter of the integrated circuit and to provide a count value as a measure of the performance parameter
  • the adaptive adder circuit includes a random number generator circuit providing a random number, a counter circuit providing a count value and a comparator circuit coupled to compare the random number and the count value and to output a compare signal indicative thereof, the compare signal being provided to the counter as an up/down count signal.
  • the adaptive adder circuit also includes a first logic circuit coupled to receive the performance parameter being measured and the compare signal. The first logic circuit outputs a first control signal which is used for controlling operation of the counter circuit.
  • the counter circuit is responsive to count up or down when the first input signal and the compare signal are at different values and the counter circuit does not count when the first input signal and the compare signal are at the same value.
  • a method of measunng a performance parameter in an integrated circuit includes providing to an adaptive adder circuit, a first input signal indicative of the performance parameter and providing a count value in the adaptive adder circuit as a measure of the performance parameter.
  • the method further includes generating a random number in a random number generator circuit, providing a counter circuit providing the count value, and comparing the random number and the count value and providing a compare signal.
  • the compare signal is provided to the counter circuit as a first control signal.
  • the method further includes logically combining the performance parameter and the compare signal to provide a second control signal for controlling operation of the counter circuit.
  • Fig. 1 shows a block diagram of an integrated circuit incorporating the present invention.
  • Fig. 2 shows an adaptive adder circuit according to one embodiment of the invention.
  • Fig. 3 shows a graph of measured and input test data for an adaptive adder circuit having an eight bit counter measuring an input data stream having a probability of 25%.
  • Fig. 4 shows a graph of measured and input test data for an adaptive adder circuit having an eight bit counter measunng an input data stream having a probability of 78%.
  • Fig. 5 shows a graph of measured and input test data for an adaptive adder circuit having a twelve bit counter measunng an input data stream having a probability of 78%.
  • Fig. 6 shows a pseudo random number generator according to one embodiment of the invention.
  • Fig 7 shows an additional embodiment of the invention in which an integrated circuit has an adaptive adder circuit being provided multiple performance parameters, a select register to select a performance parameter, a sample counter and an input/output port.
  • an integrated circuit 101 which may be a microprocessor or microcontroller having a processor on it, includes functional logic 103 such as an on-chip cache memory, a performance detecting circuit 105 that detects performance data, such as whether there was a cache hit and performance momtonng circuit 107.
  • the type of performance data measured in integrated circuit 107 is random in nature; such as the cache hit rate, or the number of cycles to read memory. These parameters vary during program execution.
  • the sequence of hit and miss data can be determined by performance detecting circuit 105 and provided as a simple 1 or 0 bit stream to performance momtonng circuit 107.
  • the probability of a 1 is the probability of a hit occurring.
  • a performance momtonng circuit such as one shown in Figure 2 can be used to integrate the probability stream and determine the relevant probability of a cache hit or other performance parameter occurnng.
  • Figure 2 shows one embodiment of an adaptive adder circuit used as performance momtonng circuit 107.
  • a counter 201 provides a count value which is compared m comparator 203 with a random number generated in random number generator circuit 205. If the counter value is greater than or equal to the random number, a 1 is generated. Large counter values are more likely to produce a 1 output from the comparator than small counter values.
  • the compare signal 204 output from the comparator is provided back to counter 201 as an up/down count signal. When the comparator indicates that the count is larger than the random number, the compare signal 204 configures the counter 201 as a down counter and when the count is less than the random number, the compare signal 204 configures counter to be an up counter.
  • the compare signal 204 is compared with the input data stream of interest.
  • the input data stream is senally provided samples of the performance parameter being measured (e.g., the cache hit information) which are provided by performance detecting circuit 105.
  • These two stochastic data streams are compared to see which one has the highest probability of being 1 That is accomplished by XORing the two data streams together in XOR gate 207.
  • That probability information is fed back to increase or decrease the counter value according to the comparator output.
  • the feedback m the illustrated embodiment is accomplished by ANDmg together clock signal 209 and the output from XOR gate 207 to provide a gated clock signal 210 to the counter Consequently, with each new comparison the counter is adjusted to produce a probability stream (from the comparator) which matches the input data stream.
  • Table 1 illustrates the action of the counter: comparator 0 1 0 1 performance 0 0 1 1 parameter data
  • the adaptive adder circuit effectively integrates the probability stream.
  • the probability stream of the parameter being measured is converted into a digital value which is held m the counter.
  • the counter value represents the probability of the parameter which is being measured.
  • This method of measunng probability has several advantages over the dual counter method. For example, there is no potential overflow and therefore no need for an overflow interrupt handler. In addition , the counter can be read at any time to give a measure of the current probability.
  • an 8-bit counter provides for a probability resolution of 0.39% (1/255)
  • increasing the resolution slows down the integration process. That results in a greater number of samples being required before a good estimate of probability can be obtained.
  • Figure 3 shows an 8-bit adaptive adder circuit used to measure an input data stream having a probability of 25%
  • the adaptive adder circuit starts with an initial value of 50% As the input data is sampled, the adaptive adder circuit's counter value heads towards the expected value After about 500 5
  • the adaptive adder circuit closely tracks the input data stream. That indicates that after as little as 500 samples the counter could be read to produce a reasonable estimate of the performance parameter being measured.
  • Figure 4 shows the 8-bit adaptive adder circuit used to measure an input data stream having a
  • the 8-bit counter was initially at a value of 128, which represents a probability of 50% (128/256). Once again, after about 500 samples the counter produces a good estimate tracking the input data.
  • the time taken by the adaptive adder circuit's counter to move from its current position to a good estimate of the performance data depends on the number of bits used by the counter.
  • An 8-bit counter takes only 256 clocks to increment through its full range.
  • a 12-bit counter takes 4096 clock pulses to cover its complete range.
  • the extra 4 bits of probability resolution requires 16 times more clocks to reach its complete counting range This directly relates to the number of samples required to measure a performance parameter Greater resolution requires a greater number of samples to achieve a measurement with similar confidence.
  • Figure 5 shows a 12-bit adaptive adder circuit used to measure an input data stream having a 78% probability The counter was initialized to a 50% probability. The results show that about 10,000 samples were required before the input and measured probability streams converge.
  • the larger adaptive adder circuit does offer greater measurement resolution; but for measunng on-chip performance parameters, an 8-bit counter may be adequate.
  • the smaller adaptive adder circuit offers the benefit of more quickly converging on the required value and better tracking of local fluctuations of the input data stream.
  • the performance parameters described so far are easy to represent by a single bit. However, other parameters need more than 1 -bit. Parameters such as the number of cycles needed to access external memory, or the number of clocks while the pipeline is stalled require more bits. Those parameters must be converted into a pulse stream containing several bits. The example below shows how several bits may be used to measure a performance parameter which has a value ranging between 0 and 4. That requires 4 serial bits as shown below
  • a parameter is measured to be 1 cycle then one serial bit of four is set. If a parameter is measured to be, e.g., 3 cycles, then 3-b ⁇ ts of the 4-bit pulse stream are set to one. This pulse stream is serially clocked into the adaptive adder circuit. There is no synchronous requirement for presenting data to the adaptive adder circuit. New parameter data can be sampled at any time and clocked into the adaptive adder circuit.
  • An 8 -bit adaptive adder circuit can measure a parameter with value range 0 to 4, and then later be used to measure a parameter with a data range of 0 to 16. In all cases the adaptive adder circuit determines the probability of the data stream being 1. For a value range of 0 to 4, a probability value of 75% represents a measured value of 3 (100% being 4).
  • any 3 bits can be set. It does not matter the order. However, varying which bits are set to represent the sampled value may have some small advantage m reducing the vanance m counter operation.
  • the adaptive adder circuit operation uses a random number generator.
  • a pseudo random number generator such as the exemplary pseudo random number generator shown in Fig. 6 can be used for this task.
  • a maximum length (m-sequence) can be produced by feeding back selected stages of an n-stage shift register The required stages are modulo-2 combined and used to produce the input signal for the first stage
  • the test data shown in Figs. 3, 4 and 5 used a 31 -bit shift register as shown in Fig. 6 with feed back being taken from stages 3 and 31 , XORed m XOR gate 603 and provided to the input of the shift register.
  • the top 8 or 12 bits where used to form the required random number.
  • Other size shift registers (other than 31 -bits) could also be used.
  • the adaptive adder circuit can be encouraged to converge on the required value by using random numbers which correlate. That can be achieved by selecting consecutive stages of the shift register and inverting the most-significant bit (MSB). Hence, the MSB of the current random number will be inverted on output while the non-mverted value becomes the next-to MSB of the next random number.
  • MSB most-significant bit
  • Fig. 7 A more economical approach is illustrated in Fig. 7 where single adaptive adder circuit 701 receives a bit stream via an input multiplexer 703. That enables one of the multiple parameters to be selected for measurement. Select register 703 controls the selection of the performance parameter to be measured.
  • the adaptive adder circuit resides on an integrated circuit which includes a microprocessor
  • select register 705 controlling parameter selection and the adaptive adder circuit counter should be accessible by the processor.
  • the select register and counter could be mapped into register, I/O or memory address space.
  • these registers may be accessible from an on- chip Software Development Port (SDP) which is an mput/output port providing test and debug access to internal registers of the processor, as is known in the art.
  • SDP Software Development Port
  • a host computer (not shown) connected to the mput/output port, can unobtrusively examine performance parameters. There is no need to instrument the target system's application or operating system software and no interrupt handler is required
  • sample counter 709 provides the ability to determine when the adaptive adder circuit can be assumed to have tracked the new input data stream. Sample counter 709 may be reset when the register selecting parameter input is updated The sample counter may then be examined via the Software Development Port or from the processor before accessing the adaptive adder circuit's counter to ensure a sufficient number of samples have been provided to the adaptive adder circuit so that the counter value tracks the actual data The sample counter should be large enough to reflect the size of the counter in the adaptive adder circuit.
  • the sample counter in the adaptive adder circuit is 12 bits, the sample counter needs to count to 10,000 or more to satisfactonly indicate when the measured data converges with the input parameter data.
  • Overflow indications m the sample counter can be utilized to indicate a satisfactory number of parameter samples have been received by the adaptive adder circuit.
  • measured performance parameters can include bus utilization by the processor.
  • bus utilization by the processor in a manner similar to cache hits. For example, for each bus cycle, a "1" is provided to the performance monitoring circuit to indicate that the processor was the bus master for the cycle and a "0" indicates that an external bus master had the bus. Thus, the probability of the processor being bus master can be determined.

Abstract

An integrated circuit (101) includes a performance monitoring circuit (107) which includes an adaptive adder circuit coupled to receive a first input signal indicative of a performance parameter of the integrated circuit and to provide a count value as a measure of the probability of the performance parameter. The adaptive adder circuit includes a random number generator circuit (205) providing a random number, a counter circuit (201) providing a count value and a comparator circuit (203) coupled to compare the random number and the count value and to output a compare signal (204) indicative thereof, the compare signal being provided to the counter as an up/down count signal. The adaptive adder circuit also includes a first logic circuit (207) coupled to receive the performance parameter being measured and the compare signal which outputs a first signal which is used for controlling operation of the counter circuit. The counter circuit is responsive to count up or down when the first input signal and the compare signal are at different values and the counter circuit does not count when the first input signal and the compare signal are at the same value.

Description

NON-INTRUSIVE PERFORMANCE MONITORING
Technical Field
The invention relates to performance monitoring in integrated circuits.
Background Art Most microprocessors used in desktop computer systems, are equipped with performance monitoring counters. These counters permit processor performance parameters to be monitored and measured. Such information is useful for performance tuning. Current techniques typically utilize two counters which simultaneously record the occurrence of prespecifϊed events. When one of the counters overflows, counting stops and an interrupt is generated Post-processmg software is used to analyze the gathered data.
Typically two large counters, of e.g., 40-bιts or more, are provided for event counting. The counters can generally be read and written from within register address space. The counters can be configured to measure such parameters as the number of data reads that hit m the cache. When configured to determined cache hits, the first counter is programmed to record the number of cache hits and the second counter is programmed to record the number of actual data reads performed. The ratio of the two numbers gives the cache hit rate for read operations. Measured performance parameters are a good estimate of future performance. Actual performance at any instant may vary widely from the measured estimate. The typical use of two large counters does not make any attempt to measure this deviation from the average.
When one of the counters reaches its limit, the overflow signal stops all counting and generates an interrupt. The software interrupt handler then records the counter values, completes post data processing and any other support work necessary.
The size of the counters is important. The larger the counter, the less frequently an interrupt is generated. Such interrupts are undesirable because they intrude into normal processor operation. A larger counter also results m greater data averagmg. Therefore, temporary fluctuation in cache hit rate may not be observed. Such temporary fluctuations may, or may not be what is of interest.
Before performance momtonng can be accomplished, an interrupt handler must be installed to deal with counter overflow. Of course, overflow can be avoided by the use of extremely large counters. But extremely large counters may be expensive to implement, unreliable or fail to produce the desired analysis It would be desirable to monitor performance parameters in an integrated circuit such as a processor without having to provide two large counters, without having to deal with counter overflow, 2
without having to provide software interrupt handlers. It would also be desirable to avoid intrusion of the performance monitoring into the normal functioning of the processor in the system.
Disclosure of Invention
Accordingly, the invention provides a new technique for gatherings and analyzing performance data with a microprocessor or microcontroller or other integrated circuit. The technique avoids the limitations imposed by fixed size counters which eventually overflow. The method is less intrusive and suitable for monitoring a wide range of performance parameters.
In a first embodiment an integrated circuit includes a performance momtonng circuit which includes an adaptive adder circuit coupled to receive a first input signal indicative of a performance parameter of the integrated circuit and to provide a count value as a measure of the performance parameter The adaptive adder circuit includes a random number generator circuit providing a random number, a counter circuit providing a count value and a comparator circuit coupled to compare the random number and the count value and to output a compare signal indicative thereof, the compare signal being provided to the counter as an up/down count signal. The adaptive adder circuit also includes a first logic circuit coupled to receive the performance parameter being measured and the compare signal. The first logic circuit outputs a first control signal which is used for controlling operation of the counter circuit. The counter circuit is responsive to count up or down when the first input signal and the compare signal are at different values and the counter circuit does not count when the first input signal and the compare signal are at the same value.
In another embodiment a method of measunng a performance parameter in an integrated circuit includes providing to an adaptive adder circuit, a first input signal indicative of the performance parameter and providing a count value in the adaptive adder circuit as a measure of the performance parameter. The method further includes generating a random number in a random number generator circuit, providing a counter circuit providing the count value, and comparing the random number and the count value and providing a compare signal. The compare signal is provided to the counter circuit as a first control signal. The method further includes logically combining the performance parameter and the compare signal to provide a second control signal for controlling operation of the counter circuit.
Brief Description of the Drawings
The present invention may be better understood, and its numerous objects, features, and advantages made apparent to those skilled in the art by referencing the accompanying drawings.
Fig. 1 shows a block diagram of an integrated circuit incorporating the present invention. Fig. 2 shows an adaptive adder circuit according to one embodiment of the invention.
Fig. 3 shows a graph of measured and input test data for an adaptive adder circuit having an eight bit counter measuring an input data stream having a probability of 25%.
Fig. 4 shows a graph of measured and input test data for an adaptive adder circuit having an eight bit counter measunng an input data stream having a probability of 78%.
Fig. 5 shows a graph of measured and input test data for an adaptive adder circuit having a twelve bit counter measunng an input data stream having a probability of 78%.
Fig. 6 shows a pseudo random number generator according to one embodiment of the invention.
Fig 7 shows an additional embodiment of the invention in which an integrated circuit has an adaptive adder circuit being provided multiple performance parameters, a select register to select a performance parameter, a sample counter and an input/output port.
Modes for Carrying out the Invention
Refernng to Figure 1 an integrated circuit 101 which may be a microprocessor or microcontroller having a processor on it, includes functional logic 103 such as an on-chip cache memory, a performance detecting circuit 105 that detects performance data, such as whether there was a cache hit and performance momtonng circuit 107. The type of performance data measured in integrated circuit 107 is random in nature; such as the cache hit rate, or the number of cycles to read memory. These parameters vary during program execution.
Consider for example, that at each memory access an on-chip cache may successfully provide the required data or may not. The sequence of hit and miss data can be determined by performance detecting circuit 105 and provided as a simple 1 or 0 bit stream to performance momtonng circuit 107. The probability of a 1 is the probability of a hit occurring. In order to measure the cache hit performance, a performance momtonng circuit such as one shown in Figure 2 can be used to integrate the probability stream and determine the relevant probability of a cache hit or other performance parameter occurnng.
Figure 2 shows one embodiment of an adaptive adder circuit used as performance momtonng circuit 107. A counter 201 provides a count value which is compared m comparator 203 with a random number generated in random number generator circuit 205. If the counter value is greater than or equal to the random number, a 1 is generated. Large counter values are more likely to produce a 1 output from the comparator than small counter values. The compare signal 204 output from the comparator is provided back to counter 201 as an up/down count signal. When the comparator indicates that the count is larger than the random number, the compare signal 204 configures the counter 201 as a down counter and when the count is less than the random number, the compare signal 204 configures counter to be an up counter.
The compare signal 204 is compared with the input data stream of interest. The input data stream is senally provided samples of the performance parameter being measured (e.g., the cache hit information) which are provided by performance detecting circuit 105. These two stochastic data streams (compare signal and input data stream) are compared to see which one has the highest probability of being 1 That is accomplished by XORing the two data streams together in XOR gate 207. When the data streams differ, there is a difference in probability. That probability information is fed back to increase or decrease the counter value according to the comparator output. The feedback m the illustrated embodiment is accomplished by ANDmg together clock signal 209 and the output from XOR gate 207 to provide a gated clock signal 210 to the counter Consequently, with each new comparison the counter is adjusted to produce a probability stream (from the comparator) which matches the input data stream.
Table 1 illustrates the action of the counter: comparator 0 1 0 1 performance 0 0 1 1 parameter data
XOR output 0 1 1 0
Figure imgf000006_0001
count action none down up none
The adaptive adder circuit effectively integrates the probability stream. The probability stream of the parameter being measured is converted into a digital value which is held m the counter. The counter value represents the probability of the parameter which is being measured.
This method of measunng probability has several advantages over the dual counter method. For example, there is no potential overflow and therefore no need for an overflow interrupt handler. In addition , the counter can be read at any time to give a measure of the current probability.
As the number of bits used by the counter and random number generator increase the probability resolution is improved. For example, an 8-bit counter provides for a probability resolution of 0.39% (1/255) However, increasing the resolution slows down the integration process. That results in a greater number of samples being required before a good estimate of probability can be obtained.
Figure 3 shows an 8-bit adaptive adder circuit used to measure an input data stream having a probability of 25% The adaptive adder circuit starts with an initial value of 50% As the input data is sampled, the adaptive adder circuit's counter value heads towards the expected value After about 500 5
samples, the adaptive adder circuit closely tracks the input data stream. That indicates that after as little as 500 samples the counter could be read to produce a reasonable estimate of the performance parameter being measured.
Dunng the generation of the presented data, the input data stream was averaged over a sliding window of 250 samples. Thus, each point shown graphed as input test data in Figs. 3-5 represent the average of the last 250 samples (or less if the number of data samples is less than 250). Thus, the graphs show how quickly the measured value produces a reasonable estimate of a sliding window average of 250 samples. Keeping this window small helps observe temporary fluctuations in both the generated test data and adaptive adder circuit response.
Figure 4 shows the 8-bit adaptive adder circuit used to measure an input data stream having a
78% probability. The 8-bit counter was initially at a value of 128, which represents a probability of 50% (128/256). Once again, after about 500 samples the counter produces a good estimate tracking the input data.
The time taken by the adaptive adder circuit's counter to move from its current position to a good estimate of the performance data depends on the number of bits used by the counter. An 8-bit counter takes only 256 clocks to increment through its full range. A 12-bit counter takes 4096 clock pulses to cover its complete range. The extra 4 bits of probability resolution requires 16 times more clocks to reach its complete counting range This directly relates to the number of samples required to measure a performance parameter Greater resolution requires a greater number of samples to achieve a measurement with similar confidence.
Figure 5 shows a 12-bit adaptive adder circuit used to measure an input data stream having a 78% probability The counter was initialized to a 50% probability. The results show that about 10,000 samples were required before the input and measured probability streams converge.
The larger adaptive adder circuit does offer greater measurement resolution; but for measunng on-chip performance parameters, an 8-bit counter may be adequate. The smaller adaptive adder circuit offers the benefit of more quickly converging on the required value and better tracking of local fluctuations of the input data stream.
The performance parameters described so far, e.g., cache hit information, are easy to represent by a single bit. However, other parameters need more than 1 -bit. Parameters such as the number of cycles needed to access external memory, or the number of clocks while the pipeline is stalled require more bits. Those parameters must be converted into a pulse stream containing several bits. The example below shows how several bits may be used to measure a performance parameter which has a value ranging between 0 and 4. That requires 4 serial bits as shown below
1 0 0 0 2 0 1 0 3 1 1 0 1 0 0 0 4
Figure imgf000008_0001
1 1 1
If a parameter is measured to be 1 cycle then one serial bit of four is set. If a parameter is measured to be, e.g., 3 cycles, then 3-bιts of the 4-bit pulse stream are set to one. This pulse stream is serially clocked into the adaptive adder circuit. There is no synchronous requirement for presenting data to the adaptive adder circuit. New parameter data can be sampled at any time and clocked into the adaptive adder circuit.
Different parameters will likely have differing value ranges The above example has a value range of 0 to 4, but another parameter may have a value range 0 to 16. There is no need for all parameters to be restncted to the same data range. Adjusting the number of data bits used for each parameters data range, helps make best use of the adaptive adder circuit's restncted probability resolution.
An 8 -bit adaptive adder circuit can measure a parameter with value range 0 to 4, and then later be used to measure a parameter with a data range of 0 to 16. In all cases the adaptive adder circuit determines the probability of the data stream being 1. For a value range of 0 to 4, a probability value of 75% represents a measured value of 3 (100% being 4).
When generating a sample value of, e.g., 3, any 3 bits can be set. It does not matter the order. However, varying which bits are set to represent the sampled value may have some small advantage m reducing the vanance m counter operation.
The adaptive adder circuit operation uses a random number generator. A pseudo random number generator such as the exemplary pseudo random number generator shown in Fig. 6 can be used for this task. A maximum length (m-sequence) can be produced by feeding back selected stages of an n-stage shift register The required stages are modulo-2 combined and used to produce the input signal for the first stage The test data shown in Figs. 3, 4 and 5 used a 31 -bit shift register as shown in Fig. 6 with feed back being taken from stages 3 and 31 , XORed m XOR gate 603 and provided to the input of the shift register. The top 8 or 12 bits where used to form the required random number. Other size shift registers (other than 31 -bits) could also be used. The adaptive adder circuit can be encouraged to converge on the required value by using random numbers which correlate. That can be achieved by selecting consecutive stages of the shift register and inverting the most-significant bit (MSB). Hence, the MSB of the current random number will be inverted on output while the non-mverted value becomes the next-to MSB of the next random number.
There are typically multiple performance parameters of interest but there is no need to have an adaptive adder circuit allocated to each parameter being measured. A more economical approach is illustrated in Fig. 7 where single adaptive adder circuit 701 receives a bit stream via an input multiplexer 703. That enables one of the multiple parameters to be selected for measurement. Select register 703 controls the selection of the performance parameter to be measured.
In one embodiment, the adaptive adder circuit resides on an integrated circuit which includes a microprocessor In that case, select register 705 controlling parameter selection and the adaptive adder circuit counter should be accessible by the processor. The select register and counter could be mapped into register, I/O or memory address space. In addition, these registers may be accessible from an on- chip Software Development Port (SDP) which is an mput/output port providing test and debug access to internal registers of the processor, as is known in the art. Via the SDP, a host computer (not shown) connected to the mput/output port, can unobtrusively examine performance parameters. There is no need to instrument the target system's application or operating system software and no interrupt handler is required
After setting the select register, an 8-bit adaptive adder circuit typically requires that about 500 or more samples be gathered before the adaptive adder circuit is assumed to have tracked the new input data stream with a satisfactory degree of accuracy. To assist with that determination, sample counter 709 provides the ability to determine when the adaptive adder circuit can be assumed to have tracked the new input data stream. Sample counter 709 may be reset when the register selecting parameter input is updated The sample counter may then be examined via the Software Development Port or from the processor before accessing the adaptive adder circuit's counter to ensure a sufficient number of samples have been provided to the adaptive adder circuit so that the counter value tracks the actual data The sample counter should be large enough to reflect the size of the counter in the adaptive adder circuit. Thus, if the counter in the adaptive adder circuit is 12 bits, the sample counter needs to count to 10,000 or more to satisfactonly indicate when the measured data converges with the input parameter data. Overflow indications m the sample counter can be utilized to indicate a satisfactory number of parameter samples have been received by the adaptive adder circuit.
In addition to the performance parameters described so far, measured performance parameters can include bus utilization by the processor. In one embodiment, a performance detecting circuit maps 8
bus utilization by the processor in a manner similar to cache hits. For example, for each bus cycle, a "1" is provided to the performance monitoring circuit to indicate that the processor was the bus master for the cycle and a "0" indicates that an external bus master had the bus. Thus, the probability of the processor being bus master can be determined.
The description of the invention set forth herein is illustrative, and is not intended to limit the scope of the invention as set forth in the following claims. Variations and modifications may be made based on the description set forth herein, without departing from the scope and spirit of the invention as set forth in the following claims.

Claims

We Claim:
1. An integrated circuit composing: a performance momtonng circuit including an adaptive adder circuit coupled to receive a first input signal indicative of a performance parameter of the integrated circuit and to provide a count value as a measure of the performance parameter.
2. The integrated circuit as recited m claim 1 wherein the adaptive adder circuit compnses: a random number generator circuit providing a random number; a counter circuit providing the count value; a comparator circuit coupled to compare the random number and the count value and to output a compare signal indicative thereof, the compare signal being provided to the counter circuit as an up/down count signal; and a first logic circuit coupled to receive the performance parameter and the compare signal and output a first control signal indicative thereof, the first control signal being provided for controlling operation of the counter circuit.
3. The integrated circuit as recited in claim 2 wherein the counter circuit is responsive to count up or down when the first input signal and the compare signal are at different values and wherein the counter circuit does not count when the first input signal and the compare signal are at the same value.
4. The integrated circuit as recited in claim 3 wherein the counter circuit is responsive to count down when the count value is greater than the random number and the first input signal is at a first value and wherein the counter is responsive to count up when the count value is less than the random number and the first input signal is at a second value.
5. The integrated circuit as recited in claim 2 further compnsmg a second logic circuit coupled to receive the compare signal and a clock signal, and outputtmg a gated clock signal to the counter circuit to selectively enable the counter circuit according to the value of the compare signal.
6 The integrated circuit as recited m claim 1 wherein the performance parameter is provided as a binary number to the performance momtonng circuit.
7 The integrated circuit as recited in claim 6 further compnsmg a performance parameter detect circuit coupled to provide the performance parameter as the first signal to the adaptive adder 10
circuit, the performance parameter detect circuit providing the first signal with a first value when a read operation results m a cache hit and a second value when another read operation results m a cache miss
8. The integrated circuit as recited in claim 1 wherein the performance parameter has a value between 0 and n, n being a number greater than 1 and wherein the performance parameter is provided to the adaptive adder circuit as a serial bit stream, with n bits being present in the bit stream to represent the value of the performance parameter.
9. The integrated circuit as recited m claim 1 wherein the integrated circuit is a microprocessor
10 The integrated circuit as recited m claim 1 wherein the performance parameter is a parameter associated with bus utilization for a bus connected to said integrated circuit.
11 The integrated circuit as recited m claim 1 wherein the performance monitoring circuit further comprises: a selector circuit receiving a plurality of performance parameters, the selector circuit being coupled to the adaptive adder circuit to provide a selected one of the performance parameters as the first input; and a programmable select register coupled to control the selector circuit, the select register being programmable to select one of the performance parameters as the first input.
12 The integrated circuit as recited m claim 11 wherein the performance momtonng circuit further comprises a sample counter, the sample counter coupled to count a number of samples of the performance parameter received by the performance monitoring circuit and to provide a sample count value indicative thereof, the sample count value indicating the accuracy of the count value in the performance momtonng circuit.
13. A method of measuring a performance parameter in an integrated circuit, comprising: providing a first input signal indicative of the performance parameter to an adaptive adder circuit: and determining a count value in the adaptive adder circuit as a probability measure of the performance parameter. 11
14 The method as recited in claim 13 wherein determining the count value further compnses: generating a random number in a random number generator circuit; providing a counter circuit providing the count value; companng the random number and the count value and providing a compare signal indicative thereof, the compare signal being provided to the counter circuit as a first control signal; and logically combining the performance parameter and the compare signal to provide a second control signal for controlling operation of the counter circuit.
15 The method as recited in claim 14 wherein providing the first control signal to the counter circuit further comprises providing the compare signal to the counter circuit as an up/down count signal, the counter circuit counting down when the count value is greater than the random number and the first input signal is at a first value and the counter circuit counting up when the count value is less than the random number and the first input signal is at a second value.
16 The method as recited in claim 15 wherein providing the second control signal further comprises- logically combining the compare signal and a clock signal to provide a gated clock signal to the counter circuit, the gated clock signal selectively enabling the counter circuit according to the value of the compare signal and the clock signal.
17 The method as recited in claim 13 wherein the performance parameter being measured is a hit rate in a cache memory and further comprising providing the performance parameter as the first signal havmg a first value when a read operation results m a cache hit and havmg a second value when another read operation results in a cache miss.
18 The method as recited in claim 13 further comprising: providing a plurality of performance parameters to a selector circuit coupled to the adaptive adder circuit; selecting one of the performance parameters to provide a selected one of the performance parameters as the first input. 12
19. The method as recited in claim 18 further comprising: counting a number of samples of the performance parameter received by the adaptive adder circuit and providing a sample count value indicative thereof; determining when to access the count value as a valid performance parameter according to the sample count value.
20. An apparatus for measunng a performance parameter of a computer system, compnsmg: means for providing a first input signal indicative of the performance parameter to an adaptive adder circuit; and means for generating a count value as a measure of the performance parameter.
21. The apparatus as recited in claim 20 wherein the means for generating the count value compnses: means for generating a random number; a counter circuit providing the count value; means for companng the random number and the count value and for providing a compare signal indicative thereof; means for controlling the counter circuit according to the compare signal and the first input signal.
22. The apparatus as recited m claim 21 further compnsmg means for selecting one of a plurality of performance parameters as the first input.
23. The apparatus as recited m claim 21 further compnsmg: means for counting a number of samples of the performance parameter received by the means for generating the count and providing a sample count value indicative thereof.
PCT/US1998/024899 1998-05-05 1998-11-23 Non-intrusive performance monitoring WO1999057640A1 (en)

Priority Applications (4)

Application Number Priority Date Filing Date Title
JP2000547546A JP2002513967A (en) 1998-05-05 1998-11-23 Non-intrusive performance monitor
EP98959548A EP1076855B1 (en) 1998-05-05 1998-11-23 Non-intrusive performance monitoring
DE69803705T DE69803705T2 (en) 1998-05-05 1998-11-23 NON-INFLUENCING PERFORMANCE MONITORING
KR1020007012246A KR20010043286A (en) 1998-05-05 1998-11-23 Non-intrusive performance monitoring

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US09/072,830 1998-05-05
US09/072,830 US6275782B1 (en) 1998-05-05 1998-05-05 Non-intrusive performance monitoring

Publications (1)

Publication Number Publication Date
WO1999057640A1 true WO1999057640A1 (en) 1999-11-11

Family

ID=22110014

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US1998/024899 WO1999057640A1 (en) 1998-05-05 1998-11-23 Non-intrusive performance monitoring

Country Status (6)

Country Link
US (2) US6275782B1 (en)
EP (1) EP1076855B1 (en)
JP (1) JP2002513967A (en)
KR (1) KR20010043286A (en)
DE (1) DE69803705T2 (en)
WO (1) WO1999057640A1 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6415243B1 (en) 1998-05-05 2002-07-02 Advanced Micro Devices, Inc. Performance monitoring and optimization using an adaptive digital circuit
US6556952B1 (en) 2000-05-04 2003-04-29 Advanced Micro Devices, Inc. Performance monitoring and optimizing of controller parameters
EP2110750A1 (en) 2008-04-18 2009-10-21 MediaTek Inc. Data access tracing
US8316214B2 (en) 2007-04-18 2012-11-20 Mediatek Inc. Data access tracing with compressed address output

Families Citing this family (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6832236B1 (en) * 1999-07-08 2004-12-14 International Business Machines Corporation Method and system for implementing automatic filesystem growth monitor for production UNIX computer system
DE19933491A1 (en) * 1999-07-09 2001-02-01 Walter Mehner Method for serial transmission of digital measurement data
US6437783B1 (en) * 1999-09-13 2002-08-20 Intel Corporation Method and system for simultaneously displaying the throughput on multiple busses
US6629170B1 (en) * 1999-11-08 2003-09-30 International Business Machines Corporation Method and apparatus for a byte lane selectable performance monitor bus
JP2001209500A (en) * 2000-01-28 2001-08-03 Fujitsu Ltd Disk device and read/write processing method threof
US6775640B1 (en) * 2000-04-28 2004-08-10 Hewlett-Packard Development Company, L.P. Performance adder for tracking occurrence of events within a circuit
US6628994B1 (en) * 2000-08-31 2003-09-30 Hewlett-Packard Development Company, L.P. Method to obtain improved performance by automatic adjustment of computer system parameters
US6895520B1 (en) 2001-03-02 2005-05-17 Advanced Micro Devices, Inc. Performance and power optimization via block oriented performance measurement and control
US6968547B2 (en) * 2002-01-29 2005-11-22 Sun Microsystems, Inc. Dynamic trap table interposition for efficient collection of trap statistics
US7168853B2 (en) * 2003-01-10 2007-01-30 International Business Machines Corporation Digital measuring system and method for integrated circuit chip operating parameters
US7275191B2 (en) * 2003-05-09 2007-09-25 Hewlett-Packard Development Company, L.P. Coverage decoder circuit for performance counter
US7475301B2 (en) * 2003-05-09 2009-01-06 Hewlett-Packard Development Company, L.P. Increment/decrement circuit for performance counter
US7107362B2 (en) * 2003-05-19 2006-09-12 Avago Technologies General Ip (Singapore) Pte. Ltd. Integrated circuit with configuration based on parameter measurement
US7146285B2 (en) * 2003-05-19 2006-12-05 Avago Technologies General Ip (Singapore) Pte. Ltd. Integrated circuit with parameter measurement
US7325034B2 (en) * 2003-09-24 2008-01-29 International Business Machines Corporation Method and apparatus for scalable peer-to-peer inquiries in a network of untrusted parties
US7519510B2 (en) * 2004-11-18 2009-04-14 International Business Machines Corporation Derivative performance counter mechanism
US7340378B1 (en) 2006-11-30 2008-03-04 International Business Machines Corporation Weighted event counting system and method for processor performance measurements
US7908493B2 (en) * 2007-06-06 2011-03-15 International Business Machines Corporation Unified management of power, performance, and thermals in computer systems
WO2009058042A1 (en) * 2007-10-29 2009-05-07 Intel Corporation A method of external performance monitoring for virtualized environments
US7984206B2 (en) * 2008-08-06 2011-07-19 Texas Instruments Incorporated System for debugging throughput deficiency in an architecture using on-chip throughput computations
US9100153B2 (en) * 2008-09-25 2015-08-04 The Royal Institution For The Advancement Of Learning/Mcgill University Methods and systems for improving iterative signal processing
US8055477B2 (en) * 2008-11-20 2011-11-08 International Business Machines Corporation Identifying deterministic performance boost capability of a computer system
US10169187B2 (en) 2010-08-18 2019-01-01 International Business Machines Corporation Processor core having a saturating event counter for making performance measurements
US9015374B2 (en) 2013-07-09 2015-04-21 Advanced Micro Devices, Inc. Virtual interrupt filter
US9715377B1 (en) * 2016-01-04 2017-07-25 International Business Machines Corporation Behavior based code recompilation triggering scheme

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4176402A (en) * 1978-05-24 1979-11-27 Sperry Rand Corporation Apparatus for simultaneously measuring a plurality of digital events employing a random number table
GB2304215A (en) * 1995-08-09 1997-03-12 Toshiba Kk Monitoring cache-miss rates for power saving

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3812478A (en) * 1971-07-31 1974-05-21 Nippon Musical Instruments Mfg Semiconductor storage device
US3887869A (en) * 1972-07-25 1975-06-03 Tau Tron Inc Method and apparatus for high speed digital circuit testing
BE856104A (en) 1977-07-05 1977-12-27 G Sojuznoe Konstruktorsko Tekh COMPUTER SPECIALIZED IN STATISTICAL PROCESSING OF INFORMATION
GB1543515A (en) 1978-01-16 1979-04-04 Sojuz Kt Bjuro P Schetnykh Mas Statistical data processing digital computer
US4409592A (en) * 1981-04-20 1983-10-11 Hunt V Bruce Multipoint packet data communication system using random access and collision detection techniques
US4608559A (en) * 1982-08-19 1986-08-26 Computer Automation, Inc. Local modulated carrier data network with a collision avoidance protocol
US4694412A (en) * 1982-09-22 1987-09-15 Intel Corporation Random number generator for use in an authenticated read-only memory
US5412587A (en) 1988-12-28 1995-05-02 The Boeing Company Pseudorandom stochastic data processing
IL96808A (en) * 1990-04-18 1996-03-31 Rambus Inc Integrated circuit i/o using a high performance bus interface
US5657253A (en) 1992-05-15 1997-08-12 Intel Corporation Apparatus for monitoring the performance of a microprocessor
US5349611A (en) * 1992-11-13 1994-09-20 Ampex Systems Corporation Recovering synchronization in a data stream
EP0689141A3 (en) * 1994-06-20 1997-10-15 At & T Corp Interrupt-based hardware support for profiling system performance
US5557548A (en) 1994-12-09 1996-09-17 International Business Machines Corporation Method and system for performance monitoring within a data processing system
US5696828A (en) * 1995-09-22 1997-12-09 United Technologies Automotive, Inc. Random number generating system and process based on chaos
US5778194A (en) 1996-04-08 1998-07-07 Symbios, Inc. Method and apparatus for measuring performance of a computer bus
US5768152A (en) 1996-08-28 1998-06-16 International Business Machines Corp. Performance monitoring through JTAG 1149.1 interface
US5835702A (en) 1996-10-21 1998-11-10 International Business Machines Corporation Performance monitor
US5835705A (en) 1997-03-11 1998-11-10 International Business Machines Corporation Method and system for performance per-thread monitoring in a multithreaded processor
US5919268A (en) * 1997-09-09 1999-07-06 Ncr Corporation System for determining the average latency of pending pipelined or split transaction requests through using two counters and logic divider
US6119075A (en) * 1997-11-26 2000-09-12 Digital Equipment Corporation Method for estimating statistics of properties of interactions processed by a processor pipeline
US6275782B1 (en) * 1998-05-05 2001-08-14 Advanced Micro Devices, Inc. Non-intrusive performance monitoring

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4176402A (en) * 1978-05-24 1979-11-27 Sperry Rand Corporation Apparatus for simultaneously measuring a plurality of digital events employing a random number table
GB2304215A (en) * 1995-08-09 1997-03-12 Toshiba Kk Monitoring cache-miss rates for power saving

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
"HARDWARE PERFORMANCE TRACE ENHANCEMENT TO INCLUDE PERFORMANCE EVENTS", IBM TECHNICAL DISCLOSURE BULLETIN, vol. 39, no. 10, October 1996 (1996-10-01), pages 87/88, XP000631336 *
"PROCESSOR PERFORMANCE MONITORING WITH A DEPICTION OF THE EFFICIENCYOF THE CACHE COHERENCY PROTOCOL OF SUPERSCALAR MICROPROCESSOR IN ANSYMMETRIC MULTIPLE PROCESSOR ENVIRONMENT", IBM TECHNICAL DISCLOSURE BULLETIN, vol. 40, no. 1, January 1997 (1997-01-01), pages 79 - 81, XP000686109 *
"SELF-ADJUSTING UTILIZATION MONITOR", IBM TECHNICAL DISCLOSURE BULLETIN, vol. 38, no. 2, 1 February 1995 (1995-02-01), New York, US, pages 371 - 372, XP000502509 *
B. GAINES: "Stochastic computer thrives on noise", ELECTRONICS., vol. 40, no. 14, 10 July 1967 (1967-07-10), NEW YORK US, pages 72 - 79, XP002092489 *
S ROHAVA: "Elements of Stochastic Systems - Instrumentation of Non-Deterministic Learning Patterns", SOVIET AUTOMATIC CONTROL, vol. 13, no. 5, 1968, New York, US, pages 67 - 69, XP002092490 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6415243B1 (en) 1998-05-05 2002-07-02 Advanced Micro Devices, Inc. Performance monitoring and optimization using an adaptive digital circuit
US6556952B1 (en) 2000-05-04 2003-04-29 Advanced Micro Devices, Inc. Performance monitoring and optimizing of controller parameters
US8316214B2 (en) 2007-04-18 2012-11-20 Mediatek Inc. Data access tracing with compressed address output
EP2110750A1 (en) 2008-04-18 2009-10-21 MediaTek Inc. Data access tracing

Also Published As

Publication number Publication date
JP2002513967A (en) 2002-05-14
DE69803705T2 (en) 2002-09-12
US6275782B1 (en) 2001-08-14
US6415243B1 (en) 2002-07-02
KR20010043286A (en) 2001-05-25
EP1076855A1 (en) 2001-02-21
EP1076855B1 (en) 2002-01-30
DE69803705D1 (en) 2002-03-14

Similar Documents

Publication Publication Date Title
EP1076855B1 (en) Non-intrusive performance monitoring
US4821178A (en) Internal performance monitoring by event sampling
US5103394A (en) Software performance analyzer
US6519310B2 (en) Hardware event based flow control of counters
US4845615A (en) Software performance analyzer
KR100280732B1 (en) Performance monitoring method and apparatus of data processing system
US6351724B1 (en) Apparatus and method for monitoring the performance of a microprocessor
EP0130469B1 (en) Internally distributed monitoring system
US5940618A (en) Code instrumentation system with non intrusive means and cache memory optimization for dynamic monitoring of code segments
US7433803B2 (en) Performance monitor with precise start-stop control
US4435759A (en) Hardware monitor for obtaining processor software/hardware interrelationships
US6341357B1 (en) Apparatus and method for processor performance monitoring
US20060005083A1 (en) Performance count tracing
US20060184832A1 (en) Method and apparatus for achieving high cycle/trace compression depth by adding width
JP2819007B2 (en) Logic analyzer
JPH0816878B2 (en) Apparatus and method for instruction sampling instrumentation
US6397354B1 (en) Method and apparatus for providing external access to signals that are internal to an integrated circuit chip package
US8185772B2 (en) Determining execution times of commands
US4731768A (en) Autoranging time stamp circuit
US7607047B2 (en) Method and system of identifying overlays
US6332117B1 (en) General event stamping scheme
US20060184776A1 (en) Methods to randomly or pseudo-randomly, without bias, select instruction for performance analysis in a microprocessor
US7480590B2 (en) Performance adder for tracking occurrence of events within a circuit
US20070168807A1 (en) Start/stop circuit for performance counter
US7315964B2 (en) Digital signal measuring apparatus and traffic observing method

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): JP KR

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE

121 Ep: the epo has been informed by wipo that ep was designated in this application
DFPE Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101)
WWE Wipo information: entry into national phase

Ref document number: 1998959548

Country of ref document: EP

ENP Entry into the national phase

Ref country code: JP

Ref document number: 2000 547546

Kind code of ref document: A

Format of ref document f/p: F

WWE Wipo information: entry into national phase

Ref document number: 1020007012246

Country of ref document: KR

WWP Wipo information: published in national office

Ref document number: 1998959548

Country of ref document: EP

WWP Wipo information: published in national office

Ref document number: 1020007012246

Country of ref document: KR

WWG Wipo information: grant in national office

Ref document number: 1998959548

Country of ref document: EP

WWW Wipo information: withdrawn in national office

Ref document number: 1020007012246

Country of ref document: KR