CN101114242A

CN101114242A - Bus data searching method

Info

Publication number: CN101114242A
Application number: CNA2006100368061A
Authority: CN
Inventors: 侯全成
Original assignee: Mitac Computer Shunde Ltd; Mitac International Corp
Current assignee: Mitac Computer Shunde Ltd; Shunda Computer Factory Co Ltd; Mitac International Corp
Priority date: 2006-07-28
Filing date: 2006-07-28
Publication date: 2008-01-30
Anticipated expiration: 2026-07-28
Also published as: CN100530114C

Abstract

The invention discloses a bus data search method. A target waveform pattern is searched out from the waveform patterns which are measured and obtained from the bus. The invention comprises following steps: firstly, the waveform patterns are regulated into the same data comparison time interval with the target waveform pattern; the waveform patterns and the target waveform pattern are respectively quantized into data awaiting comparison and simulation data with sampling time points; the difference between the simulation data and the data awaiting comparison of each sampling time point is calculated to obtain a first error value. The simulation data or the data awaiting comparison is regulated to a next calculation reference point, and again the difference between the simulation data and the data awaiting comparison of each sampling time point is calculated to obtain a second error value. When the second error value is smaller than the first error value, the data awaiting comparison corresponding to the second error value is taken out, which is determined to be the target waveform pattern.

Description

Bus data searching method

Ji Intraoperative field

The present invention is about a kind of waveform patterns data search method, particularly a kind of mean square deviation (meansquare error, MSE) bus sequence data method for searching of algorithm of utilizing.

Bei Jing Ji Intraoperative

At present (for example at sequence signal, PCI-express or SATA) detection often judge signal transmission quality by eye pattern (eyediagram), measure by oscillograph and to pick out some particular profile section in a lot of sequence data, cooperate presenting of eye pattern to understand signal transmission quality, and present mode is when seeking certain data blocks, move by artificial mode and to compare the sequence data that measures, judge the most similar waveform patterns with human eye, with the data segments of decision desire searching.

Please refer to the TaiWan, China patent announcement number No. 535421, this patent discloses a kind of new block comparison displacement estimation (Block-matching motion estimation) method, have one group of size two formulas symmetry and search example edition (Search pattern), search example edition (Large cellular searchpattern with large-scale nido earlier, LCSP) do fast searching on a large scale at search window (Search window) intra block, then, search example edition (Small rectangular search pattern with small-sized side's shape, SRSP) accurately search at being decided to contract in the zone to do among a small circle by LCSP, just can obtain the motion vector position.Use the example edition of searching of good definition during because of search, need do the universe search, so can reduce the search number of blocks to search window; Be symmetry because of searching example edition again, thus the block searched between two steps, need do not repeated, more can further reduce the search number of blocks, therefore, just can reach and both reduce search time, can correctly estimate the effect of displacement again.

Though said method provides a kind of special pattern method for searching, can find out the block data that meets example edition fast, can't be applied on the search process of waveform patterns of bus.

Therefore, how to provide a kind of bus data searching method, with fast searching in the data that obtain in measurement with pick out specific data segment, become one of researchist's problem to be solved.

Summary of the invention

In view of above problem, fundamental purpose of the present invention is to provide a kind of bus data searching method, obtain recently by the computing machine calculation like the figure matched position, provide tester's fast searching sequence data thus, with the bus test operation of being correlated with.

Therefore, for reaching above-mentioned purpose, the bus data searching method that the present invention disclosed hunts out the target waveform pattern in the waveform patterns that measurement obtains on bus, include the following step: at first, adjusting waveform patterns and target waveform pattern is identical data comparison time interval; Quantized waveform pattern and target waveform pattern are data to be compared and the emulated data with sampling time point respectively; Calculate the emulated data of each sampling time point and the difference of data to be compared,, adjust emulated data or data to be compared and calculate reference point to next to obtain first error amount; Calculate the emulated data of each sampling time point and the difference of data to be compared once more, to obtain second error amount; And, take out the data to be compared of the second error amount correspondence when second error amount during less than first error amount, be the target waveform pattern to determine data to be compared, and store second error amount in buffer.

Wherein in the difference of the emulated data of calculating each sampling time point once more and data to be compared, after the step that obtains one second error amount, more include and confirm that a time calculating reference point is the final step of calculating reference point, in confirming that an inferior calculating reference point is not when finally calculating reference point, then get back to adjust emulated data or the extremely inferior calculating reference point of data to be compared.

Wherein when second error amount is not less than first error amount, taking out the data to be compared of the first error amount correspondence, is the target waveform pattern to determine data to be compared.

By this bus data searching method, obtain optimum waveform pattern match position by mean square deviation (MSE) algorithm, in metric data, to pick out the data segments that desire is sought, promote the search speed and the accuracy of sequence data thus.

Description of drawings

Fig. 1 is the flow chart of steps of the bus data searching method of the embodiment of the invention;

Fig. 2 A is the bus waveform patterns comparison synoptic diagram of the embodiment of the invention; And

Fig. 2 B is the bus waveform patterns comparison synoptic diagram of the embodiment of the invention.

Embodiment

Please refer to Fig. 1, flow chart of steps for the bus data searching method of the embodiment of the invention, hunt out a target waveform pattern in the waveform patterns that measurement obtains on bus, include the following step: at first, adjusting waveform patterns is an identical data comparison time interval (step 100) with the target waveform pattern, because waveform patterns may be different with the sampling time point of target waveform pattern, so need elder generation that the sampling time point of two waveform patterns is adjusted into unanimity (being identical data comparison time interval), to carry out follow-up data search operation.

Next, respectively quantized waveform pattern and target waveform pattern for data to be compared with sampling time point (for example, 010101100010......) and emulated data is (for example, 1000110101) (step 101), the analog waveform patterns that measurement is got is converted to digital data kenel, the data-switching operation that can adopt analog/digital conversion (ADC) circuit to be correlated with on the practice.

Calculate the emulated data on each sampling time point and the difference of data to be compared, to obtain first error amount (step 102), and store first error amount in the buffer (not shown), wherein first error amount adopts mean square deviation (MSE) calculation formula to obtain, and is as follows:

\sqrt{{(X_{1} - Y_{1})}^{2} + {(X_{2} - Y_{2})}^{2} + . . . {(X_{n} - Y_{n})}^{2}},

X wherein ₁Be first emulated data on first sampling time point, X ₂Be second emulated data on second sampling time point, X _nBe the n emulated data on the n sampling time point, the n value is decided on degree of accuracy and complexity, that is to say that the n value can be adjusted voluntarily and be set by the user, and Y ₁Be first data to be compared on first sampling time point, Y ₂Be second data to be compared on second sampling time point, Y _nBe the n data to be compared on the n sampling time point, the n value is decided on degree of accuracy and complexity, that is to say that the n value can be adjusted voluntarily and be set by the user.

Then, adjust emulated data or data to be compared and calculate reference point (step 103) to next, the aforementioned calculation reference point refers to one of them sampling time point of emulated data or data to be compared is added up 1, in other words, be about to emulated data or data to be compared to next time point translation (shown in Fig. 2 A and Fig. 2 B, being among the figure) to next time point translation with emulated data.

Calculate the emulated data on each sampling time point and the difference of data to be compared once more, to obtain second error amount (step 104), and second error amount of obtaining can be stored in the buffer (not shown), and wherein second error amount adopts mean square deviation (MSE) calculation formula to obtain, and is as follows:

\sqrt{{(X_{1} - Y_{m})}^{2} + {(X_{2} - Y_{m + 1})}^{2} + . . . {(X_{n} - Y_{n + m - 1})}^{2}},

X wherein ₁Be first emulated data on first sampling time point, X ₂Be second emulated data on second sampling time point, X _nBe the n emulated data on the n sampling time point, the n value is decided on degree of accuracy and complexity, that is to say that the n value can be adjusted voluntarily and be set by the user, and Y _mBe the m data to be compared on the m sampling time point of an inferior calculating reference point, Y _M+1Be the m+1 data to be compared on the m+1 sampling time point of an inferior calculating reference point, Y _N+m-1Be that inferior one calculates the n+m-1 data to be compared on the n+m-1 sampling time point of reference point, n value is decided on degree of accuracy and complexity, that is to say that the n value can be adjusted voluntarily and be set by the user, decides and the data that the m value obtains on measurement are total.

Judge whether to calculate reference point for final one? (step 105) then gets back to step 103 when really admitting a fault final one when calculating reference point, continues to adjust emulated data or data to be compared are calculated reference point to next; When confirming as final one when calculating reference point, then then read first error amount in the buffer, to judge that second error amount is less than first error amount? (step 106), wherein when second error amount during less than first error amount, take out the data to be compared of the second error amount correspondence, to determine that these data to be compared are target waveform pattern (step 107), and store second error amount in buffer, when second error amount is not less than first error amount, then taking out the data to be compared of the first error amount correspondence, is the target waveform pattern to determine these data to be compared.

Please refer to Fig. 2 A and Fig. 2 B, it is the bus waveform patterns comparison synoptic diagram of the embodiment of the invention, at first, measurement obtains waveform patterns 10 from the bus (not shown), then, the data comparison time interval 10a of waveform patterns 10 is adjusted into identical size with the data comparison time interval 20a of target waveform pattern 20, the sampling time point of waveform patterns 10 is (t0, t1, t2, t3, t4, t5, t6......tn), and the sampling time point of target waveform pattern 20 is (t0, t1, t2, t3 and t4), carry out error amount calculating after waveform patterns 10 and target waveform pattern 20 quantized respectively, to calculate back pairing waveform patterns 10 sections of minimum error values again and take out, the waveform patterns 10 of this section is the target waveform pattern of desiring to look for 20.

In addition, target waveform pattern 20 can adopt the waveform patterns of corresponding K handle for instance, and the type of K handle includes: K28.5, " COM ", the abbreviation of Comma is in order to the management as transmission channel (lane) and binding initialization (link initialization); K27.7, " STP ", the abbreviation of Star TLP is in order to indicate the beginning of TLP package; K28.2, " SDP ", the abbreviation of Start DLLP is in order to indicate the beginning of DLLP package; K29.7, " END " is in order to indicate the end of TLP package or DLLP package; K30.7, " EDB ", the abbreviation of EnD Bad is in order to indicate the end of " nullified TLP " package; K28.0, " SKP " is when two PORT COM transfer rates are not used simultaneously by way of compensation ... or the like, therefore, the method that discloses by the embodiment of the invention can pick out the certain data blocks of sequence data in the bus fast, avoid manually inspecting cause the problem of identification mistake.

This bus data searching method obtains optimum waveform pattern match position by mean square deviation (MSE) algorithm, to pick out the data segments that desire is sought in metric data, promotes the search speed and the accuracy of sequence data thus.

Claims

1. a bus data searching method hunts out a target waveform pattern in the waveform patterns that measurement obtains on this bus, includes:

Adjusting this waveform patterns and this target waveform pattern is identical data comparison time interval;

Quantize this waveform patterns and this target waveform pattern respectively for having one of sampling time point data to be compared and an emulated data;

Calculate this emulated data of each this sampling time point and the difference of these data to be compared, to obtain one first error amount;

Adjust this emulated data or this data to be compared to a time calculating reference point;

Calculate this emulated data of each this sampling time point and the difference of these data to be compared once more, to obtain one second error amount; And

When this second error amount during less than this first error amount, take out these data to be compared of this second error amount correspondence, be this target waveform pattern with these data to be compared of decision.

2. bus data searching method according to claim 1, it is characterized in that, in the difference of this emulated data of calculating each this sampling time point once more with these data to be compared, after the step that obtains one second error amount, more include this time of affirmation one and calculate the step of reference point for final calculating reference point.

3. as the bus data searching method as described in the claim 2, it is characterized in that, this bus data searching method comprises in addition: confirm that it is not when finally calculating reference point that reference point is calculated in this time, then gets back to this emulated data of adjustment or this data to be compared step to an inferior calculating reference point.

4. bus data searching method according to claim 1, it is characterized in that, this bus data searching method comprises in addition: when this second error amount is not less than this first error amount, take out these data to be compared of this first error amount correspondence, with the step that determines that these data to be compared are this target waveform pattern.

5. bus data searching method according to claim 1 is characterized in that, this time calculated reference point and referred to each this sampling time point of this emulated data is added up 1 one by one.

6. bus data searching method according to claim 1 is characterized in that, in the difference of this emulated data of calculating each this sampling time point with these data to be compared, after the step that obtains one first error amount, stores this first error amount in a buffer.

7. as the bus data searching method as described in the claim 6, it is characterized in that, in the difference of this emulated data of calculating each this sampling time point once more, after the step that obtains one second error amount, read this first error amount in this buffer with these data to be compared.

8. as the bus data searching method as described in the claim 7, it is characterized in that, in when this second error amount during less than this first error amount, take out these data to be compared of this second error amount correspondence, after determining that these data to be compared are the step of this target waveform pattern, store this second error amount in this buffer.

9. bus data searching method according to claim 1 is characterized in that, this first error amount adopts a mean square deviation calculation formula to obtain.

10. bus data searching method according to claim 1 is characterized in that, this second error amount adopts a mean square deviation calculation formula to obtain.