US20120046929A1 - Statistical Design with Importance Sampling Reuse - Google Patents

Statistical Design with Importance Sampling Reuse Download PDF

Info

Publication number
US20120046929A1
US20120046929A1 US12859871 US85987110A US2012046929A1 US 20120046929 A1 US20120046929 A1 US 20120046929A1 US 12859871 US12859871 US 12859871 US 85987110 A US85987110 A US 85987110A US 2012046929 A1 US2012046929 A1 US 2012046929A1
Authority
US
Grant status
Application
Patent type
Prior art keywords
origin
new
samples
importance sampling
gravity
Prior art date
Legal status (The legal status 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 status listed.)
Abandoned
Application number
US12859871
Inventor
Rajiv V. Joshi
Rouwaida N. Kanj
Sani R. Nassif
Carl J. Radens
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
International Business Machines Corp
Original Assignee
International Business Machines Corp
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

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/07Responding to the occurrence of a fault, e.g. fault tolerance
    • G06F11/0703Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
    • G06F11/0751Error or fault detection not based on redundancy
    • G06F11/0754Error or fault detection not based on redundancy by exceeding limits
    • G06F11/076Error or fault detection not based on redundancy by exceeding limits by exceeding a count or rate limit, e.g. word- or bit count limit
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/008Reliability or availability analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/07Responding to the occurrence of a fault, e.g. fault tolerance
    • G06F11/0703Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
    • G06F11/079Root cause analysis, i.e. error or fault diagnosis
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/50Computer-aided design
    • G06F17/5009Computer-aided design using simulation
    • G06F17/5022Logic simulation, e.g. for logic circuit operation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/50Computer-aided design
    • G06F17/5009Computer-aided design using simulation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2217/00Indexing scheme relating to computer aided design [CAD]
    • G06F2217/10Probabilistic or stochastic CAD

Abstract

A mechanism is provided for reusing importance sampling for efficient cell failure rate estimation of process variations and other design considerations. First, the mechanism performs a search across circuit parameters to determine failures with respect to a set of performance variables. For a single failure region, the initial search may be a uniform sampling of the parameter space. Mixture importance sampling (MIS) efficiently may estimate the single failure region. The mechanism then finds a center of gravity for each metric and finds importance samples. Then, for each new origin corresponding to a process variation or other design consideration, the mechanism finds a suitable projection and recomputes new importance sampling (IS) ratios.

Description

    BACKGROUND
  • The present application relates generally to an improved data processing apparatus and method and more specifically to mechanisms for simulation based characterization of circuits using statistical design with importance sampling reuse.
  • As memory array architectures are pushed to their practical limits by increasing requirements for density and speed, accurately estimating the cell failure rate of a design becomes increasingly critical. Since a finite number of redundant rows and/or columns is available to replace those containing defective cells, a number of failed cells above this level of redundancy will yield a defective device. The number of defective devices, or device yield is then directly related to the cell failure rate. The larger arrays being fabricated today have increasingly stringent failure rate control requirements. For example, in order to achieve a yield of 90% in a one-million cell array without redundancy, a failure rate below five standard deviations (5σ) must be held.
  • Traditional techniques such as Monte-Carlo analysis produce accurate results at a cost of a large number of iterations, due to the random sampling of the entire probability space of the independent variables that are treated in the analysis. As the cell failure rate decreases, the number of samples and iterations required for accurate analysis becomes increasingly large, because of the relatively sparse distribution of samples in the distribution tail(s) that correspond to failed cells. The effect of circuit changes on cell readability and writability, as well as minimum read and write cycle times and margins, are difficult to estimate at very low failure rate levels. Such low failure rates cause further complications for adjusting designs to achieve the best result.
  • Techniques other than Monte-Carlo analysis have been implemented for estimating cell failure rates, each with related drawbacks. Sensitivity analysis is a well-known technique in which the gradients of the various independent variables are used to determine the bounds of the non-failure confidence region. However, accurate estimates of the failure rate are not typically produced by sensitivity analysis, as sensitivity analysis by its very nature cannot determine the exact overlapping impact of all independent variables on the cell failure rate at once. Another technique that can accurately estimate the failure rate is the grid analysis approach, in which the grid size can be made arbitrarily small. However, the number of simulations increases exponentially with the number of independent variables and typically a large amount of custom coded program control (scripting) must be employed to direct the analysis.
  • SUMMARY
  • In one illustrative embodiment, a method, in a data processing system, is provided for determining failure rate of a device using importance sampling reuse. The method comprises performing, by the data processing system, a uniform sampling over a random sample space for a metric for the device with respect to an origin to form a set of samples. The set of samples comprises one or more failing samples. The method further comprises determining, by the data processing system, a center of gravity of the one or more failing samples with respect to the origin and determining, by the data processing system, importance samples based on the center of gravity of the one or more failing samples. The method further comprises selecting a new origin, recomputing, by the data processing system, new importance sampling weight ratios for the new origin, and determining, by the data processing system, a failure rate for the device based on the new importance sampling weight ratios for the new origin.
  • In other illustrative embodiments, a computer program product comprising a computer usable or readable medium having a computer readable program is provided. The computer readable program, when executed on a computing device, causes the computing device to perform various ones, and combinations of, the operations outlined above with regard to the method illustrative embodiment.
  • In yet another illustrative embodiment, a system/apparatus is provided. The system/apparatus may comprise one or more processors and a memory coupled to the one or more processors. The memory may comprise instructions which, when executed by the one or more processors, cause the one or more processors to perform various ones, and combinations of, the operations outlined above with regard to the method illustrative embodiment.
  • These and other features and advantages of the present invention will be described in, or will become apparent to those of ordinary skill in the art in view of, the following detailed description of the example embodiments of the present invention.
  • BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
  • The invention, as well as a preferred mode of use and further objectives and advantages thereof, will best be understood by reference to the following detailed description of illustrative embodiments when read in conjunction with the accompanying drawings, wherein:
  • FIG. 1 is a schematic diagram of a memory cell within an array of memory cells that can be modeled in accordance with an illustrative embodiment;
  • FIGS. 2A-2B and 3A-3B are graphs illustrating techniques employed in illustrative embodiments;
  • FIG. 4 is a flowchart illustrating operation of a centroid-locating technique in accordance with an illustrative embodiment;
  • FIG. 5 is a flowchart illustrating operation of a random direction vector technique in accordance with an illustrative embodiment;
  • FIG. 6 is a graph depicting global variability space in accordance with an illustrative embodiment;
  • FIGS. 7A and 7B are graphs illustrating global and local variation in two dimensions in accordance with an illustrative embodiment;
  • FIG. 8 illustrates how reusing the old set of samples for a new origin may underestimate fails in accordance with an illustrative embodiment;
  • FIGS. 9A and 9B illustrate alternative approaches to determining fails for a new origin in accordance with illustrative embodiments;
  • FIG. 10 illustrates an example application to importance sampling in accordance with an example embodiment;
  • FIG. 11 illustrates an example application for importance samples for multiple metrics in accordance with an example embodiment;
  • FIG. 12 is a graph illustrating importance sampling reuse for a set of process variations in accordance with an illustrative embodiment;
  • FIG. 13 illustrates an example application of importance sampling reuse for manufacturing variability in accordance with an example embodiment;
  • FIG. 14 is a graph illustrating the manufacturing variability space in accordance with an example embodiment;
  • FIG. 15 is a graph illustrating results for an example static random access memory in accordance with an illustrative embodiment;
  • FIG. 16 is a flowchart illustrating operation of a mechanism for statistical design with importance sampling reuse in accordance with an illustrative embodiment; and
  • FIG. 17 is a block diagram of an example data processing system in which aspects of the illustrative embodiments may be implemented.
  • DETAILED DESCRIPTION
  • The illustrative embodiments provide a mechanism for reusing importance sampling for efficient cell failure rate estimation of process variations and other design considerations. First, the mechanism performs a search across circuit parameters to determine failures with respect to a set of performance variables. For a single failure region, the initial search may be a uniform sampling of the parameter space. Mixture importance sampling (MIS) efficiently may estimate the single failure region. The mechanism then finds a center of gravity for each metric/region and finds importance samples. Then, for each new origin corresponding to a process variation or other design consideration, the mechanism finds a suitable projection and recomputes new importance sampling (IS) ratios.
  • Importance sampling is a general technique for estimating properties of a particular distribution, while only having samples generated from a different distribution rather than the distribution of interest. Depending on the application, the term may refer to the process of sampling from this alternative distribution, the process of inference, or both. The idea behind importance sampling is that certain values of the input random variables in a simulation have more impact on the metric being estimated than others. If these “important” values are emphasized by sampling more frequently, then the estimator variance can be reduced. Hence, the basic methodology in importance sampling is to choose a distribution which encourages the important values. This use of biased distributions will result in a biased estimator if it is applied directly in the simulation. However, the simulation outputs are weighted to correct for the use of the biased distribution, and this ensures that the new importance sampling estimator is unbiased.
  • The fundamental issue in implementing importance sampling simulation is the choice of the biased distribution which encourages the important regions of the input variables. Choosing or designing a good biased distribution is the art of importance sampling. The rewards for a good distribution can be huge run-time savings; the penalty for a bad distribution can be longer run times than for a general Monte-Carlo simulation without importance sampling.
  • More particularly, the mechanism of the illustrative embodiments may first perform uniform sampling until a predetermined number of failing samples are encountered. Then, the mechanism may perform importance sampling simulation to determine a biased (distorted) distribution for performing a second sampling. The mechanism may then weight the results to determine failure rate or yield information. Finally, in accordance with an illustrative embodiment, the mechanism may reuse the samples from the biased distribution with respect to a new origin to determine the yield for the new origin. In an alternative embodiment, the mechanism may reuse the samples from the uniform sampling to determine a new biased distribution for the new origin and then weight the results to determine the yield for the new origin.
  • Thus, the illustrative embodiments concern techniques for overcoming the limitations of traditional Monte-Carlo analysis for circuits where the failure rate of the circuit being analyzed is very low. For instance, with respect to circuits having large arrays of identical cells, the cells are generally the determining factor in the failure rate, but only as a totality of the cells. Because millions of cells may be incorporated in a memory array, even very low failure rates contribute significantly to the failure rate of the individual memory devices or other devices that incorporate memory such arrays.
  • It is necessary to analyze the cell design and process variations at the extreme end of the distribution of actual cell parameters in order to gain meaningful information that can accurately predict device yields and permit improvement of the cells in order to improve device yields. The techniques used in the prior art either require exhaustive computation and storage, or do not perform well once there are more than a small number of variable design and process parameters, such as device areas or lengths and widths, doping densities, threshold voltages and other measures of design and process parameters. The illustrative embodiments provide a mechanism for using mixture importance sampling (MIS), which is a known technique, in a manner that effectively models memory cells.
  • MIS uses a mixture of two or more distributions for generating sample values, with at least one of the distributions biased to generate samples in a region of interest. In the illustrative embodiments the sample values are input vectors of design- and process-dependent electrical circuit parameters that determine the performance of the memory cell. The performance is measured in terms of operational performance values such as read and write delay time, writability (i.e., can the cell value be changed during a write) and read stability (i.e., will the cell hold its value during reads), as well as the margins associated with the read and write times, under the operational and circuit parametric conditions simulated. Because a priori knowledge about what regions in N-space might be of interest is not generally easily obtained (where N is the number of variable parameters), the illustrative embodiments provide a front-end mechanism to the MIS analysis that identifies and quantifies a particular region or regions of interest for further study via the MIS technique. The result is that the computational overhead and storage associated with the analysis is greatly reduced, while yielding the desired accuracy with respect to the failure mechanism(s) being studied.
  • In order to locate a region of interest that provides information about a particular failure mechanism, a priori information about the failure mechanisms to expect is useful. If it is known that a single dominant failure mode is present, such as in many static random access memory (SRAM) cell designs, then the identification of the region of interest is simplified. If it is known that multiple regions of interest having statistically significant impact will be present, then the analysis can proceed until all regions of interest are identified. If it is not known how many failure mechanisms will be of interest and/or significant within the probability space to be explored, then techniques must be employed that can handle either the single-region or multiple region cases.
  • In the multiple region case, there are also two alternatives for applying the MIS technique. Either the regions of interest can be studied independently, or the sampling function for the MIS technique can include multiple sampling centers concentrated on the approximate centers of the multiple regions. The latter technique can be used to avoid error in failure rate prediction due to overlap of the regions of interest in one or more dimensions.
  • With reference now to the figures, and in particular with reference to FIG. 1, an example memory cell that can be modeled by a method in accordance with an illustrative embodiment is shown in an array that includes other cells 5. Transistors P10, N10, P11, and N11 form a cross-coupled static latch that provides the storage of a value in the cell. Transistors N12 and N13 provide for access to the value in response to a wordline select signal WL. Bitlines BLT (true bitline) and BLC (complement bitline) couple all cells in a column, so that when a row is selected by signal WL, only one row cell from each column is exposed to the memory logic. For a write operation, bitlines BLC and BLT are charged to voltages corresponding to the desired state of the memory cell and WL is activated (pulsed), setting the state of the latch formed by transistors P10, N10, P11 and N11. For a read operation, the bitlines BLC and BLT are previously charged to opposite state predetermined voltages (generally Vdd and ground), and to commence the read, WL is pulsed and a sense amplifier coupled to bitlines BLC and BLT determines the stored state by differential comparison of bitlines BLC and BLT. Depending on the relative strengths of the transistors P10-11 and N10-13, the cell will exhibit varying ability to perform to predetermined read/write cycle times and may be unstable in that the cell value does not remain constant after a write or when being read. As operating frequencies are increased and device sizes correspondingly decreased, the variations take on a statistically significantly greater range causing failure of an increasing number of devices in a lot. The illustrative embodiments are directed toward an efficient method for statistically analyzing the design of memory cells so that the yield of memory cells can be predicted accurately and further so that yields may be improved by selecting optimized nominal values for the device parameters and other environmental parameters such as operating voltages.
  • While the illustrated cell is an example of a cell of order 4 that may be analyzed and improved by a method according to an embodiment of the invention, it should be understood that the techniques illustrated herein may be applied to memory cells of any order and design and to circuits other than memory circuits, as well. (Order as used herein refers to the number of devices that implement the storage element of the cell exclusive of the bitline access transistors.)
  • Referring now to FIG. 2A, a graph illustrating the purpose and techniques of the illustrative embodiments is shown. A distribution 10 of operational performance values, such as the above-mentioned delay times and writability/read stability measures, generally following a Gaussian shape extends past 5 standard deviations (5σ) on either side of a mean value. The input parameter values are generated by a similar Gaussian distribution of samples as generated around a nominal vector of cell parameter values by a Monte-Carlo algorithm.
  • Failure zone 12A in the graph is located past the 5σ point and is shown as a shaded area. The yield of the cell modeled by distribution 10 can be predicted from the graph, and thus the yield of the overall device. However, the accuracy near failure zone 12A is limited due to the relatively sparse distribution of samples in the tails of distribution 10. The illustrative embodiments use MIS to concentrate sampling within one or more failure zones, so that more accurate estimates of yield are produced.
  • Referring now to FIG. 2B, a graph showing the operation of the illustrative embodiments is shown. Once a region of interest is identified in the device and process-dependent parameter space, a mixture 10C of distributions 10, 10A, and 10B is employed to improve the density of samples in zone 12A, while ensuring that gaps are not left in the distribution that might otherwise miss other failures. The mixture sampling function distribution 10C can be expressed by:

  • g λ(x)=λ1 p(x)+λ2 U(x)+(1−λ1−λ2)p(x−μs),
  • where λ1 and λ2 are coefficients used to control the mixture, which can be determined by the position of a new sampling function center μs 14 that is used to improve the concentration of samples around μs. Note that μs is not the center of mixture sampling function distribution 10C, but rather the center of gaussian distribution 10A forming part of sampling function distribution 10C. Uniform distribution U(x) 10B is also included in the mixture, which helps in ensuring that some samples are present for all values within the interval over which uniform distribution 10B extends. The choice of coefficients, in combination with the inclusion of the uniform distribution is made so that the number of samples in the region of interest is increased, but no “dead spots” are present in the analysis.
  • The result of the mixture sampling function is to generate a relatively larger number of samples over zone 12A (as compared to the distribution of FIG. 2A), yielding a much more precise estimate of the impact of zone 12A on. Results show a more than 10× improvement (or greater, in proportion to the distance of zone 12A from the nominal mean μ) in convergence time for estimates of the edges of failure zone 12A, which results in a corresponding improvement in convergence of the calculation of overall device yield.
  • While generally the statistical analysis detailed above will be conducted independently over the operational performance variables being studied, it is possible to conduct a combined pass/fail analysis over the parameter space in which no information about the particular operational performance variables associated with each failed point is retained, but the overall desirability of a particular design can be directly observed with respect to process variations. MIS analysis can then be conducted with one or more mean-shifting distributions included to precisely predict the yield.
  • Referring now to FIG. 3A, a first technique for locating sampling function center μs 14, is depicted. Since memory cell operational performance variable failure regions generally will extend to the bounds of the analysis range (with respect to the parameter deviation) from the point at which the failures occur, failure regions can be ascribed to failure region boundaries beyond which the probability of failure only increases. The illustration is two-dimensional but in actual practice, a larger number of dimensions are actually being evaluated simultaneously. A number of sample points are generated by uniformly sampling the variable memory cell parameter space, corresponding to the dots shown in the depicted example. Quasi-random sampling techniques can be used to improve the spread of the samples across the parameter space. In the depicted example, hollow dots depict non-failing points and solid dots depict failing points. After a threshold number of failures is accumulated, which may be qualified by observing their proximity in the parameter space if multiple failure regions might be present (in order to group the failures), a particular failure region of interest 20B is identified and the vector centroid 14 of that region computed and used as new sampling function center μs 14 for subsequent MIS analysis. The vector centroid is computed from the vector distances of the points as the vector of average distance in each parameter space. There may be multiple failure regions for some applications of the techniques of the illustrative embodiments including other failure regions such as 20A, and those regions can be ignored if not of interest, or may be further evaluated with independent MIS analysis around their centers.
  • The above-described technique is especially applicable to the study of memory cell designs that have a single dominant failure region of interest. However, if it is not known in advance that there will be a single dominant region, the positions of the failure samples in parameter vector space can be observed and the samples grouped into one or more groups as mentioned above. If a group is much more distant from the nominal parameter vector, then that group may be discarded as being due to a relatively unimportant failure mechanism. The technique of the illustrative embodiments can be used to obtain better information about multiple failure mechanisms by the above grouping technique or discarding of groups.
  • A threshold number of samples can then be collected for each group to be studied and either an independent set of MIS analyses can be conducted for each group, or the above MIS sampling distribution function can be modified to follow:

  • g λ(x)=λ1 p(x)+λ2 U(x)+λ3 p(x−μ s1)+(1−λ1−λ2−λ3)p(x−μs2),
  • where λ1, λ2 and λ3 are coefficients used to control the mixture and new sampling function centers μs1, μs2 are used to improve the concentration of samples around two regions of interest. If more than two regions of interest are present, the above sampling function can be expanded to include other mean shifting values and their associated sampling function kernels in the above sampling function.
  • Referring now to FIG. 3B, another technique adapted for treating circuits having multiple dominant failure regions is depicted, e.g., circuits for which multiple failure mechanisms have fairly equivalent statistical significance. The technique may also be used for single failure region analysis, as well. A vector of nominal memory cell parameter values 16 is used as a starting point for random generation of a sufficient initial number of vector directions 18 along which cell failure analysis proceeds until failure points are detected. A sufficient number of vectors is generated so that failure regions will not be missed by the analysis. Gaussian latin hypercube sampling can be used to ensure uniform placement of the vectors in all directions.
  • The analysis then proceeds away from nominal vector 16 as illustrated. The generally monotonic behavior of the circuits as the parameters vary in one direction away from the nominal ensures that failing points will only be encountered at and beyond the boundaries of the failure regions 20A, 20B. The new sampling function center μs 14 for subsequent MIS analysis is determined either by the mean vector 14B corresponding to the group of boundary points 22, or an estimated vector 14A is extrapolated within failure region of interest 20B from the location of the boundary points. After a first iteration, if the boundaries of failure regions 20A, 20B are not sufficiently defined, a local set of random vectors is generated to enhance the set of samples around the boundaries. After boundaries 20A, 20B are sufficiently defined μs 14 is chosen as described above. As in the center of gravity technique, regions of interest that are more distant from the nominal vector can be discarded as relatively unimportant failure mechanisms. The technique illustrated in FIG. 3B also provides better information about multiple failure mechanisms by locating the boundaries of the regions of interest. If the boundaries are not well defined after the above analysis has been attempted, then the entire process can be repeated with a larger set of initial random vectors.
  • FIG. 4 is a flowchart illustrating operation of a centroid-locating technique in accordance with an illustrative embodiment. The flowchart of FIG. 4 corresponds to the centroid-locating technique described above with reference to FIG. 3A. First, a statistical analysis across ranges of memory circuit parameters is performed for multiple performance variables (block 30) and once a threshold number of failing samples is accumulated (possibly qualified as being in a single failure region if required) (block 32) the centroid of the parameters corresponding to the group of failures is determined (block 34). Finally, a new distribution of input parameter vectors is selected in conformity with the determined centroid and the nominal vector (block 36) and an MIS analysis is performed (block 38) using the new mixture distribution to gain more accurate results in the region of the failures and a faster convergence of the overall yield computation, and operation ends.
  • FIG. 5 is a flowchart illustrating operation of a random vector technique in accordance with an illustrative embodiment. The flowchart of FIG. 5 corresponds to the random vector technique described above with reference to FIG. 3B. First, random vector directions are generated around the nominal value of the memory circuit parameters (block 50). The memory cell is simulated along the vectors until a failure is detected, corresponding to a failure region boundary (block 52). Then, a mean point is computed from multiple failure region boundary points or a point is estimated within the failure region from the boundary points (block 54). An MIS distribution function is then selected (block 56) as a mixture of a distribution around the nominal parameter values and the mean point computed in block 54 (along with the uniform distribution), and an MIS analysis is performed (block 58) using the new mixture distribution to gain more accurate results in the region of the failures and a faster convergence of the overall yield computation, and operation ends.
  • Variability is a key problem in circuit design. Statistical analysis and simulations are needed to analyze design yield. For this, importance sampling was developed as a fast statistical analysis tool versus regular Monte-Carlo analysis. Often, circuit designers face situations where they revisit a circuit design with multiple design considerations. For example, design centering, optimization, and manufacturing variability require a lot of what-if statistical analysis studies.
  • For the design centering and manufacturing variability space, often global variation space is a subset of the random sample space. In accordance with an illustrative embodiment, a mechanism is provided to minimize function evaluations (simulations) and to reuse importance samples. For different global variation corners, the mechanism may infer yield information from samples obtained around the old origin.
  • FIG. 6 is a graph depicting global variability space in accordance with an illustrative embodiment. The disk represents the global variability space defined by the following:

  • x=−z,
  • where x, y, and z are random variables. The mechanism of the illustrative embodiments may find importance samples for origin 602 and may recomputed new importance sample ratios for new origins 604, which are “corners” on the boundary of the global variability space.
  • FIGS. 7A and 7B are graphs illustrating global and local variation in two dimensions in accordance with an illustrative embodiment. More particularly, FIG. 7A illustrates failure direction with respect to the origin by regular sampling. FIG. 7B illustrates the operation of using the old set of samples with respect to a new origin. As seen in FIG. 7B, simply reusing the old set of samples may result in underestimating fails.
  • FIG. 8 illustrates how reusing the old set of samples for a new origin may underestimate fails in accordance with an illustrative embodiment. The distance of the new origin to the failing samples (e.g., the center of gravity (COG)) can underestimate the fails, because the distance along the COG direction 802 is greater than the distance 804 of the new origin to the failing samples along the same direction of the origin to the COG.
  • FIGS. 9A and 9B illustrate alternative approaches to determining fails for a new origin in accordance with illustrative embodiments. As shown in FIG. 9A, one approach is to project the new origin onto the line from the origin to the COG. This approach maintains failing samples close to the projected new origin. As shown in FIG. 9B, another approach is to move the population of samples, along the orthogonal direction to the line from the origin to the COG. This approach approximates the dominant fail direction in the proximity of the old origin via the COG of fails. The approach of FIG. 9B provides a fixed pattern of fails without enforcing or modeling using a fixed boundary. This approach is particularly reasonable when the new origin is not too far from the origin.
  • FIG. 10 illustrates an example application to importance sampling in accordance with an example embodiment. In the depicted example, area 1002 shows the real importance samples with respect to the origin. Area 1004 shows the shifted importance samples with respect to the new origin. FIG. 11 illustrates an example application for importance samples for multiple metrics in accordance with an example embodiment. In the example shown in FIG. 11, each metric is identified by a dominant center of gravity of fails.
  • In accordance with the illustrative embodiments, the mechanism re-evaluates importance sampling with respect to the new projected origin. The importance sampling weight function of x with respect to the origin is defined as follows:
  • w ( x ) = Πexp ( - 0.5 * ( x - 0 σ ) 2 ) / exp ( - 0.5 * ( x - x COG σ ) 2 )
  • where x is process variation variable of the device, xCOG is the center of gravity of the one or more failing samples, and σ is the standard deviation of x. The importance sampling weight function of x with respect to the new projected origin (np) is defined as follows:
  • w ( x ) = Πexp ( - 0.5 * ( x - x np σ ) 2 ) / exp ( - 0.5 * ( x - x COG σ ) 2 )
  • where x is process variation variable of the device, xnp is a new point of the projected origin for the process variation variable x, xCOG is the center of gravity of the one or more failing samples, and σ is the standard deviation of x; the equations assume that the process variation variables are Gaussian. In case of non-gaussian variables, transformation are used to normalize the process variations via cumulative distribution functions mapping tables, or more complex weight functions can be adopted. Most importantly if the probability density function is f(x), then the shifted weights are proportional to f(x−xnp)/f(x−xcog). The equations above also assume that the standard deviation, σ, of the process variation variables remain the same. In case there is a change in σ associated with origin of process variation change, then the weights are modified accordingly.
  • w ( x ) = Π σ σ np exp ( - 0.5 ( x - x np σ np ) 2 ) / exp ( - 0.5 ( x - x COG σ ) 2 )
  • where x is process variation variable of the device, xnp is a new point of the projected origin for the process variation variable x, xCOG is the center of gravity of the one or more failing samples, σ is the standard deviation of x, and σnp is the standard deviation associated with the new point of origin.
  • FIG. 12 is a graph illustrating importance sampling reuse for a set of process variations in accordance with an illustrative embodiment. In the example shown in FIG. 12, there are three variables, in this instance for a pass gate (pg), a pull-up transistor (pu), and a pull-down transistor (pd). The mechanism of the illustrative embodiments may perform MIS analysis at an origin and predict yield at different corners 1202 with no additional simulations.
  • Thus, the mechanism of the illustrative embodiments may be applied reuse importance sampling to determine yield given manufacturing variability. The mechanism may estimate yield at various corners. Designers currently study multiple corner combinations using conventional methods. For example, designers may maintain a table of a hundred or more vdd/vcc points for which yield is being computed. The mechanism of the illustrative embodiments allow designers to more easily and accurately estimate yield gradients and sensitivity to manufacturing variability space and to even estimate weighted equivalent yield within the manufacturing variability space at each of these points with minimum computation/simulation effort.
  • FIG. 13 illustrates an example application of importance sampling reuse for manufacturing variability in accordance with an example embodiment. The mechanism may perform importance sampling reuse for various corners 1302 within the manufacturing variability space. The standard deviation within the manufacturing variability space (σmv) may be much less than the standard deviation for random dopant fluctuation (σRDF). The small standard deviation helps in terms of better accuracy. The corners 1302 may be defined by vt changes in two-dimensional or three-dimensional space. Importance sampling may be performed in six-dimensional space.
  • FIG. 14 is a graph illustrating the manufacturing variability space in accordance with an example embodiment. The mechanism of the illustrative embodiment may estimate yield for each corner with no additional simulations. The mechanism may reuse samples from “nominal” corner analysis rather than one origin. The example shown in FIG. 14 assumes three-dimensional importance sampling and two-dimensional global variability space.
  • The mechanism of the illustrative embodiments may be applied to design centering. The mechanism may determine yield as vt centers of design are moved around. This may enable identifying optimal design centers. For instance, the mechanism of the illustrative embodiments may be part of an adaptive search scheme (or hybrid), which is good to guide gradient, to provide initial solution, etc. True sampling could be performed to certify the yield determined using the importance sampling reuse. In another example embodiment, the mechanism of the illustrative embodiments may be used to determine yield for negative biased temperature instability (NBTI) or positive biased temperature instability (PBTI).
  • FIG. 15 is a graph illustrating results for an example static random access memory in accordance with an illustrative embodiment. As can be seen, the writability and stability for importance sampling reuse are a good approximation for the true estimate. The estimated bounds are large enough for manufacturing variability.
  • As will be appreciated by one skilled in the art, the present invention may be embodied as a system, method, or computer program product. Accordingly, aspects of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, micro-code, etc.) or an embodiment combining software and hardware aspects that may all generally be referred to herein as a “circuit,” “module” or “system.” Furthermore, aspects of the present invention may take the form of a computer program product embodied in any one or more computer readable medium(s) having computer usable program code embodied thereon.
  • Any combination of one or more computer readable medium(s) may be utilized. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CDROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.
  • A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in a baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
  • Computer code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, radio frequency (RF), etc., or any suitable combination thereof.
  • Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java™, Smalltalk™, C++, or the like, and conventional procedural programming languages, such as the “C” programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).
  • Aspects of the present invention are described below with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to the illustrative embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
  • These computer program instructions may also be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other devices to function in a particular manner, such that the instructions stored in the computer readable medium produce an article of manufacture including instructions that implement the function/act specified in the flowchart and/or block diagram block or blocks.
  • The computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus, or other devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
  • FIG. 16 is a flowchart illustrating operation of a mechanism for statistical design with importance sampling reuse in accordance with an illustrative embodiment. Operation begins, and the mechanism performs uniform sampling (block 1602). The mechanism finds a center of gravity (COG) for each metric (block 1604). Then, the mechanism finds importance samples (block 1606). For each new origin, the mechanism finds a suitable projection (bock 1608) and recomputes new importance sampling ratios (block 1610). Thereafter, operation ends.
  • In an alternative embodiment, in block 1610, the mechanism may project a new center of gravity with respect to the new origin. In another embodiment, the mechanism may perform a second stage of importance sampling given the same uniform sampling from block 1602.
  • The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
  • The illustrative embodiments may be utilized in many different types of data processing environments including a distributed data processing environment, a single data processing device, or the like. In order to provide a context for the description of the specific elements and functionality of the illustrative embodiments, FIG. 17 is provided hereafter as an example environment in which aspects of the illustrative embodiments may be implemented. While the description of FIG. 17 will focus primarily on a single data processing device implementation, this is only an example and is not intended to state or imply any limitation with regard to the features of the present invention. To the contrary, the illustrative embodiments are intended to include distributed data processing environments and embodiments.
  • FIG. 17 is a block diagram of an example data processing system in which aspects of the illustrative embodiments may be implemented. Data processing system 1700 is an example of a computer in which computer usable code or instructions implementing the processes for illustrative embodiments of the present invention may be located.
  • In the depicted example, data processing system 1700 employs a hub architecture including north bridge and memory controller hub (NB/MCH) 1702 and south bridge and input/output (I/O) controller hub (SB/ICH) 1704. Processing unit 1706, main memory 1708, and graphics processor 1710 are connected to NB/MCH 1702. Graphics processor 1710 may be connected to NB/MCH 1702 through an accelerated graphics port (AGP).
  • In the depicted example, local area network (LAN) adapter 1712 connects to SB/ICH 1704. Audio adapter 1716, keyboard and mouse adapter 1720, modem 1722, read only memory (ROM) 1724, hard disk drive (HDD) 1726, CD-ROM drive 1730, universal serial bus (USB) ports and other communication ports 1732, and PCI/PCIe devices 1734 connect to SB/ICH 1704 through bus 1738 and bus 1740. PCI/PCIe devices may include, for example, Ethernet adapters, add-in cards, and PC cards for notebook computers. PCI uses a card bus controller, while PCIe does not. ROM 1724 may be, for example, a flash basic input/output system (BIOS).
  • HDD 1726 and CD-ROM drive 1730 connect to SB/ICH 1704 through bus 1740. HDD 1726 and CD-ROM drive 1730 may use, for example, an integrated drive electronics (IDE) or serial advanced technology attachment (SATA) interface. Super I/O (SIO) device 1736 may be connected to SB/ICH 1704.
  • An operating system runs on processing unit 1706. The operating system coordinates and provides control of various components within the data processing system 1700 in FIG. 17. As a client, the operating system may be a commercially available operating system such as Microsoft® Windows® XP (Microsoft and Windows are trademarks of Microsoft Corporation in the United States, other countries, or both). An object-oriented programming system, such as the Java™ programming system, may run in conjunction with the operating system and provides calls to the operating system from Java™ programs or applications executing on data processing system 1700 (Java is a trademark of Sun Microsystems, Inc. in the United States, other countries, or both).
  • As a server, data processing system 1700 may be, for example, an IBM® eServer™ System p® computer system, running the Advanced Interactive Executive (AIX®) operating system or the LINUX® operating system (eServer, System p, and AIX are trademarks of International Business Machines Corporation in the United States, other countries, or both while LINUX is a trademark of Linus Torvalds in the United States, other countries, or both). Data processing system 1700 may be a symmetric multiprocessor (SMP) system including a plurality of processors in processing unit 1706. Alternatively, a single processor system may be employed.
  • Instructions for the operating system, the object-oriented programming system, and applications or programs are located on storage devices, such as HDD 1726, and may be loaded into main memory 1708 for execution by processing unit 1706. The processes for illustrative embodiments of the present invention may be performed by processing unit 1706 using computer usable program code, which may be located in a memory such as, for example, main memory 1708, ROM 1724, or in one or more peripheral devices 1726 and 1730, for example.
  • A bus system, such as bus 1738 or bus 1740 as shown in FIG. 17, may be comprised of one or more buses. Of course, the bus system may be implemented using any type of communication fabric or architecture that provides for a transfer of data between different components or devices attached to the fabric or architecture. A communication unit, such as modem 1722 or network adapter 1712 of FIG. 17, may include one or more devices used to transmit and receive data. A memory may be, for example, main memory 1708, ROM 1724, or a cache such as found in NB/MCH 1702 in FIG. 17.
  • Those of ordinary skill in the art will appreciate that the hardware in FIG. 17 may vary depending on the implementation. Other internal hardware or peripheral devices, such as flash memory, equivalent non-volatile memory, or optical disk drives and the like, may be used in addition to or in place of the hardware depicted in FIG. 17. Also, the processes of the illustrative embodiments may be applied to a multiprocessor data processing system, other than the SMP system mentioned previously, without departing from the spirit and scope of the present invention.
  • Moreover, the data processing system 1700 may take the form of any of a number of different data processing systems including client computing devices, server computing devices, a tablet computer, laptop computer, telephone or other communication device, a personal digital assistant (PDA), or the like. In some illustrative examples, data processing system 1700 may be a portable computing device which is configured with flash memory to provide non-volatile memory for storing operating system files and/or user-generated data, for example. Essentially, data processing system 1700 may be any known or later developed data processing system without architectural limitation.
  • Thus, the illustrative embodiments provide a mechanism for reusing importance sampling for efficient cell failure rate estimation of process variations and other design considerations. First, the mechanism performs a search across circuit parameters to determine failures with respect to a set of performance variables. For a single failure region, the initial search may be a uniform sampling of the parameter space. Mixture importance sampling (MIS) efficiently may estimate the single failure region. The mechanism then finds a center of gravity for each metric and finds importance samples. Then, for each new origin corresponding to a process variation or other design consideration, the mechanism finds a suitable projection and recomputes new importance sampling (IS) ratios.
  • As noted above, it should be appreciated that the illustrative embodiments may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment containing both hardware and software elements. In one example embodiment, the mechanisms of the illustrative embodiments are implemented in software or program code, which includes but is not limited to firmware, resident software, microcode, etc.
  • A data processing system suitable for storing and/or executing program code will include at least one processor coupled directly or indirectly to memory elements through a system bus. The memory elements can include local memory employed during actual execution of the program code, bulk storage, and cache memories which provide temporary storage of at least some program code in order to reduce the number of times code must be retrieved from bulk storage during execution.
  • Input/output or I/O devices (including but not limited to keyboards, displays, pointing devices, etc.) can be coupled to the system either directly or through intervening I/O controllers. Network adapters may also be coupled to the system to enable the data processing system to become coupled to other data processing systems or remote printers or storage devices through intervening private or public networks. Modems, cable modems and Ethernet cards are just a few of the currently available types of network adapters.
  • The description of the present invention has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiment was chosen and described in order to best explain the principles of the invention, the practical application, and to enable others of ordinary skill in the art to understand the invention for various embodiments with various modifications as are suited to the particular use contemplated.

Claims (25)

    What is claimed is:
  1. 1. A method, in a data processing system, for determining failure rate of a device using importance sampling reuse, the method comprising:
    performing, by the data processing system, a uniform sampling over a random sample space for a metric for the device with respect to an origin to form a set of samples, wherein the set of samples comprises one or more failing samples;
    determining, by the data processing system, a center of gravity of the one or more failing samples with respect to the origin;
    determining, by the data processing system, importance samples based on the center of gravity of the one or more failing samples;
    selecting a new origin; recomputing, by the data processing system, new importance sampling weight ratios for the new origin; and
    determining, by the data processing system, a failure rate for the device based on the new importance sampling weight ratios for the new origin.
  2. 2. The method of claim 1, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    finding a projected origin; and
    recomputing new importance sampling weight ratios with respect to the projected origin.
  3. 3. The method of claim 2, wherein finding the projected origin comprises:
    determining a line passing through the origin and the center of gravity of the one or more failing samples; and
    projecting the new origin onto the line passing through the origin and the center of gravity of the one or more failing samples.
  4. 4. The method of claim 1, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    finding a set of projected samples with respect to the new origin; and
    recomputing new importance sampling ratios based on the projected samples.
  5. 5. The method of claim 4, wherein finding the set of projected samples comprises:
    determining a line passing through the origin and the center of gravity of the one or more failing samples; and
    move the set of samples in a direction orthogonal to the line passing through the origin and the center of gravity of the one or more failing samples.
  6. 6. The method of claim 1, wherein recomputing new importance sampling weight ratios for the new origin comprises computing a weight function, wherein the weight function is as follows:
    w ( x ) = Π σ σ np exp ( - 0.5 ( x - x np σ np ) 2 ) / exp ( - 0.5 ( x - x COG σ ) 2 )
    where x is process variation variable of the device, xnp is a new point of a projected origin for the process variation variable x, xCOG is the center of gravity of the one or more failing samples, σ is the standard deviation of x, and σnp is the standard deviation associated with the new point of projected origin.
  7. 7. The method of claim 1, further comprising:
    repeating selecting a new origin, recomputing new importance sampling weight ratios for the new origin, and determining a failure rate for the device based on the new importance sampling weight ratios for the new origin for a set of process variations.
  8. 8. The method of claim 1, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    determining importance samples based on the center of gravity of the one or more failing samples and the new origin.
  9. 9. A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to:
    perform a uniform sampling over a random sample space for a metric for the device with respect to an origin to form a set of samples, wherein the set of samples comprises one or more failing samples;
    determine a center of gravity of the one or more failing samples with respect to the origin;
    determine importance samples based on the center of gravity of the one or more failing samples;
    recompute new importance sampling weight ratios for a selected new origin; and
    determine a failure rate for the device based on the new importance sampling weight ratios for the new origin.
  10. 10. The computer program product of claim 9, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    finding a projected origin; and
    recomputing new importance sampling weight ratios with respect to the projected origin.
  11. 11. The computer program product of claim 10, wherein finding the projected origin comprises:
    determining a line passing through the origin and the center of gravity of the one or more failing samples; and
    projecting the new origin onto the line passing though the origin and the center of gravity of the one or more failing samples.
  12. 12. The computer program product of claim 9, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    finding a set of projected samples with respect to the new origin; and
    recomputing new importance sampling ratios based on the projected samples.
  13. 13. The computer program product of claim 12, wherein finding the set of projected samples comprises:
    determining a line passing through the origin and the center of gravity of the one or more failing samples; and
    move the set of samples in a direction orthogonal to the line passing through the origin and the center of gravity of the one or more failing samples.
  14. 14. The computer program product of claim 9, wherein recomputing new importance sampling weight ratios for the new origin comprises computing a weight function, wherein the weight function is as follows:
    w ( x ) = Π σ σ np exp ( - 0.5 ( x - x np σ np ) 2 ) / exp ( - 0.5 ( x - x COG σ ) 2 )
    where x is process variation variable of the device, xnp is a new point of a projected origin for the process variation variable x, xCOG is the center of gravity of the one or more failing samples, σ is the standard deviation of x, and σnp is the standard deviation associated with the new point of projected origin.
  15. 15. The computer program product of claim 9, wherein the computer readable program further causes the computing device to:
    repeat selecting a new origin, recomputing new importance sampling weight ratios for the new origin, and determining a failure rate for the device based on the new importance sampling weight ratios for the new origin for a set of process variations.
  16. 16. The computer program product of claim 9, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    determining importance samples based on the center of gravity of the one or more failing samples and the new origin.
  17. 17. The computer program product of claim 9, wherein the computer readable program is stored in a computer readable storage medium in a data processing system and wherein the computer readable program was downloaded over a network from a remote data processing system.
  18. 18. The computer program product of claim 9, wherein the computer readable program is stored in a computer readable storage medium in a server data processing system and wherein the computer readable program is downloaded over a network to a remote data processing system for use in a computer readable storage medium with the remote system.
  19. 19. An apparatus, comprising:
    a processor; and
    a memory coupled to the processor, wherein the memory comprises instructions which, when executed by the processor, cause the processor to:
    perform a uniform sampling over a random sample space for a metric for the device with respect to an origin to form a set of samples, wherein the set of samples comprises one or more failing samples;
    determine a center of gravity of the one or more failing samples with respect to the origin;
    determine importance samples based on the center of gravity of the one or more failing samples;
    recompute new importance sampling weight ratios for a selected new origin; and
    determine a failure rate for the device based on the new importance sampling weight ratios for the new origin.
  20. 20. The apparatus of claim 19, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    finding a projected origin; and
    recomputing new importance sampling weight ratios with respect to the projected origin.
  21. 21. The apparatus of claim 20, wherein finding the projected origin comprises:
    determining a line passing through the origin and the center of gravity of the one or more failing samples; and
    projecting the new origin onto the line passing through the origin and the center of gravity of the one or more failing samples.
  22. 22. The apparatus of claim 19, wherein recomputing new importance sampling weight ratios for the new origin comprises:
    finding a set of projected samples with respect to the new origin; and
    recomputing new importance sampling ratios based on the projected samples.
  23. 23. The apparatus of claim 22, wherein finding the set of projected samples comprises:
    determining a line passing through the origin and the center of gravity of the one or more failing samples; and
    move the set of samples in a direction orthogonal to the line passing through the origin and the center of gravity of the one or more failing samples.
  24. 24. The apparatus of claim 19, wherein recomputing new importance sampling weight ratios for the new origin comprises computing a weight function, wherein the weight function is as follows:
    w ( x ) = Π σ σ np exp ( - 0.5 ( x - x np σ np ) 2 ) / exp ( - 0.5 ( x - x COG σ ) 2 )
    where x is process variation variable of the device, xnp is a new point of a projected origin for the process variation variable x, xCOG is the center of gravity of the one or more failing samples, σ is the standard deviation of x, and σnp is the standard deviation associated with the new point of projected origin.
  25. 25. The apparatus of claim 19, wherein the instructions further cause the processor to:
    repeat selecting a new origin, recomputing new importance sampling weight ratios for the new origin, and determining a failure rate for the device based on the new importance sampling weight ratios for the new origin for a set of process variations.
US12859871 2010-08-20 2010-08-20 Statistical Design with Importance Sampling Reuse Abandoned US20120046929A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US12859871 US20120046929A1 (en) 2010-08-20 2010-08-20 Statistical Design with Importance Sampling Reuse

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US12859871 US20120046929A1 (en) 2010-08-20 2010-08-20 Statistical Design with Importance Sampling Reuse
US14242418 US9348680B2 (en) 2010-08-20 2014-04-01 Statistical design with importance sampling reuse
US15161462 US20160266950A1 (en) 2010-08-20 2016-05-23 Statistical Design with Importance Sampling Reuse

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14242418 Continuation US9348680B2 (en) 2010-08-20 2014-04-01 Statistical design with importance sampling reuse

Publications (1)

Publication Number Publication Date
US20120046929A1 true true US20120046929A1 (en) 2012-02-23

Family

ID=45594758

Family Applications (3)

Application Number Title Priority Date Filing Date
US12859871 Abandoned US20120046929A1 (en) 2010-08-20 2010-08-20 Statistical Design with Importance Sampling Reuse
US14242418 Active US9348680B2 (en) 2010-08-20 2014-04-01 Statistical design with importance sampling reuse
US15161462 Pending US20160266950A1 (en) 2010-08-20 2016-05-23 Statistical Design with Importance Sampling Reuse

Family Applications After (2)

Application Number Title Priority Date Filing Date
US14242418 Active US9348680B2 (en) 2010-08-20 2014-04-01 Statistical design with importance sampling reuse
US15161462 Pending US20160266950A1 (en) 2010-08-20 2016-05-23 Statistical Design with Importance Sampling Reuse

Country Status (1)

Country Link
US (3) US20120046929A1 (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120116850A1 (en) * 2010-11-10 2012-05-10 International Business Machines Corporation Causal modeling of multi-dimensional hierachical metric cubes
US20140215274A1 (en) * 2010-08-20 2014-07-31 International Business Machines Corporation Statistical Design with Importance Sampling Reuse
US8806418B1 (en) 2013-06-19 2014-08-12 Freescale Semiconductor, Inc. Scaled sigma sampling
US20140278309A1 (en) * 2013-03-15 2014-09-18 International Business Machines Corporation Selective importance sampling
JP2014229311A (en) * 2013-05-27 2014-12-08 三星電子株式会社Samsung Electronics Co.,Ltd. Simulation system and method and computer system including simulation system
WO2014204577A1 (en) * 2013-06-21 2014-12-24 California Institute Of Technology Determination of electronic circuit robustness
US20160063156A1 (en) * 2014-08-26 2016-03-03 International Business Machines Corporation Multi-dimension variable predictive modeling for yield analysis acceleration
CN105701293A (en) * 2016-01-13 2016-06-22 北京航空航天大学 Method for generating multistage task system fault physical simulation sections of avionic devices

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160116527A1 (en) * 2014-10-27 2016-04-28 Qualcomm Incorporated Stochastic and topologically aware electromigration analysis methodology
WO2017176258A1 (en) * 2016-04-05 2017-10-12 Hewlett Packard Enterprise Development Lp Simulation based on precomputed results of the simulation

Citations (38)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020035462A1 (en) * 2000-07-24 2002-03-21 Makoto Kidera Method of and device for simulation
US20030004679A1 (en) * 2001-01-08 2003-01-02 Tryon Robert G. Method and apparatus for predicting failure in a system
US20030061152A1 (en) * 2001-09-26 2003-03-27 De Rabi S. System and method for determining Value-at-Risk using FORM/SORM
US20040199444A1 (en) * 2003-01-23 2004-10-07 Joel Woodcock Method of characterizing financial benefits based upon input operational parameters having uncertainties
US20050273308A1 (en) * 2004-06-07 2005-12-08 Houston Theodore W Statistical evaluation of circuit robustness separating local and global variation
US7016825B1 (en) * 2000-10-26 2006-03-21 Vextec Corporation Method and apparatus for predicting the failure of a component
US20060142988A1 (en) * 2004-12-28 2006-06-29 Satoru Akiyama Design methodology and manufacturing method for semiconductor memory
US20060150129A1 (en) * 2004-12-10 2006-07-06 Anova Solutions, Inc. Stochastic analysis process optimization for integrated circuit design and manufacture
US7096140B2 (en) * 2003-10-30 2006-08-22 Kabushiki Kaisha Toshiba Test system, test method and test program for an integrated circuit by IDDQ testing
US20060229856A1 (en) * 2005-04-11 2006-10-12 Nicolas Burrus Systems, devices, and methods for diffusion tractography
US20070097763A1 (en) * 2003-11-10 2007-05-03 Renesas Technology Corp. Manufacturing method of semiconductor integrated circuit device
US7272808B1 (en) * 2005-06-08 2007-09-18 Advanced Micro Devices, Inc. On-chip variability impact simulation and analysis for circuit performance
US20070220455A1 (en) * 2006-03-14 2007-09-20 Joshi Rajiv V Method and computer program for efficient cell failure rate estimation in cell arrays
US20080015827A1 (en) * 2006-01-24 2008-01-17 Tryon Robert G Iii Materials-based failure analysis in design of electronic devices, and prediction of operating life
US20080059143A1 (en) * 2005-12-12 2008-03-06 Hsien-Yen Chiu Hierarchical stochastic analysis process optimization for integrated circuit design and manufacture
US20080140363A1 (en) * 2006-12-12 2008-06-12 Chung-Kai Lin Unified Model for process variations in integrated circuits
US20080141190A1 (en) * 2006-12-06 2008-06-12 Qualcomm Incorporated Process Variation Tolerant Memory Design
US20090070716A1 (en) * 2007-09-12 2009-03-12 Joshi Rajiv V System and method for optimization and predication of variability and yield in integrated ciruits
US7508697B1 (en) * 2007-05-09 2009-03-24 Purdue Research Foundation Self-repairing technique in nano-scale SRAM to reduce parametric failures
US20090182536A1 (en) * 2003-07-14 2009-07-16 Morris Joel M System and method for estimating probabilities of events
US20090248387A1 (en) * 2008-03-28 2009-10-01 Carnegie Mellon University Method and apparatus for sampling and predicting rare events in complex electronic devices, circuits and systems
US20100010744A1 (en) * 2008-07-11 2010-01-14 Schlumberger Technology Corporation Monte carlo method for laplace inversion of nmr data
US20100153086A1 (en) * 2008-12-11 2010-06-17 Jedat Inc. System for creating parameter information, system for estimating yields, program and recording medium
US7827018B2 (en) * 2007-11-16 2010-11-02 International Business Machines Corporation Method and computer program for selecting circuit repairs using redundant elements with consideration of aging effects
US20110054856A1 (en) * 2009-09-01 2011-03-03 International Business Machines Corporation Equivalent Device Statistical Modeling for Bitline Leakage Modeling
US7920995B2 (en) * 2005-06-29 2011-04-05 Siemens Aktiengesellschaft Probabilistic design tool for optimizing a technical system
US20110153271A1 (en) * 2009-12-23 2011-06-23 Cadence Design Systems, Inc. Methods and systems for high sigma yield estimation
US20110153272A1 (en) * 2009-12-23 2011-06-23 Cadence Design Systems, Inc. Methods and systems for high sigma yield estimation using reduced dimensionality
US8010325B2 (en) * 2008-04-25 2011-08-30 Microsoft Corporation Failure simulation and availability report on same
US20110307438A1 (en) * 2010-06-14 2011-12-15 Fern E Acu A Ee Ndez Mart E Acu I Ee Nez Juan Luis High-dimensional data analysis
US20110313747A1 (en) * 2010-06-22 2011-12-22 International Business Machines Corporation Technology Computer-Aided Design (TCAD)-Based Virtual Fabrication
US20120150500A1 (en) * 2010-12-08 2012-06-14 Fujitsu Limited Optimization processing method and apparatus
US8214190B2 (en) * 2009-04-13 2012-07-03 International Business Machines Corporation Methodology for correlated memory fail estimations
US8275596B2 (en) * 2006-12-08 2012-09-25 Globalfoundries Inc. Method for robust statistical semiconductor device modeling
US8285522B1 (en) * 2006-01-24 2012-10-09 Vextec Corporation Materials-based failure analysis in design of electronic devices
US20120317656A1 (en) * 2011-04-05 2012-12-13 International Business Machines Corporation Two-party private estimation of dataset similarity
US20130159947A1 (en) * 2011-12-16 2013-06-20 International Business Machines Corporation Guiding design actions for complex failure modes
US20130226544A1 (en) * 2010-10-27 2013-08-29 Solido Design Automation Inc. Method and system for identifying rare-event failure rates

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5943661A (en) * 1991-07-11 1999-08-24 Texas Instruments Incorporated Hybrid neural network classifier, systems and methods
US5235506A (en) * 1991-08-30 1993-08-10 The United States Of America As Represented By The Secretary Of The Navy Process which aids to the laying out of locations of a limited number of 100, personnel and equipments in functional organization
US7050936B2 (en) * 2001-09-06 2006-05-23 Comverse, Ltd. Failure prediction apparatus and method
US7747624B2 (en) * 2002-05-10 2010-06-29 Oracle International Corporation Data summarization
US6999599B2 (en) * 2002-06-07 2006-02-14 Microsoft Corporation System and method for mode-based multi-hypothesis tracking using parametric contours
US7474999B2 (en) * 2002-12-23 2009-01-06 Cadence Design Systems, Inc. Method for accounting for process variation in the design of integrated circuits
US20070166847A1 (en) * 2006-01-17 2007-07-19 Honeywell International Inc. Boxes for Soft Error Rate Calculation
WO2009076600A3 (en) 2007-12-12 2009-08-20 Esteban G Tabak System, method and computer-accessible medium for normalizing databases through mixing
US20090248368A1 (en) * 2008-03-26 2009-10-01 Livermore Software Technology Corporation Systems and methods of constructing Radial Basis Function (RBF) based meta-models used in engineering design optimization
US7996719B2 (en) * 2008-10-24 2011-08-09 Microsoft Corporation Expressing fault correlation constraints
US8365118B2 (en) * 2009-06-03 2013-01-29 International Business Machines Corporation Broken-spheres methodology for improved failure probability analysis in multi-fail regions
US20120046929A1 (en) * 2010-08-20 2012-02-23 International Business Machines Corporation Statistical Design with Importance Sampling Reuse
US20140244563A1 (en) * 2011-11-10 2014-08-28 Hitachi, Ltd. Operation information prediction computer, operation information prediction method and program
US8832497B2 (en) * 2012-02-07 2014-09-09 A.L.D. Advanced Logistics Development Ltd Methods, apparatus and systems for performing dynamic fault tree analysis

Patent Citations (50)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020035462A1 (en) * 2000-07-24 2002-03-21 Makoto Kidera Method of and device for simulation
US7016825B1 (en) * 2000-10-26 2006-03-21 Vextec Corporation Method and apparatus for predicting the failure of a component
US7480601B2 (en) * 2000-10-26 2009-01-20 Vextec Corporation Methods and apparatus for predicting the failure of a component, and for determining a grain orientation factor for a material
US7006947B2 (en) * 2001-01-08 2006-02-28 Vextec Corporation Method and apparatus for predicting failure in a system
US20060009951A1 (en) * 2001-01-08 2006-01-12 Vextec Corporation Method and apparatus for predicting failure in a system
US20030004679A1 (en) * 2001-01-08 2003-01-02 Tryon Robert G. Method and apparatus for predicting failure in a system
US20030061152A1 (en) * 2001-09-26 2003-03-27 De Rabi S. System and method for determining Value-at-Risk using FORM/SORM
US20040199444A1 (en) * 2003-01-23 2004-10-07 Joel Woodcock Method of characterizing financial benefits based upon input operational parameters having uncertainties
US20090182536A1 (en) * 2003-07-14 2009-07-16 Morris Joel M System and method for estimating probabilities of events
US7096140B2 (en) * 2003-10-30 2006-08-22 Kabushiki Kaisha Toshiba Test system, test method and test program for an integrated circuit by IDDQ testing
US20070097763A1 (en) * 2003-11-10 2007-05-03 Renesas Technology Corp. Manufacturing method of semiconductor integrated circuit device
US20050273308A1 (en) * 2004-06-07 2005-12-08 Houston Theodore W Statistical evaluation of circuit robustness separating local and global variation
US20060150129A1 (en) * 2004-12-10 2006-07-06 Anova Solutions, Inc. Stochastic analysis process optimization for integrated circuit design and manufacture
US20060142988A1 (en) * 2004-12-28 2006-06-29 Satoru Akiyama Design methodology and manufacturing method for semiconductor memory
US20060229856A1 (en) * 2005-04-11 2006-10-12 Nicolas Burrus Systems, devices, and methods for diffusion tractography
US7272808B1 (en) * 2005-06-08 2007-09-18 Advanced Micro Devices, Inc. On-chip variability impact simulation and analysis for circuit performance
US7920995B2 (en) * 2005-06-29 2011-04-05 Siemens Aktiengesellschaft Probabilistic design tool for optimizing a technical system
US20080059143A1 (en) * 2005-12-12 2008-03-06 Hsien-Yen Chiu Hierarchical stochastic analysis process optimization for integrated circuit design and manufacture
US8285522B1 (en) * 2006-01-24 2012-10-09 Vextec Corporation Materials-based failure analysis in design of electronic devices
US8290753B2 (en) * 2006-01-24 2012-10-16 Vextec Corporation Materials-based failure analysis in design of electronic devices, and prediction of operating life
US20080015827A1 (en) * 2006-01-24 2008-01-17 Tryon Robert G Iii Materials-based failure analysis in design of electronic devices, and prediction of operating life
US20080195325A1 (en) * 2006-03-14 2008-08-14 Joshi Rajiv V System and computer program for efficient cell failure rate estimation in cell arrays
US7380225B2 (en) * 2006-03-14 2008-05-27 International Business Machines Corporation Method and computer program for efficient cell failure rate estimation in cell arrays
US20070220455A1 (en) * 2006-03-14 2007-09-20 Joshi Rajiv V Method and computer program for efficient cell failure rate estimation in cell arrays
US20080141190A1 (en) * 2006-12-06 2008-06-12 Qualcomm Incorporated Process Variation Tolerant Memory Design
US7979832B2 (en) * 2006-12-06 2011-07-12 Qualcomm Incorporated Process variation tolerant memory design
US8275596B2 (en) * 2006-12-08 2012-09-25 Globalfoundries Inc. Method for robust statistical semiconductor device modeling
US20080140363A1 (en) * 2006-12-12 2008-06-12 Chung-Kai Lin Unified Model for process variations in integrated circuits
US7508697B1 (en) * 2007-05-09 2009-03-24 Purdue Research Foundation Self-repairing technique in nano-scale SRAM to reduce parametric failures
US20090070716A1 (en) * 2007-09-12 2009-03-12 Joshi Rajiv V System and method for optimization and predication of variability and yield in integrated ciruits
US7827018B2 (en) * 2007-11-16 2010-11-02 International Business Machines Corporation Method and computer program for selecting circuit repairs using redundant elements with consideration of aging effects
US20090248387A1 (en) * 2008-03-28 2009-10-01 Carnegie Mellon University Method and apparatus for sampling and predicting rare events in complex electronic devices, circuits and systems
US8155938B2 (en) * 2008-03-28 2012-04-10 Carnegie Mellon University Method and apparatus for sampling and predicting rare events in complex electronic devices, circuits and systems
US8010325B2 (en) * 2008-04-25 2011-08-30 Microsoft Corporation Failure simulation and availability report on same
US20100010744A1 (en) * 2008-07-11 2010-01-14 Schlumberger Technology Corporation Monte carlo method for laplace inversion of nmr data
US8374839B2 (en) * 2008-12-11 2013-02-12 Jedat Inc. System for creating parameter information, system for estimating yields, program and recording medium
US20100153086A1 (en) * 2008-12-11 2010-06-17 Jedat Inc. System for creating parameter information, system for estimating yields, program and recording medium
US8214190B2 (en) * 2009-04-13 2012-07-03 International Business Machines Corporation Methodology for correlated memory fail estimations
US8346528B2 (en) * 2009-09-01 2013-01-01 International Business Machines Corporation Equivalent device statistical modeling for bitline leakage modeling
US20110054856A1 (en) * 2009-09-01 2011-03-03 International Business Machines Corporation Equivalent Device Statistical Modeling for Bitline Leakage Modeling
US8219355B2 (en) * 2009-12-23 2012-07-10 Cadence Design Systems, Inc. Methods and systems for high sigma yield estimation
US20110153272A1 (en) * 2009-12-23 2011-06-23 Cadence Design Systems, Inc. Methods and systems for high sigma yield estimation using reduced dimensionality
US20110153271A1 (en) * 2009-12-23 2011-06-23 Cadence Design Systems, Inc. Methods and systems for high sigma yield estimation
US20110307438A1 (en) * 2010-06-14 2011-12-15 Fern E Acu A Ee Ndez Mart E Acu I Ee Nez Juan Luis High-dimensional data analysis
US8515724B2 (en) * 2010-06-22 2013-08-20 International Business Machines Corporation Technology computer-aided design (TCAD)-based virtual fabrication
US20110313747A1 (en) * 2010-06-22 2011-12-22 International Business Machines Corporation Technology Computer-Aided Design (TCAD)-Based Virtual Fabrication
US20130226544A1 (en) * 2010-10-27 2013-08-29 Solido Design Automation Inc. Method and system for identifying rare-event failure rates
US20120150500A1 (en) * 2010-12-08 2012-06-14 Fujitsu Limited Optimization processing method and apparatus
US20120317656A1 (en) * 2011-04-05 2012-12-13 International Business Machines Corporation Two-party private estimation of dataset similarity
US20130159947A1 (en) * 2011-12-16 2013-06-20 International Business Machines Corporation Guiding design actions for complex failure modes

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Efficient Rare Event Simulation: A Tutorial on Importance Sampling; Michele Pagano et al; July 2005; Pgs. 1-244 *
IEEE paper "Robust Importance Sampling for Efficient SRAM Yield AnalysisTakanori Date et al, (IEEE March 2010, Pgs: 15-21) *
Importance Sampling for the Simulation of Highly Reliable Markovian SystemsAuthor(s): Perwez ShahabuddinSource: Management Science, Vol. 40, No. 3 (Mar., 1994), pp. 333-352 *
Sequential Importance Sampling for Low-Probability and High-Dimensional SRAM Yield Analysis; Kentaro Katayama et al; 7-11 Nov.2010 IEEE; Page(s): 703 - 708 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140215274A1 (en) * 2010-08-20 2014-07-31 International Business Machines Corporation Statistical Design with Importance Sampling Reuse
US9348680B2 (en) * 2010-08-20 2016-05-24 International Business Machines Corporation Statistical design with importance sampling reuse
US20120116850A1 (en) * 2010-11-10 2012-05-10 International Business Machines Corporation Causal modeling of multi-dimensional hierachical metric cubes
US20140278309A1 (en) * 2013-03-15 2014-09-18 International Business Machines Corporation Selective importance sampling
US20140278296A1 (en) * 2013-03-15 2014-09-18 International Business Machines Corporation Selective importance sampling
US9576085B2 (en) * 2013-03-15 2017-02-21 International Business Machines Corporation Selective importance sampling
US9460243B2 (en) * 2013-03-15 2016-10-04 International Business Machines Corporation Selective importance sampling
JP2014229311A (en) * 2013-05-27 2014-12-08 三星電子株式会社Samsung Electronics Co.,Ltd. Simulation system and method and computer system including simulation system
US8806418B1 (en) 2013-06-19 2014-08-12 Freescale Semiconductor, Inc. Scaled sigma sampling
WO2014204577A1 (en) * 2013-06-21 2014-12-24 California Institute Of Technology Determination of electronic circuit robustness
US20160063156A1 (en) * 2014-08-26 2016-03-03 International Business Machines Corporation Multi-dimension variable predictive modeling for yield analysis acceleration
CN105701293A (en) * 2016-01-13 2016-06-22 北京航空航天大学 Method for generating multistage task system fault physical simulation sections of avionic devices

Also Published As

Publication number Publication date Type
US9348680B2 (en) 2016-05-24 grant
US20160266950A1 (en) 2016-09-15 application
US20140215274A1 (en) 2014-07-31 application

Similar Documents

Publication Publication Date Title
Blaauw et al. Statistical timing analysis: From basic principles to state of the art
US6378112B1 (en) Verification of design blocks and method of equivalence checking of multiple design views
Mukhopadhyay et al. Modeling of failure probability and statistical design of SRAM array for yield enhancement in nanoscaled CMOS
Agarwal et al. Statistical analysis of SRAM cell stability
US7137080B2 (en) Method for determining and using leakage current sensitivities to optimize the design of an integrated circuit
Kwong et al. Variation-driven device sizing for minimum energy sub-threshold circuits
Wang et al. Analyzing static and dynamic write margin for nanometer SRAMs
US20070016881A1 (en) Automation method and system for assessing timing based on Gaussian slack
US7882471B1 (en) Timing and signal integrity analysis of integrated circuits with semiconductor process variations
Chen et al. Yield-driven near-threshold SRAM design
Dhillon et al. Soft-error tolerance analysis and optimization of nanometer circuits
Agarwal et al. The impact of random device variation on SRAM cell stability in sub-90-nm CMOS technologies
US6327557B1 (en) Method and system for creating electronic circuitry
Naseer et al. Critical charge characterization for soft error rate modeling in 90nm SRAM
US20060203581A1 (en) Efficient method and computer program for modeling and improving static memory performance across process variations and environmental conditions
Rao et al. Computing the soft error rate of a combinational logic circuit using parameterized descriptors
US20130226544A1 (en) Method and system for identifying rare-event failure rates
Sylvester et al. Variability in nanometer CMOS: Impact, analysis, and minimization
US20090248387A1 (en) Method and apparatus for sampling and predicting rare events in complex electronic devices, circuits and systems
US6378113B1 (en) Black box transparency in a circuit timing model
Zhang et al. Analytical modeling of SRAM dynamic stability
US20100250187A1 (en) Method and system for analyzing performance metrics of array type circuits under process variability
US20070226572A1 (en) Soft error rate analysis system
Ramanarayanan et al. Modeling soft errors at the device and logic levels for combinational circuits
US7380225B2 (en) Method and computer program for efficient cell failure rate estimation in cell arrays

Legal Events

Date Code Title Description
AS Assignment

Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:JOSHI, RAJIV V.;KANJ, ROUWAIDA N.;NASSIF, SANI R.;AND OTHERS;SIGNING DATES FROM 20100812 TO 20100818;REEL/FRAME:024908/0927