The significant process parameter variations occurring during fabrication of high performance sequential circuits, such as microprocessors, are posing relevant uncertainties on the power that such circuits will consume in the field, while executing workloads typical for the diverse products they are oriented to (e.g., cellular phones, notebooks, servers, etc). On the other hand, different kinds of products have different constraints on the maximal power that could be consumed during the execution of typical workloads, due to diverse needs in terms of charge autonomy, heat dissipation, etc. Consequently, the power that will be consumed by microprocessors during the execution of typical workloads in the field needs to be accurately characterized at the end of fabrication. Such a power consumption characterization (hereinafter referred to as “power binning”), will enable to classify microprocessors in “power bins”, each one containing microprocessors suitable for different kinds of products, thus enabling to introduce them all into the market for different kinds of products. Based on these considerations, in this paper we propose an approach to characterize accurately at the end of fabrication, and at low-cost (in terms of characterization time), the power that microprocessors will consume in the in-field during the execution of workloads typical for different kinds of products. Our approach exploits scan-based Logic Built-In Self-Test (LBIST) to apply to microprocessors' sequential blocks test vectors that induce on their internal nodes an activity factor (AF) similar to that experienced during the in-field execution of workloads typical for different kinds of products, thus enabling to perform power binning by simply measuring their consumed power. Our approach enables to scale the AF from 0 percent up to 97.6 percent (on average for the considered benchmark circuits) compared to conventional LBIST, with a granularity of the 2 percent, thus enabling to emulate accurately the AF induced by workloads typical of a wide range of products. We propose a hardware implementation for our approach requiring a limited area overhead (lower than 3 percent) over conventional LBIST.

The generation of significant power droop (PD) during at-speed test performed by Logic Built-In Self Test (LBIST) is a serious concern for modern ICs. In fact, the PD originated during test may delay signal transitions of the circuit under test (CUT): an effect that may be erroneously recognized as delay faults, with consequent erroneous generation of test fails and increase in yield loss. In this paper, we propose a novel scalable approach to reduce the PD during at-speed test of sequential circuits with scan-based LBIST using the launch-on-capture scheme. This is achieved by reducing the activity factor of the CUT, by proper modification of the test vectors generated by the LBIST of sequential ICs. Our scalable solution allows us to reduce PD to a value similar to that occurring during the CUT in field operation, without increasing the number of test vectors required to achieve a target fault coverage (FC). We present a hardware implementation of our approach that requires limited area overhead. Finally, we show that, compared with recent alternative solutions providing a similar PD reduction, our approach enables a significant reduction of the number of test vectors (by more than 50%), thus the test time, to achieve a target FC.

During at-speed test of high performance sequential ICs using scan-based Logic BIST, the IC activity factor (AF) induced by the applied test vectors is significantly higher than that experienced during its in field operation. Consequently, power droop (PD) may take place during both shift and capture phases, which will slow down the circuit under test (CUT) signal transitions. At capture, this phenomenon is likely to be erroneously recognized as due to delay faults. As a result, a false test fail may be generated, with consequent increase in yield loss. In this paper, we propose two approaches to reduce the PD generated at capture during at-speed test of sequential circuits with scan-based Logic BIST using the Launch-On-Shift scheme. Both approaches increase the correlation between adjacent bits of the scan chains with respect to conventional scan-based LBIST. This way, the AF of the scan chains at capture is reduced. Consequently, the AF of the CUT at capture, thus the PD at capture, is also reduced compared to conventional scan-based LBIST. The former approach, hereinafter referred to as Low-Cost Approach (LCA), enables a 50 percent reduction in the worst case magnitude of PD during conventional logic BIST. It requires a small cost in terms of area overhead (of approximately 1.5 percent on average), and it does not increase the number of test vectors over the conventional scan-based LBIST to achieve the same Fault Coverage (FC). Moreover, compared to three recent alternative solutions, LCA features a comparable AF in the scan chains at capture, while requiring lower test time and area overhead. The second approach, hereinafter referred to as High-Reduction Approach (HRA), enables scalable PD reductions at capture of up to 87 percent, with limited additional costs in terms of area overhead and number of required test vectors for a given target FC, over our LCA approach. Particularly, compared to two of the three recent alternative solutions mentioned above, HRA enables a significantly lower AF in the scan chains during the application of test vectors, while requiring either a comparable area overhead or a significantly lower test time. Compared to the remaining alternative solutions mentioned above, HRA enables a similar AF in the scan chains at capture (approximately 90 percent lower than conventional scan-based LBIST), while requiring a significantly lower test time (approximately 4.87 times on average lower number of test vectors) and comparable area overhead (of approximately 1.9 percent on average).

Global Signal Vulnerability (GSV) analysis is a novel method for assessing the susceptibility of modern microprocessor state elements to failures in the field of operation. In order to effectively allocate design for reliability resources, GSV analysis takes into account the high degree of architectural masking exhibited in modern microprocessors and ranks state elements accordingly. The novelty of this method lies in the way this ranking is computed. GSV analysis operates either at the Register Transfer (RT-) or at the Gate-Level, offering increased accuracy in contrast to methods which compute the architectural vulnerability of registers through high-level simulations on performance models. Moreover, it does not rely on extensive Statistical Fault Injection (SFI) campaigns and lengthy executions of workloads to completion in RT- or Gate-Level designs, which would make such analysis prohibitive. Instead, it monitors the behavior of key global microprocessor signals in response to a progressive stuck-at fault injection method during partial workload execution. Experimentation with the Scheduler and Reorder Buffer modules of an Alpha-like microprocessor and a modern Intel microprocessor corroborates that GSV analysis generates a near-optimal ranking, yet is several orders of magnitude faster than existing RT- or Gate-Level approaches.

THE continuous scaling of microelectronic technology, while allowing to integrate increasingly complex and high performance systems on a die, poses new challenges to their reliable operation in the field, due to the increased likelihood of faults and aging phenomena possibly occurring in the field and compromising the system’s correct operation. Several on-line testing and error/fault resilience techniques have been employed in the past to implement highly reliable, fault tolerant systems for mission critical applications, in areas like space, military, automotive, medical, banking, etc. However, new faults and aging phenomena occurring in the field are posing unique on-line testing and error/fault resilience challenges even for mainstream applications, where cost is a crucial factor. This mandates the development and adoption of innovative solutions optimized for cost, power and area.

Post-silicon validation is a necessary step in a design's verification process. Pre-silicon techniques such as simulation and emulation are limited in scope and volume as compared to what can be achieved on the silicon itself. Some parts of the verification, such as full-system functional verification, cannot be practically covered with current pre-silicon technologies. This panel brings together experts from industry, academia, and EDA to review the differences and similarities between pre- and post-silicon, discuss how the fundamental aspects of verification are affected by these differences, and explore how the gaps between the two worlds can be bridged.

Functional test content to screen for electrical marginalities during silicon validation are not generated with the goal of identifying speed-limiting paths, adversely affecting the quality and efficiency of validation. We propose a methodology to generate functional tests to excite pre-silicon timing-critical paths along with environmental effects such as voltage droop. These tests are to replace random/function-targeted content as the source for identifying speed failures during silicon validation. The effectiveness of this methodology is demonstrated through silicon experiments on a recent processor.

This paper presents a concurrent error detection technique targeted towards control logic in a processor with emphasis on low area overhead. Rather than detect all modeled transient faults, the technique selects faults which have a high probability of causing damage to the architectural state of the processor and protects the circuit against these faults. Fault detection is achieved through a series of assertions. Each assertion is an implication from inputs to the outputs of a combinational circuit. Fault simulation experiments performed on control logic modules of an industrial processor suggest that high reduction in damage causing faults can be achieved with a low overhead.

We propose an on-line testing approach for the control logic of high performance microprocessors. Rather than adding information redundancy (in the form of error detecting codes), we propose to look for the information redundancy (referred to as Function-Inherent Codes) that the microprocessor control logic may inherently have, due to its required functionality. We will show that this allows to achieve on-line testing at significant savings in terms of area and power consumption, and with lower or comparable impact on system performance and design costs, compared to alternate, traditional on-line testing approaches.

Test catches defective parts. Traditional DFT makes test-related activities more cost effective. Debug and diagnosis discovers why a design or part did not work. Yield relates to how many good parts can be shipped thus how many dollars can be received. As one would expect, all these areas are related. The issue becomes: Can information from one operation be geared to better another? This panel specifically considers how test/DFT can aid in debug & diagnosis. The discussion is then extended to capture the impact on yield management in riskier contemporary and future processes. Over the past few decades, each of the above domains have gone through cycles of research, development, maturity and overhaul to adapt to changes in the process & product environment. In fact, because increasing degrees of uncertainty is introduced with modern nanometer processes, debug & diagnosis and yield management are becoming more recognized and valued portions of the product creation flow. To forward this, test and DFT functions are experiencing re-development. In building, executing and sustaining a competence, key issues include time & information. That is, the problem must be understood, solved in for the short term and solved for the long term. In addition, solution scope must be properly framed and, if possible, adaptable to unexpected changes in the process and tooling environments. It is well recognized that information availability and access are pillars of a debug/diagnosis solution. DFT structures, design automation tooling, test programs, ATE applications and physical failure analysis are means to a bulk of this data. The follow through is the assimilation of this data into a yield management and test quality system. As the topic is broad, panel of test experts will focus on practical experiences concerning the use of test, crafted DFT or test environments to facilitate debug and diagnosis. Precise developments in tying debug/diagnosis to yield management will also be discussed.

Testing today of a several hundred million transistor system-on-chip with analog, RF blocks, many processor cores and tens of memories is a huge task. What the test technology be like in year 2020 with hundreds of billions of transistors on a single chip? Can we get there with tweaks to today's technology? While the exact nature of the circuit styles, architectural innovations and product innovations in year 2020 are highly speculative at this point, we examine the impact of likely design and process technology trends on testing methods

Moore’s law has enabled an ever increasing number of transistors in a given area, and this is expected to continue at least for the next decade. The additional transistors have enabled increased functionality in the products resulting in a richer user experience, while staying roughly within the same cost envelope. Over the last several years, innovations in design micro-architecture, signaling technologies, circuit styles, process technologies, power delivery and packaging have enabled us to further utilize the increased transistor count. Designer productivity has also been greatly improved due to EDA innovations such as high-level design languages, logic synthesis, verification, timing analysis, physical design and design re-use. As a result, the design complexity has steadily increased.

Driven by market applications in the areas of computing, networking, storage, optical, wireless, portable, and consumer electronics, semiconductor chips today are as diverse as ever. Confluence of multiple applications and rapid integration has also driven the heterogeneity of chips. Test methods have evolved with the products. However, the basic goals in testing remain the same: quality of product, recurring and non-recurring costs and time to market. In this paper we try to catalog some commonly used test methods, identify their associated DFT requirements and trends in terms of tester requirements. Given the diversity of semiconductors chips today such as various PLDs, volatile and non-volatile memories, analog, mixed signal, FPGA, ASIC, SOC, MEMs and processors, it is impossible for a paper of this nature to be fully comprehensive. So we limit our focus on processor, ASIC and SOCs.

An automatic test pattern generation program (ATPG) is described for large, application-specific integrated circuits (ASICs) designed with a scan path technique. This program was implemented with an improved deterministic test generation algorithm that makes use of a split model for circuit representation and multiple testability heuristics for efficient search. The program can also interface to a hardware accelerator to speed up the test compaction process.>