Search

Nano-crossbar arrays have emerged to achieve high performance computing beyond the limits of current CMOS with the drawback of higher fault rates. They offer area and power efficiency in terms of their easy-to-fabricate and dense physical structures. They consist of regularly placed crosspoints as computing elements, which behave as diode, memristor, field effect transistor, or novel four-terminal switching devices. In this study, we establish a complete design framework for crossbar circuits explaining and analyzing every step of the process. We comparatively elaborate on these technologies in the sense of their capabilities for computation regarding area including a new logic synthesis technique for memristors, fault tolerance including a novel paradigm for four-terminal devices, delay, and power consumption. As a result, this study introduces a synthesis methodology that considers basic technology preference for switching crosspoints and fault rates of the given crossbar as well as their effects on performance metrics including power, delay, and area.

International audience; Switching lattices are two-dimensional arrays composed of two or four-terminals switches organized as a crossbar array. The idea of using regular two-dimensional arrays of switches for Boolean function implementation was proposed by Akers in 1972. Recently, with the advent of a variety of emerging nanoscale technologies, lattices have found a renewed interest. Emerging technologies allow more complex function integration, thanks to their smaller sizes and advanta geous properties such as zero leakage current, capability to retain data when in power-off state, and almost unlimit edendurance, to name just a few appealing features. Also, implementation of new computing paradigms combining memory and logic becomes possible. However, emerging technologies show a non-negligible defect ratio and higher sensitivity to process and environment variations. The reliability challenges in adopting these technologies need to be investigated. In this paper, we analyze the resilience of switching lattices to stuck-at-fault model (SAF). We first identify the critical switches through an elaborated sensitivity methodology and extensive analysis of the lattice. Next, we propose several techniques to improve lattice resilience in the face of these types of faults, that can be implemented after lattice logic optimization steps.

A switching lattice is a two-dimensional array of four-terminal switches implemented in its cells. Each switch is linked to the four neighbors and is connected with them when the switch is ON, or is disconnected when the switch is OFF. Recently, with the advent of a variety of emerging nanoscale technologies based on regular arrays of switches, lattices of multi-terminal switches, originally introduced by Akers in 1972, have found a renewed interest. In this paper, the testability under the Cellular Fault Model (CFM) of switching lattices is defined and analyzed. Moreover, some techniques for improving the testability of lattices are discussed and experimentally evaluated.

Switching lattices are two-dimensional arrays composed by four-terminal switches (crossbar arrays). The idea of using regular two-dimensional arrays of switches to implement Boolean functions was proposed by Akers in 1972. Recently, with the advent of a variety of emerging nanoscale technologies, lattices have found a renewed interest. Switching lattices can have a non-negligible defective ratio. In this paper, we analyze the fault tolerance of switching lattices under the stuck-at-fault model (SAFM). We first identify the critical switches with a sensitivity analysis of the lattice. We then propose some techniques to improve the resilience to faults, which are implemented as a post preprocessing step after logic optimization.

A front-end ASIC for 4D tracking is presented. The prototype includes the block necessary to build a pixel front-end chain for timing measurement, as independent circuits. The architecture includes a charge-sensitive amplifier, a discriminator with programmable threshold, and a time- to-digital converter. The blocks were designed with target specifications in mind including: an area occupation of 55 μm × 55 μm, a power consumption tens of micro ampere per channel and timing a resolution of at least 100 ps. The prototype has been designed and integrated in 28 nm CMOS technology. The presented design is part of the TimeSpOT project which aims to reach a high-resolution particle tracking both in space and in time, in order to provide front-end circuitry suitable for next generation colliders.

We present the development of a test ASIC, named Timespot1, designed in CMOS 28-nm technology, featuring a 32x32 pixel matrix and a pitch of 55 μm, The ASIC is conceived as the first prototype in a series, capable to read-out pixels with timing capabilities in the range of 30 ps and below. Each pixel is endowed with a charge amplifier, a discriminator and a Time-to-Digital-Converter, capable of time resolutions below 20 ps and read-out rates (per pixel) around 3 MHz. The timing performance are obtained respecting a power budget of about 50 µW per pixel, corresponding to a power density of approximately 2 W/cm2• This feature makes the Timespot1 approach an interesting solution for vertex detectors of the next generation of colliders, where high space and time resolutions will be mandatory requirements to cope with the huge amount of tracks per event to be detected and processed.

Future high luminosity colliders will require front-end electronics with unprecedented performance, both in space and time resolution (tens of micrometers and tens of picoseconds) and in radiation hardness (tens of megagray). Moreover, the high number of events will generate an enormous quantity of data (some terabits per second), and the limited bandwidth requires to perform data selection as close as possible to the front-end stage, to reduce the amount of data transmitted and stored for off-line analysis.The TimeSpOT (TIME and SPace real-time Operating Tracker) project, funded by INFN, is developing a complete demonstrator of a tracking device including all the features needed for future high luminosity experiments.In this presentation, we describe the first prototype of the readout electronics in 28 nm CMOS technology. The modules of the front-end circuitry have been designed and integrated in a test chip, which will allow us to characterize each block separately, and to connect them in a processing chain to evaluate the overall performance.

International audience; The RD53A large scale pixel demonstrator chip has been developed in 65 nm CMOS technology by the RD53 collaboration, in order to face the unprecedented design requirements of the pixel 2 phase upgrades of the CMS and ATLAS experiments at CERN. This prototype chip is designed to demonstrate that a set of challenging specifications can be met, such as: high granularity (small pixels of 50×50 or 25× 100 µm2) and large pixel chip size (~2x2 cm2), high hit rate (3 GHz/cm2), high readout speed, very high radiation levels (500 Mrad - 1 Grad) and operation with serial powering. Furthermore, coping with the long latency of the trigger signal (~12.5 µs), used to select only events of interest in order to achieve sustainable output data rates, requires increased buffering resources in the limited pixel area. The RD53A chip has been fabricated in an engineer run. It integrates a matrix of 400×192 pixels and features various design variations in the analog and digital pixel matrix for testing purposes. This paper presents an overview of the chip architecture and of the methodologies used for efficient design of large complex mixed signal chips for harsh radiation environments. Experimental results obtained from the characterization of the RD53A chip are reported to demonstrate that design objectives have been achieved. Moreover, design improvements and new features being developed in the RD53B framework for final ATLAS and CMS production chips are discussed

International audience; This paper presents the characterization of the new Associative Memory chip (version 7) designed and fabricated in 28 nm CMOS. The design aims at: enhancing links from/to FPGAs; increasing bandwidth thanks to full custom LVDS transceivers; and reducing power consumption and silicon area by means of new memory cells designed with full-custom approach. The design was submitted in December 2016; the prototypes were fabricated and packaged in a 17 × 17 Ball Grid Array (BGA) standalone package. Prototype characterization confirms the chip functionality. The final chip will be assembled in a System In Package (SiP) together with a bare FPGA die.

Nano-crossbar arrays have emerged as area and power efficient structures with an aim of achieving high performance computing beyond the limits of current CMOS. Due to the stochastic nature of nano-fabrication, nano arrays show different properties both in structural and physical device levels compared to conventional technologies. Mentioned factors introduce random characteristics that need to be carefully considered by synthesis process. For instance, a competent synthesis methodology must consider basic technology preference for switching elements, defect or fault rates of the given nano switching array and the variation values as well as their effects on performance metrics including power, delay, and area. Presented synthesis methodology in this study comprehensively covers the all specified factors and provides optimization algorithms for each step of the process.

This paper presents a new Content Addressable Memory (CAM) cell, designed in two variants (inverting and non-inverting), and optimized for power consumption. The cells are designed in 28 nm CMOS technology, using a fully-CMOS approach. If one of the bits in a word is not matching, the corresponding cell sends a "kill" signal to the cascaded cell to inhibit further switching. Thanks to this feature, the designed CAM array requires less than 0.7 fJ/bit per comparison. A prototype containing 217 words of 18 bits each has been fabricated and it has proven to be functional.

In this paper we propose a novel approach to the synthesis of minimal-sized lattices, based on the decomposition of logic functions. Since the decomposition allows to obtain circuits with a smaller area, our idea is to decompose the Boolean functions according to generalizations of the classical Shannon decomposition, then generate the lattices for each component function, and finally implement the original function by a single composed lattice obtained by gluing together appropriately the lattices of the component functions. In particular we study the two decomposition schemes defining the bounded-level logic networks called P-circuits and EXOR-Projected Sums of Products (EP-SOPs). Experimental results show that about 34% of our benchmarks achieve a smaller area when implemented using the P-circuit decomposition for switching lattices, with an average gain of at least 25%, and about 27% of our benchmarks achieve a smaller area when implemented using the EP-SOP decomposition, with an average gain of at least 22%.

The paper presents the research activities in microelectronics, aiming at improving detection capabilities of future High Energy Physics (HEP) experiments. The output of this research will be the development of novel integrated circuits, to enhance the performance of electronic systems for the High Luminosity Large Hadron Collider (HL-LHC). In particular, the main research activities are focused on monolithic pixel arrays, on new digital architectures for pixel readout in 65 nm CMOS, and on associative memories for several interdisciplinary applications, such as fast tracking for trigger, DNA sequencing, magnetic resonance and image analysis.

International audience; RD53A is a large scale 65 nm CMOS pixel demonstrator chip that has been developed by the RD53 collaboration for very high rate (3 GHz/cm$^2$) and very high radiation levels (500 Mrad, possibly 1 Grad) for ATLAS and CMS phase 2 upgrades. It features serial powering operation and design variations in the analog and digital pixel matrix for different testing purposes. The design and verification of RD53A are described together with an outline of the plans to develop final pixel chips for the two experiments.

This paper proposes a novel population count circuit for Associative Memories (AM)s. Currently, AM chips requires a large number of silicon area for the population count circuitry. For this reason, is necessary an optimization in terms of area for the future AM devices to have a better memory density. A population count circuit counts how many blocks of the AM are in a matching state. If the count sum is greater than a preconfigured threshold, the output wire is set to ‘1’, otherwise is set to ‘0’. In the existing circuits there are, in addition, several control signals that are used to increase the circuit flexibility, but these controls require a large number of transistors and interconnections. The purpose of the proposed circuit is to reduce the number of transistors and interconnections complexity, with the final aim to reduce the occupied silicon area.

Multi-terminal switching lattices are typically exploited for modeling switching nano-crossbar arrays that lead to the design and construction of emerging nanocomputers. In this paper we propose a switching lattice optimization method for a special class of “regular” Boolean functions, called autosymmetric functions. Autosymmetry is a property that is frequent enough within Boolean functions to be interesting in the synthesis process. Each autosymmetric function can be synthesized through a new function (called restriction), depending on less variables and with a smaller on-set, which can be computed in polynomial time. In this paper we describe how to exploit the autosymmetry property of a Boolean function in oder to obtain a smaller lattice representation in a reduced minimization time. The original Boolean function can be constructed through a composition of the restriction with some EXORs of subsets of the input variables. Similarly, the lattice implementation of the function can be constructed using some external lattices for the EXORs, whose outputs will input the lattice implementing the restriction. Finally, the output of the restriction lattice corresponds to the output of the original function. Experimental results show that the total area of the obtained lattices is often significantly reduced. Moreover, in many cases, the computational time necessary to minimize the restriction is smaller than the time necessary to perform the lattice synthesis of the entire function.

In this paper we propose a novel approach to the synthesis of minimal-sized lattices, based on the decomposition of logic functions. Since the decomposition allows to obtain circuits with a smaller area, our idea is to decompose Boolean functions with separate lattices, according to the P-circuits decomposition scheme, and then to implement the decomposed blocks with physically separated regions in a single lattice. Experimental results show that about 35% of the considered benchmarks achieve a smaller area when implemented using the proposed decomposition for switching lattices, with an average gain of at least 24%.

In this paper we describe a Content Addressable Memory architecture designed in 28 nm CMOS technology and based on the 65 nm XORAM cell previously developed. The cell is composed by two main blocks: a 6T SRAM, and a 4T XOR logic gate. Each XORAM cell makes a bitwise comparison between input data and stored data. The memory is organized in 18-bit words, and all the 18 XOR outputs bits must have a low logic value to trigger a high logic value of the single bit match line. A 18-input NOR gate performs this operation. The memory operation is triggered by the change of the least significant bit of the 18-bit input word, which is delayed w.r.t. the other bits. In this way, the logic does not require any clock. The proposed architecture is based on CMOS combinational logic, and it does not require any precharge operation, nor control and timing logic. The Associative Memory block is useful for several pattern recognition tasks, such as track recognition in high energy physics experiments, and image recognition for medical applications.

This paper describes the AM06 chip, which is a highly parallel processor for pattern recognition in the ATLAS high energy physics experiment. The AM06 contains memory banks that store data organized in 18 bit words; a group of 8 words is called "pattern". Each AM06 chip can store up to 131 072 patterns. The AM06 is a large chip, designed in 65 nm CMOS, and it combines full-custom memory arrays, standard logic cells and serializer/deserializer IP blocks at 2 Gbit/s for input/output communication. The overall silicon area is 168 mm2 and the chip contains about 421 million transistors. The AM06 receives the detector data for each event accepted by Level-1 trigger, up to 100 kHz, and it performs a track reconstruction based on hit information from channels of the ATLAS silicon detectors. Thanks to the design of a new associative memory cell and to the layout optimization, the AM06 consumption is only about 1 fJ/bit per comparison. The AM06 has been fabricated and successfully tested with a dedicated test system.

In this paper we describe a Content Addressable Memory (CAM) architecture based on a new custom cell, called XORAM. The cell is composed by two main blocks: a 6T-SRAM, and a 4T-XOR logic gate. Each XORAM cell compares the input data on the bit line with the data stored in the 6T-SRAM cell. The output matching bit is obtained by performing a NOR operation between all bits of the XORAM cells storing the word. The proposed architecture is based on a fully-CMOS combinational logic, and it does nor require any precharge operation or control and timing logic. A compact full-custom layout has been designed for a memory organized in 18-bit words, to reduce both area and power consumption. Compared with a conventional selective precharge match-line technique, the proposed circuit occupies less area. Simulation results demonstrate that power consumption is reduced by a factor of 8.

Beyond CMOS, new technologies are emerging to extend electronic systems with features unavailable to silicon-based devices. Emerging technologies provide new logic and interconnection structures for computation, storage and communication that may require new design paradigms, and therefore trigger the development of a new generation of design automation tools. In the last decade, several emerging technologies have been proposed and the time has come for studying new ad-hoc techniques and tools for logic synthesis, physical design and testing. The main goal of this project is developing a complete synthesis and optimization methodology for switching nano-crossbar arrays that leads to the design and construction of an emerging nanocomputer. New models for diode, FET, and four-terminal switch based nanoarrays are developed. The proposed methodology implements logic, arithmetic, and memory elements by considering performance parameters such as area, delay, power dissipation, and reliability. With combination of logic, arithmetic, and memory elements a synchronous state machine (SSM), representation of a computer, is realized. The proposed methodology targets variety of emerging technologies including nanowire/nanotube crossbar arrays, magnetic switch-based structures, and crossbar memories. The results of this project will be a foundation of nano-crossbar based circuit design techniques and greatly contribute to the construction of emerging computers beyond CMOS. The topic of this project can be considered under the research area of “Emerging Computing Models” or “Computational Nanoelectronics”, more specifically the design, modeling, and simulation of new nanoscale switches beyond CMOS.

Beyond CMOS, new technologies are emerging to extend electronic systems with features unavailable to silicon-based devices. Emerging technologies provide new logic and interconnection structures for computation, storage and communication that may require new design paradigms, and therefore trigger the development of a new generation of design automation tools. In the last decade, several emerging technologies have been proposed and the time has come for studying new ad-hoc techniques and tools for logic synthesis, physical design and testing. The main goal of this project is developing a complete synthesis and optimization methodology for switching nano-crossbar arrays that leads to the design and construction of an emerging nanocomputer. New models for diode, FET, and four-terminal switch based nanoarrays are developed. The proposed methodology implements both arithmetic and memory elements, necessitated by achieving a computer, by considering performance parameters such as area, delay, power dissipation, and reliability. With combination of arithmetic and memory elements a synchronous state machine (SSM), representation of a computer, is realized. The proposed methodology targets variety of emerging technologies including nanowire/nanotube crossbar arrays, magnetic switch-based structures, and crossbar memories. The results of this project will be a foundation of nano-crossbar based circuit design techniques and greatly contribute to the construction of emerging computers beyond CMOS. The topic of this project can be considered under the research area of "Emerging Computing Models" or "Computational Nanoelectronics", more specifically the design, modeling, and simulation of new nanoscale switches beyond CMOS.

International audience; This work discusses the design and the main results relevant to the characterization of analogfront-end processors in view of their operation in the pixel detector readout chips of ATLAS andCMS at the High-Luminosity LHC. The front-end channels presented in this paper are part ofRD53A, a large scale demonstrator designed in a 65 nm CMOS technology by the RD53 collaboration. The collaboration is now developing the full-sized readout chips for the actual experiments. Some details on the improvements implemented in the analog front-ends are provided inthe paper.