We survey recent developments in the design of large-capacity content-addressable memory (CAM). A CAM is a memory that implements the lookup-table function in a single clock cycle using dedicated comparison circuitry. CAMs are especially popular in network routers for packet forwarding and packet classification, but they are also beneficial in a variety of other applications that require high-speed table lookup. The main CAM-design challenge is to reduce power consumption associated with the large amount of parallel active circuitry, without sacrificing speed or memory density. In this paper, we review CAM-design techniques at the circuit level and at the architectural level. At the circuit level, we review low-power matchline sensing techniques and searchline driving approaches. At the architectural level we review three methods for reducing power consumption.

https://www.pagiamtzis.com/pubs/pagiamtzis-jssc2006.pdf

Content-addressable memory (CAM) circuits and architectures: a tutorial and survey

The new era of cognitive computing brings forth the grand challenge of developing systems capable of processing massive amounts of noisy multisensory data. This type of intelligent computing poses a set of constraints, including real-time operation, low-power consumption and scalability, which require a radical departure from conventional system design. Brain-inspired architectures offer tremendous promise in this area. To this end, we developed TrueNorth, a 65 mW real-time neurosynaptic processor that implements a non-von Neumann, low-power, highly-parallel, scalable, and defect-tolerant architecture. With 4096 neurosynaptic cores, the TrueNorth chip contains 1 million digital neurons and 256 million synapses tightly interconnected by an event-driven routing infrastructure. The fully digital 5.4 billion transistor implementation leverages existing CMOS scaling trends, while ensuring one-to-one correspondence between hardware and software. With such aggressive design metrics and the TrueNorth architecture breaking path with prevailing architectures, it is clear that conventional computer-aided design (CAD) tools could not be used for the design. As a result, we developed a novel design methodology that includes mixed asynchronous–synchronous circuits and a complete tool flow for building an event-driven, low-power neurosynaptic chip. The TrueNorth chip is fully configurable in terms of connectivity and neural parameters to allow custom configurations for a wide range of cognitive and sensory perception applications. To reduce the system’s communication energy, we have adapted existing application-agnostic very large-scale integration CAD placement tools for mapping logical neural networks to the physical neurosynaptic core locations on the TrueNorth chips. With that, we have successfully demonstrated the use of TrueNorth-based systems in multiple applications, including visual object recognition, with higher performance and orders of magnitude lower power consumption than the same algorithms run on von Neumann architectures. The TrueNorth chip and its tool flow serve as building blocks for future cognitive systems, and give designers an opportunity to develop novel brain-inspired architectures and systems based on the knowledge obtained from this paper.

TrueNorth: Design and Tool Flow of a 65 mW 1 Million Neuron Programmable Neurosynaptic Chip

A fully integrated, programmable mixed-signal transceiver comprising a radio frequency integrated circuit (RFIC) which is frequency and protocol agnostic with digital inputs and outputs, the transceiver being programmable and configurable for multiple radio frequency bands and standards and being capable of connecting to many networks and service providers. The RFIC does not use spiral inductors and instead includes transmission line inductors allowing for improved scalability. Components of the transceiver are programmable to allow the transceiver to switch between different frequency bands of operating. Frequency switching can be accomplished though the content of digital registers coupled to the components.

Programmable radio transceiver

An integrated circuit device includes: a RAM block including a plurality of wordlines, a plurality of bitlines, a plurality of memory cells, and a wordline control circuit; and a data line driver block which drives a plurality of data line groups of a display panel based on data supplied from the RAM block. The data line driver block includes first to Nth (N is an integer larger than one) divided data line driver blocks, each of the first to Nth divided data line driver blocks driving a different data line group of the data line groups. The wordline control circuit drives an identical wordline N times from among the wordlines in one horizontal scan period of the display panel. The first to Nth divided data line drivers are disposed along a first direction in which the bitlines extend.

Integrated circuit device and electronic instrument

Deep neural networks are efficient at learning from large sets of labelled data, but struggle to adapt to previously unseen data. In pursuit of generalized artificial intelligence, one approach is to augment neural networks with an attentional memory so that they can draw on already learnt knowledge patterns and adapt to new but similar tasks. In current implementations of such memory augmented neural networks (MANNs), the content of a network’s memory is typically transferred from the memory to the compute unit (a central processing unit or graphics processing unit) to calculate similarity or distance norms. The processing unit hardware incurs substantial energy and latency penalties associated with transferring the data from the memory and updating the data at random memory addresses. Here, we show that ternary content-addressable memories (TCAMs) can be used as attentional memories, in which the distance between a query vector and each stored entry is computed within the memory itself, thus avoiding data transfer. Our compact and energy-efficient TCAM cell is based on two ferroelectric field-effect transistors. We evaluate the performance of our ferroelectric TCAM array prototype for one- and few-shot learning applications. When compared with a MANN where cosine distance calculations are performed on a graphics processing unit, the ferroelectric TCAM approach provides a 60-fold reduction in energy and 2,700-fold reduction in latency for a single memory search operation. A compact ternary content-addressable memory cell, which is based on two ferroelectric field-effect transistors, can provide memory augmented neural networks with improved energy and latency performance compared with traditional approaches based on graphics processing units.

Ferroelectric ternary content-addressable memory for one-shot learning

This paper presents a hybrid-type TCAM architecture which can utilize the benefits of both NOR and NAND-type TCAM cells: high speed and low power. A hidden bank selection scheme is proposed to activate limited amount of cells during the search operation avoiding additional timing penalty. Match fine repeaters and sub-match fine scheme are used for fast NAND search operation. A test chip with 144-kb TCAM capacity is implemented using 0.1-/spl mu/m 1.2-V CMOS process to verify the proposed schemes. It shows 2.2 ns of match evaluation time on a 144-bit data search with 0.7 fJ/bit/search energy efficiency.

A 0.7-fJ/bit/search 2.2-ns search time hybrid-type TCAM architecture

Disclosed is a programmable impedance control circuit, comprising a voltage divider, the voltage divider comprising an MOS array supplied with a first voltage and an external resistance having an external impedance equal to N times said external resistance. The voltage divider outputs a second voltage. A reference voltage generator is provided for generating a third voltage corresponding to N/(N+M) times said first voltage as a reference voltage for said second voltage, and wherein M times internal impedance is used for N times external impedance (N=M or N≠M).

Programmable impedance control circuit

An impedance control circuit that reduces the impedance variance when an external impedance generated from an external resistor is matched to internal impedance. In one aspect, an impedance control circuit comprises an external resistor for establishing a first reference voltage; a comparator for comparing the first reference voltage with a second reference voltage and outputting an impedance corresponding to the result of the comparison; and a PMOS current source connected to a constant-voltage source and to the output of the comparator, wherein the PMOS current source generates a current that corresponds to the impedance of the comparator.

Impedance control circuit

The proliferation of high-performance mobile devices is growing exponentially around the world. The essential driving force behind these ubiquitous devices is the high-performance CPU. To satisfy ever growing consumer demand for higher performance, a high-speed and multicore CPU configuration is mandatory, while energy efficiency is equally important for longer battery life. Furthermore, in order to support the wide range of performance (compute-intensive tasks, as well as less intensive tasks) required by today's mobile devices, a heterogeneous dual-CPU configuration comprising a high-performance CPU and an energy-efficient CPU can be one of the most energy efficient solutions for accomplishing both high-intensity and low-intensity tasks.

28nm high- metal-gate heterogeneous quad-core CPUs for high-performance and energy-efficient mobile application processor

An output impedance control circuit of a semiconductor device. A first transistor is connected to a pad and a level controller controls a gate voltage of the first transistor in response to a voltage of the pad and a reference voltage. A MOS array is connected between the pad and a power supply voltage and supplies current to the pad in response to an impedance control code. A first control circuit generates the impedance control code in response to whether a voltage of the pad is converging to the reference voltage. A second control circuit controls a pull-up impedance of the output buffer circuit in response to the first impedance control code when a voltage of the pad is converging to the reference voltage.

Uk-Rae Cho

Papers

A 0.7-fJ/bit/search 2.2-ns search time hybrid-type TCAM architecture

Programmable impedance control circuit

Impedance control circuit

28nm high- metal-gate heterogeneous quad-core CPUs for high-performance and energy-efficient mobile application processor

Semiconductor device with programmable impedance control circuit