Stanley E. Schuster

Power-aware microarchitecture: design and modeling challenges for next-generation microprocessors

The ability to estimate power consumption during early-stage definition and trade-off studies is a key new methodology enhancement. Opportunities for saving power can be exposed via microarchitecture-level modeling, particularly through clock-gating and dynamic adaptation. In this paper we describe the approach of using energy-enabled performance simulators in early design. We examine some of the emerging paradigms in processor design and comment on their inherent power-performance characteristics.

1 µm MOSFET VLSI technology: Part IV—Hot-electron design constraints

An approach is described for determining the hot-electron-limited voltages for silicon MOSFETs of small dimensions. The approach was followed in determining the room-temperature and the 77 K hot-electron-limited voltages for a device designed to have a minimum channel length of 1 µm. The substrate hot-electron limits were determined empirically from measurements of the emission probabilities as a function of voltage using devices of reentrant geometry. The channel hot-electron limits were determined empirically from measurements of the injection current as a function of voltage and from long-term stress experiments. For the 1 µm design considered, the channel hot-electron limits are lower than the substrate hot-electron limits. The maximum voltage,V_{DS} = V_{GS}, is 4.75 V at room temperature (25°C) and 3.5 V at 77 K. More details of the voltage limits as well as the approach for determining them are discussed. Examples of circuits designed with these devices to operate within these hot-electron voltage limits are also discussed.

Multiple word/bit line redundancy for semiconductor memories

Multiple word/bit line redundancy techniques at the chip level are shown to be powerful enough to obtain good yields for chips with much higher rates of faults/chip than without redundancy. This is possible because, in many instances, chips which are rejected as being bad still have a high percentage of usable bits on them. The redundancy techniques described consist of putting spare decoders and spare word and bit lines on a chip in order to be able to replace defective lines of the chip with good lines while still maintaining the same address. Based on a first-pass design of a 16K chip, a significant improvement in the number of usable bits per wafer appears possible. The leverage for improvement is shown to be strongly dependent upon the type of cell, the layout, and the technology used.

Dynamically tuning processor resources with adaptive processing

By using adaptive processing to dynamically tune major microprocessor resources, developers can achieve greater energy efficiency with reasonable hardware and software overhead while avoiding undue performance loss. Adaptive processors require few additional transistors. Further, because adaptation occurs only in response to infrequent trigger events, the decision logic can be placed into a low-leakage state until such events occur.

An Adaptive Issue Queue for Reduced Power at High Performance

Increasing power dissipation has become a major constraint for future performance gains in the design of microprocessors. In this paper, we present the circuit design of an issue queue for a superscalar processor that leverages transmission gate insertion to provide dynamic low-cost configurability of size and speed. A novel circuit structure dynamically gathers statistics of issue queue activity over intervals of instruction execution. These statistics are then used to change the size of an issue queue organization on-the-fly to improve issue queue energy and performance. When applied to a fixed, full-size issue queue structure, the result is up to a 70% reduction in energy dissipation. The complexity of the additional circuitry to achieve this result is almost negligible. Furthermore, self-timed techniques embedded in the adaptive scheme can provide a 56% decrease in cycle time of the CAM array read of the issue queue when we change the adaptive issue queue size from 32 entries (largest possible) to 8 entries (smallest possible in our design).

A 2-ns cycle, 3.8-ns access 512-kb CMOS ECL SRAM with a fully pipelined architecture

The authors describe a 512 K CMOS static RAM (SRAM) with emitter-coupled-logic (ECL) interfaces which has a 2-ns cycle time and a 3.8-ns access time, both of which are valid for random READ/WRITE operations. The CMOS technology and the physical organization of the chip are briefly discussed, and the pipelined architecture of the chip is described. Detailed measurements of internal chip waveforms demonstrating 2-ns cycle time operation are presented. The impact of wire RC delays on performance is discussed. Circuit examples that demonstrate the implementation of the pipelined architecture are included. Measurements of operating margins, access time, and cycle time are outlined. >

Synchronous interlocked pipelines

Locality principles are becoming paramount in controlling advancement of data through pipelined systems. Achieving fine grained power down and progressive pipeline stalls at the local stage level is therefore becoming increasingly, important to enable lower dynamic power consumption while keeping introduced switching noise under control as well as avoiding global distribution of timing critical stall signals. It has long been known that the interlocking properties of as asynchronous pipelined systems have a potential to provide such benefits. However it has not been understood how such interlocking can be achieved in synchronous pipelines. This paper presents a novel technique based on local clock gating and synchronous handshake protocols that achieves stage level interlocking characteristics in synchronous pipelines similar to that of asynchronous pipelines. The presented technique is directly applicable to traditional synchronous pipelines and works equally well for two-phase clocked pipelines based on transparent latches, as well as one-phase clocked pipelines based on master-slave latches.

A circuit level implementation of an adaptive issue queue for power-aware microprocessors

Increasing power dissipation has become a major constraint for future performance gains in the design of microproces sors In this paper we present the circuit design of an issue queue for a superscalar processor that leverages transmis sion gate insertion to provide dynamic low cost con gura bility of size and speed A novel circuit structure dynami cally gathers statistics of issue queue activity over intervals of instruction execution These statistics are then used to change the size of an issue queue organization on the y to improve issue queue energy and performance When applied to a xed full size issue queue structure the result is up to a reduction in energy dissipation The complexity of the additional circuitry to achieve this result is almost neg ligible Furthermore self timed techniques embedded in the adaptive scheme can provide a decrease in cycle time of the CAM array read of the issue queue when we change the adaptive issue queue size from entries largest possible to entries smallest possible in our design

Fast CMOS ECL receivers with 100-mV worst-case sensitivity

CMOS emitter-coupled logic (ECL) receiver circuits consisting of a differential-amplifier stage and a CMOS inverter are shown to convert 100-mV input signals to on-chip CMOS levels even with worst-case parameter variations in a 5-V 1- mu m technology. Two different receiver circuits are used to cover a range of power supply options; a third circuit provides a comparison case. The differential amplifiers feature built-in feedback compensation for common-mode parameter variations. The differential input devices are designed with large widths, minimum channel lengths, and an interleaved layout to enhance gain, speed, and margin for differential mismatches. The simplicity of the circuits and the effectiveness of the built-in compensation facilitate analysis. Partitioning and simplifying assumptions are used to thoroughly test the worst case without complex simulations, while providing insight into the design process. >

Asynchronous interlocked pipelined CMOS circuits operating at 3.3-4.5 GHz

Chip performance, power, noise, and clock synchronization are becoming formidable challenges as microprocessor performance moves into the GHz regime and beyond. Interlocked pipelined CMOS (IPCMOS), an asynchronous clocking technique, helps address these challenges. This paper shows how a typical block (e.g., Block D) is interlocked with all the blocks with which it interacts. In the forward direction, dedicated Valid signals emulate the worst-case path through each driving block and thus determine when data can be latched within the typical block. In the reverse direction, Acknowledge signals indicate that data has been received by the subsequent blocks and that new data may be processed within the typical block. In this interlocked approach local clocks are generated only when there is an operation to perform.