Vincent R. von Kaenel

A high-speed, low-power clock generator for a microprocessor application

This paper discusses the design of the clock generator for the Alpha 21264. As the speed performances are of primary concern in the whole design, the clock-generator jitter and phase misalignment must be as low as possible in a very noisy environment. A dedicated on-chip voltage regulator based on a bandgap reference has been designed to reduce the effect of supply noise on the clock generator. To avoid a large voltage drop across the power-supply bond wires during the startup sequence, the core frequency can be increased by steps in one period of the core clock, with a limited frequency overshoot and no missing pulses. The circuit has been implemented in a CMOS 0.35 /spl mu/m process. The voltage-controlled-oscillator frequency range is between 350 MHz and 2.8 GHz, with a peak-to-peak cycle-to-cycle jitter lower than 16 ps. While booting Unix on a system, the maximum phase misalignment is lower than /spl plusmn/100 ps.

A 4-GHz clock system for a high-performance system-on-a-chip design

A digital systems clocks must have not only low jitter, but also well-controlled duty cycles in order to facilitate versatile clocking techniques. Power-supply noise is often the most common and dominant source of jitter on a phase-locked loops (PLL) output clock. Jitter can be minimized by regulating the supply to the PLLs noise-sensitive analog circuit blocks in order to filter out supply noise. This paper introduces a PLL-based clock generator intended for use in a high-speed highly integrated system-on-a-chip design. The generator produces clocks with accurate duty cycles and phase relationships by means of a high-speed divider design. The PLL also achieves a power-supply rejection ratio (PSRR) greater than 40 dB while operating at frequencies exceeding 4 GHz. The high level of noise rejection exceeds that of earlier designs by using a combination of both passive and active filtering of the PLLs analog supply voltage. The PLL system has been integrated in a 0.15-/spl mu/m single-poly 5-metal digital CMOS technology. The measured performance indicates that at a 4-GHz output frequency the circuit achieves a PSRR greater than 40 dB. The peak cycle-to-cycle jitter is 25 ps at 700 MHz and a 2.8-GHz VCO frequency with a 500-mV step on the regulators 3.3-V supply. The total power dissipated by the prototype is 130 mW and its active area is 1.48/spl times/1.00 mm/sup 2/.

A low-cost 300 MHz RISC CPU with attached media processor

This custom CPU derived from the StrongARM/sup TM/ 110 is capable of more than 2 billion 16 b operations per second (2 BOPs). Starting with the original design, an attached media processor (AMP) is integrated along with a synchronous DRAM memory controller and separate I/O bus. In addition, several enhancements are made to the CPU and cache subsystem and the chip is reduced from 0.35 /spl mu/m to 0.28 /spl mu/m technology. The chip includes 3.3M transistors and measures 60 mm/sup 2/. It dissipates less than 3 W at 300 MHz at 2.0 V internal, 3.3 V I/O. The chip supports dynamic clock frequency switching for reduced operating power during low performance demands. There are 333 separately conditioned clocks on the chip. For battery powered applications, Vdd is reduced to achieve <0.5 W operation at 150 MHz. The chip is pseudo-static and supports clock stop and IDDQ testing.

Power-Efficient Dual-Supply 64kB L1 Caches in a 65nm CMOS Technology

The 64 kB LI caches on the PA6T-1682M SoC CPU are composed of common data and tag structures fabricated in a 65 nm CMOS process and deliver a 1.5 cycle read latency with 32 GB/s bandwidth at 2 GHz. Several features optimize cache performance and power including power-down safe level shifters, streamlined dual-supply bitslices, fine-grain clock gating, and a centralized tag floorplan.