Power density continues to increase exponentially with each new technology generation, posing a major challenge for thermal management in modern processors. Much past work has examined microarchitectural policies for reducing total chip power, but these techniques alone are insufficient if not aimed at mitigating individual hotspots. The industry's current trend has been toward multicore architectures, which provide additional opportunities for dynamic thermal management. This paper explores various thermal management techniques that exploit the distributed nature of multicore processors. We classify these techniques in terms of core throttling policy, whether that policy is applied locally to a core or to the processor as a whole, and process migration policies. We use Turandot and a HotSpot-based thermal simulator to simulate a variety of workloads under thermal duress on a 4-core PowerPCTMprocessor. Using benchmarks from the SPEC 2000 suite we characterize workloads in terms of instruction throughput as well as their effective duty cycles. Among a variety of options we find that distributed controltheoretic DVFS alone improves throughput by 2.5X under our test conditions. Our final design involves a PI-based core thermal controller and an outer control loop to decide process migrations. This policy avoids all thermal emergencies and yields an average of 2.6X speedup over the baseline across all workloads.

http://bnrg.eecs.berkeley.edu/~randy/Courses/CS294.F07/5.1.pdf

Techniques for Multicore Thermal Management: Classification and New Exploration

Power systems for modern complementary metal-oxide-semiconductor (CMOS) technology are becoming harder to design. One design methodology is to identify a target impedance to be met across a broad frequency range and specify components to meet that impedance. The impedance versus frequency profiles of the power distribution system components including the voltage regulator module, bulk decoupling capacitors and high frequency ceramic capacitors are defined and reduced to simulation program with integrated circuit emphasis (SPICE) models. A sufficient number of capacitors are placed in parallel to meet the target impedance. Ceramic capacitor equivalent series resistance (ESR) and ESL are extremely important parameters in determining how many capacitors are required. SPICE models are then analyzed in the time domain to find the response to load transients.

/pdf/power-distribution-system-design-methodology-and-capacitor-lhl3j5lhjo.pdf

Power distribution system design methodology and capacitor selection for modern CMOS technology

In this paper we compare the performance of area equivalent small, medium, and large-scale multithreaded chip multiprocessors (CMTs) using throughput-oriented applications. We use area models based on SPARC processors incorporating these architectural features. We examine CMTs with in-order scalar processor cores, 2-way or 4-way in-order superscalar cores, private primary instruction and data caches, and a shared secondary cache. We explore a large design space, ranging from processor-intensive to cache-intensive CMTs. We use SPEC JBB2000, TPC-C, TPC-W, and XML Test to demonstrate that the scalar simple-core CMTs do a better job of addressing the problems of low instruction-level parallelism and high cache miss rates that dominate Web service middleware and online transaction processing applications. For the best overall CMT performance, smaller cores with lower performance, so called "mediocre" cores, maximize the total number of CMT cores and outperform CMTs built from larger, higher performance cores.

/pdf/maximizing-cmp-throughput-with-mediocre-cores-1sms4ar9nn.pdf

Maximizing CMP throughput with mediocre cores

Power and ground planes are required to have low impedance over a wide range of frequencies. Parallel ground and power planes in multilayer printed-circuit boards exhibit multiple resonances, which increase the impedance and also the radiation from the edge of the board. Resistive termination along the board edges reduces the resonance peaks. Simple and straightforward design expressions, simulated self and transfer impedances, as well as measured impedance plots are presented for power-ground planes.

Reducing simultaneous switching noise and EMI on ground/power planes by dissipative edge termination

A method, apparatus, and system for providing operating power and transient suppression power to a microelectronic device are disclosed. The system includes a primary regulator to supply nominal operating power and to respond to relatively slow transient events and a transient suppression regulator to respond to fast transient power events. The system also includes a sense circuit to detect when a transient event occurs and to send a signal to the transient suppression regulator to supply or sink current to the load in response to a sensed transient power event.

Method and apparatus for providing wideband power regulation to a microelectronic device

Complementary metal-oxide-semiconductor (CMOS) microprocessors operating in the hundreds of megahertz create significant current deltas due to the variation in switching activity front clock cycle to clock cycle. In addition to the high-frequency voltage variations more commonly discussed, a lower frequency noise component is also produced that lasts from 50-200 ns which we refer to as mid-frequency noise. In this paper, we discuss the design of IBM's CMOS S/390 computer for control of mid-frequency noise. This machine has a 10-way multiprocessor on a 127 mm by 127 mm multichip module (MCM) on a FR4 board. The chips on the MCM cause a current step of tens of Amps in a few cycles that can be sustained for many cycles. The power distribution and decoupling capacitors must supply that current without disturbing the voltage level at the circuits. The design of the system power distribution and modeling and verification of mid-frequency noise in this system is presented.

Modeling, simulation, and measurement of mid-frequency simultaneous switching noise in computer systems

A method and apparatus for generating a voltage that is proportional to an absolute temperature (PTAT voltage). A current generator for generating a current that is proportional to absolute temperature (PTAT current) has an internal resistance and two diodes. The PTAT current is proportional to the resistance, and the temperature coefficient of the PTAT current is defined by the ratio of diode current densities. A feedback circuit has a source follower that is connected to the current generator for driving the output node with a regulated PTAT current wherein the PTAT current is mirrored accurately, providing a constant Vref.

Power Supply Insensitive PTAT Voltage Generator

offers significantly increased performance over previous POWER designs by incorporating simultaneous multithreading, an enhanced memory subsystem, and extensive RAS and power management support. The 276M transistor processor is implemented in 130nm silicon-on-insulator technology with 8-level of Cu metallization and operates at >1.5 GHz.

Design and Implementation of the POWER5 TM Microprocessor

CMOS microprocessors operating in the hundreds of megahertz create significant current deltas due to the variation in switching activity from clock cycle to clock cycle. In addition to the high-frequency voltage variations more commonly discussed, a lower frequency noise component is also produced that lasts from 50 to 200 ns which we refer to as mid-frequency noise. In this paper, we discuss the design of IBM's CMOS S/390 computer for control of mid-frequency noise. This machine has a 10-way multiprocessor on a 127 mm by 127 mm MCM on a FR4 board. The chips on the MCM cause a current step of tens of Amps in a few cycles that can be sustained for many cycles. The power distribution and decoupling capacitors must supply that current without disturbing the voltage level at the circuits. The design of the system power distribution and modeling and verification of mid-frequency noise in this system is presented.

Mid-frequency simultaneous switching noise in computer systems

An integrated circuit having a temperature sensitive circuit (TSC) to generate a signal indicative of the substrate temperature near the TSC The integrated circuit has circuitry configured to receive a TSC signal from at least one TSC and to convert the TSC signal to a signal indicative of the integrated circuit's temperature The thermal control circuit compares the integrated circuit temperature to a threshold and produces a corrective action signal when the temperature exceeds the threshold The corrective action signal is provided to corrective action circuitry preferably configured to modify the operation of the IC to reduce the IC temperature in proximity to the corresponding TSC

Paul D. Muench

Papers

Modeling, simulation, and measurement of mid-frequency simultaneous switching noise in computer systems

Power Supply Insensitive PTAT Voltage Generator

Design and Implementation of the POWER5 TM Microprocessor

Mid-frequency simultaneous switching noise in computer systems

Thermally aware integrated circuit