Power/Performance Bits: Dec. 31

Approximate computing
With the potential to double efficiency and reduce energy consumption, Purdue University and NEC Laboratories America researchers are developing computers capable of “approximate computing” to perform calculations good enough for certain tasks that don’t require perfect accuracy.

The need for approximate computing is driven by a fundamental shift in the nature of computing workloads, and the need for new sources of efficiency, the researchers said. They reminded that computers were first designed to be precise calculators that solved problems where they were expected to produce an exact numerical value but the demand for computing today is driven by very different applications. At the same time, there is an explosion in digital data searched, interpreted, and mined by data centers. As such, a growing number of applications are designed to tolerate “noisy” real-world inputs and use statistical or probabilistic types of computations.

The nature of these computations is different from the traditional computations where a precise answer is needed. Instead, since there is no golden answer, the best match is sought to provide results that are of acceptable quality – but not perfect.

Computers today are designed to compute precise results even when it is not necessary therefore Purdue researchers have developed a range of hardware techniques to demonstrate approximate computing, showing a potential for improvements in energy efficiency. The researchers have shown how to apply approximate computing to programmable processors and have shown how to design a programmable processor to perform approximate computing.

They achieved this by altering the instruction set between software and hardware. Quality fields added to the instruction set allow the software to tell the hardware the level of accuracy needed for a given task. They have created a prototype programmable processor called Quora based on this approach.

The secret to superconductivity
With the potential to advance current superconductor technology including quantum computers, MRI, high-precision magnetometry, levitating high-speed trains, and lossless power lines, University of British Columbia researchers have discovered a prevalent electronic state that controls the behavior of high-temperature superconducting copper-oxide ceramics. This state also reveals the universal existence of so-called “charge-density-waves,” which are static ripples formed by the self-organization of electrons in the material’s normal state. The ripples create the conditions for superconductivity.

The understanding of superconductivity in the cuprate family has been hindered by the diversity of intertwining electronic orders, according to the researchers. The findings suggest the existence of a universal charge-ordering that is common to all cuprate materials, and uncover its connection to the emergence of superconducting behavior.

They believe the work also proves that two techniques — resonant X-ray scattering or scanning tunneling microscopy — can be used interchangeably to probe the mysteries of charge-density-waves. These are fundamental, but very subtle, features which leave only a faint spectroscopic fingerprint and the success in detecting these tiny ripples in the electron distribution demonstrates the far-reaching power of these complementary techniques, and their pivotal role in advancing the understanding of quantum materials.

Superconductivity is the phenomenon of electricity flowing with no resistance, which occurs in some materials at very low temperatures. High-temperature cuprate superconductors are capable of conducting electricity without resistance at record temperatures, higher than the boiling point of liquid nitrogen. Because of their unrivalled characteristics, they are believed to be the best candidates to advance current superconductor technology.