Power/Performance Bits: Feb. 21


Harvesting energy from multiple sources Researchers from the University of Oulu in Finland found a particular type of perovskite, KBNNO, has the right properties to extract energy from multiple sources simultaneously. While perovskites are particularly known for their use as solar cells, certain minerals in the perovskite family show piezoelectric and pyroelectric (harvesting energy from ... » read more

Power/Performance Bits: Jan. 31


Microbial nanowires Microbiologists at the University of Massachusetts Amherst report that they have discovered a new type of microbial nanowire, the protein filaments that bacteria use to make electrical connections with other microbes or minerals. The team was motivated by the potential for improved "green" conducting materials for electronics. According to Derek Lovley, professor of... » read more

Design and Analysis of Soft-Error ResilienceMechanisms for GPU Register File


Modern graphics processing units (GPUs) are using increasingly larger register file (RF) which occupies a large fraction of GPU core area and is very frequently accessed. This makes RF vulnerable to soft-errors (SE). In this paper, we present two techniques for improving SE resilience of GPU RF. First, we propose compressing the RF values for reducing the number of vulnerable bits. We leverage... » read more

Addressing Read-Disturbance Issue In STT-RAM By Data Compression And Selective Duplication


In deep sub-micron region, STT-RAM (spin transfer torque RAM) shows read-disturbance error (RDE) which presents acrucial reliability challenge. We present SHIELD, a technique to mitigate RDE in STT-RAM LLCs (last level caches). SHIELD uses datacompression to reduce cache-write traffic and restore requirement. Also, SHIELD keeps two copies of data blocks compressed to lessthan half the block si... » read more

Power/Performance Bits: Sept. 27


Self-organizing circuits Researchers studying the behavior of nanoscale materials at the Department of Energy's Oak Ridge National Laboratory discovered that due an unusual feature of certain complex oxides called phase separation, individual nanoscale regions can behave as self-organized circuit elements, which could support new multifunctional types of computing architectures. "Within a... » read more

Manufacturing Bits: Sept. 13


Direct-write liquid litho The Department of Energy’s Oak Ridge National Laboratory has developed what could be called direct-write liquid lithography. In the lab, researchers have modified a scanning transmission electron microscope (STEM). Then, using the STEM as an e-beam tool, researchers have devised a technology that enables the direct write of tiny features in “microfabricated liq... » read more

Manufacturing Bits: Aug. 9


Faster FEBIDs Focused electron beam induced deposition (FEBID) is generating steam in the industry. Still in the R&D stage, FEBID makes use of an electron beam from a scanning electron microscope. Basically, it decomposes gaseous molecules, which, in turn, deposit materials and structures on a surface at the nanoscale. One of the big applications is a futuristic manufacturing technology... » read more

Complete Control Through Software


As Industrial Internet of Things (IIoT) applications proliferate in critical infrastructure, such as the power grid and water supply, the importance of the underlying software and the availability of an open-source platform for app development is coming to the forefront. This marks a significant shift, particularly in the industrial and commercial world, where software historically has playe... » read more

Manufacturing Bits: June 21


Atomic sculpting Oak Ridge National Laboratory has combined a scanning transmission electron microscope (STEM) with new electronic controls. This tool enables the construction, or the atomic sculpting, of 3D-like feature sizes down to 1nm and 2nm. To achieve these dimensions, the STEM is controlled with a special set of programmable electronics. This, in turn, enables the STEM to tunnel in... » read more

Rethinking Memory


Getting data in and out of memory is as important as the speed and efficiency of a processor, but for years design teams managed to skirt the issue because it was quicker, easier and less expensive to boost processor clock frequencies with a brute-force approach. That worked well enough prior to 90nm, and adding more cores at lower clock speeds filled the gap starting at 65nm. After that, th... » read more

← Older posts