Power/Performance Bits: Aug. 4

Assessing code similarity
Researchers from Intel, MIT, and Georgia Institute of Technology created an automated engine designed to learn what a piece of software intends to do by studying the structure of the code and analyzing syntactic differences of other code with similar behavior.

The machine inferred code similarity (MISIM) program, a subset of Intel’s work on machine programming, was driven by the rise of heterogeneous computing and the difficulty of finding programmers who can code at an expert level across multiple architectures. Machine programming aims to improve development productivity through the use of automated tools.

“Intel’s ultimate goal for machine programming is to democratize the creation of software. When fully realized, MP will enable everyone to create software by expressing their intention in whatever fashion that’s best for them, whether that’s code, natural language or something else. That’s an audacious goal, and while there’s much more work to be done, MISIM is a solid step toward it,” said Justin Gottschlich, principal scientist and director/founder of Machine Programming Research at Intel.

MISIM can accurately determine when two pieces of code perform a similar computation, even when those pieces use different data structures and algorithms. “This is an important step toward the grander vision of machine programming,” Gottschlich said.

Differentiating it from other code-similarity systems, MISIM uses a context-aware semantic structure (CASS), which aims to lift out what the code actually does. Intel says that unlike other existing approaches, CASS can be configured to a specific context, allowing it to capture information that describes the code at a higher level and provide more specific insight into what the code does rather than how it does it.

Once the code’s structure is integrated into CASS, a neural network gives similarity scores to pieces of code based on the jobs they are designed to carry out, even if they are different in structure. The researchers found that MISIM was able to identify similar pieces of code up to 40x more accurately than prior state-of-the-art systems.

The team is continuing work on MISIM and on developing the ability to recognize the intent behind a simple algorithm input by a developer and offer candidate codes that are semantically similar but with improved performance. They are also working with software groups at Intel to see how it could be deployed in a day-to-day environment.

Shadow harvesting
Researchers at the National University of Singapore developed a way to harness shadows for energy generation. The ‘shadow-effect energy generator’ (SEG) makes use of the contrast in illumination between lit and shadowed areas.

“Shadows are omnipresent, and we often take them for granted. In conventional photovoltaic or optoelectronic applications where a steady source of light is used to power devices, the presence of shadows is undesirable, since it degrades the performance of devices. In this work, we capitalized on the illumination contrast caused by shadows as an indirect source of power. The contrast in illumination induces a voltage difference between the shadowed and illuminated sections, resulting in an electric current. This novel concept of harvesting energy in the presence of shadows is unprecedented,” said Tan Swee Ching, an assistant professor in the NUS Department of Materials Science and Engineering.

The shadow-effect energy generator contains a set of SEG cells arranged on a flexible and transparent plastic film. Each SEG cell is a thin film of gold deposited on a silicon wafer. The SEG can be fabricated at a lower cost compared to commercial silicon solar cells.

“When the whole SEG cell is under illumination or in shadow, the amount of electricity generated is very low or none at all. When a part of the SEG cell is illuminated, a significant electrical output is detected. We also found that the optimum surface area for electricity generation is when half of the SEG cell is illuminated and the other half in shadow, as this gives enough area for charge generation and collection respectively,” said Andrew Wee, a professor from the NUS Department of Physics.

Based on laboratory experiments, the team’s four-cell SEG is twice as efficient when compared with commercial silicon solar cells under the effect of shifting shadows. The harvested energy from the SEG in the presence of shadows created under indoor lighting conditions is sufficient to power a digital watch (1.2 V).

In addition, the team also showed that the SEG can serve as a self-powered sensor for monitoring moving objects. When an object passes by the SEG, it casts an intermittent shadow on the device and triggers the sensor to record the presence and movement of the object.

In the next phase of research, the team will experiment with other materials, besides gold, to reduce the cost of the SEG.