Power/Performance Bits: May 21

More speculative vulnerabilities
Security researchers at the Graz University of Technology, KU Leuven, Cyberus Technology, and Worcester Polytechnic Institute point to two new speculative execution vulnerabilities related to Meltdown and Spectre.

The first, which they dubbed ZombieLoad, uses a similar approach to Meltdown. After preparing tasks in parallel, the processor needs to discard the ones it won’t be using or hasn’t been granted permissions for. Due to the way processors are designed, they always have to pass on data, even if it is not correct. The check for permission only happens once sensitive processing steps, which depend on assumptions made by the computer system, have already been prepared. “In the split second between the command and the check, using this new form of attack we can see the pre-loaded data from other programs,” said Daniel Gruss, a security researcher at TU Graz. In other words, the researchers can read what the computer is currently processing.

The team was part of the effort to deploy a patch for the Meltdown vulnerability, but say this new potential attack could be more difficult to remedy. “Every CPU has multiple cores, and each of these cores is also split in two. This means several programs can run simultaneously. We think that one of these two parts of each core has to be disabled,” Gruss said. That would mean a 50% drop in performance. Or in clouds, which are also vulnerable to this method of attack, 50% fewer potential users on the same hardware.

According to the researchers, all processors developed by Intel that were manufactured between 2012 and the beginning of 2018 are affected.

The second vulnerability, store-to-leak forwarding, also reads pre-loaded data by exploiting the efficient way in which computer processors function. “The computer assumes that I want to use the data which I have just written to the processor again right away. So it keeps it in the buffer for faster access,” explained Gruss. This functionality can also be used to determine the architecture of the computer processor and find the exact location where the operating system is running. “If I know exactly where the processor is running the operating system, then I can launch targeted attacks against flaws in the operating system.”

Intel is currently working on solutions to these vulnerabilities.

Stretchy supercapacitors
Researchers at Michigan State University, Duke University, U.S. Naval Research Laboratory, Huazhong University of Science and Technology, and MIT created highly stretchable supercapacitors for powering wearable electronics using forests of crumpled carbon nanotubes.

The new supercapacitor demonstrated solid performance and stability, even when it is stretched to 800% of its original size for thousands of stretching/relaxing cycles.

“The key to success is the innovative approach of crumpling vertically aligned CNT arrays, or CNT forests,” said Changyong Cao, director of MSU’s Soft Machines and Electronics Laboratory and an assistant professor of mechanical engineering and electrical and computer engineering at MSU. “Instead of having a flat thin film strictly constrained during fabrication, our design enables the three-dimensionally interconnected CNT forest to maintain good electrical conductivity, making it much more efficient, reliable and robust.”

For emerging wearable tech to advance, it needs improved power sources. Now researchers from MSU have provided a potential solution via crumpled carbon nanotube forests, or CNT forests. (Image courtesy of MSU)

Plus, the 3D interconnected CNT forest has a larger surface area and can be easily modified with nanoparticles or adapted to other designs.

“It’s more robust; it’s truly a design breakthrough,” said Cao. “Even when it’s stretched up to 300% along each direction, it still conducts efficiently. Other designs lose efficiency, can usually be stretched in only one direction or malfunction completely when they are stretched at much lower levels.”

While the crumpled carbon nanotube forest design performed better than most other CNT-based supercapacitors, the team sees room for improvement by impregnating the design with metal oxide nanoparticles to improve the specific capacitance and energy density.