PhD student Kiumars Aryana is investigating the potential of implementing a new type of memory – phase change memory – to discover how to minimize power consumption and degradation of computing materials over time. Aryana’s findings could pave the way for tactics to be used in neuromorphic computing, a type of AI that mimics neural networks in the brain. It could revolutionize the way artificial intelligence, including everyday devices like Alexa and Siri, is used across the world.
PCM is a form of computer memory that stores data by changing the phase of a particular alloy. An efficient and durable PCM allows for increasingly faster processing speeds, which helps improve current industry standard AI speeds. This requires processing at an even faster rate than the human brain.
Faster processing will allow the computer to process information in a fraction of a second and be able to identify things quickly, just like the human brain does. As data processing capabilities advance, artificial intelligence has the potential to reach extended capabilities.
Although the technology we use today is sophisticated and complex, power consumption and memory efficiency are major limitations that have so far impeded artificial intelligence – and our collective ability to quickly solve problems – to reach the next level. Everyday computers have long-term storage and short-term RAM storage, and in order to save data for long-term use, it needs to be saved in long-term memory. Computers typically use transistors to transmit these electrical signals internally.
The use of two separate memory systems is limited, and researchers are trying to find a way to combine the two memory systems, eliminating the need for them to communicate with each other. To do this, Aryana and other PhD students in the physics department are experimenting with neuromorphic computing.
“We don’t have a section [in our brain] which stores memory and a [separate] section to process information,” Aryana said. “We process information between neurons. We are trying to create an in-memory processing engine where you process the information in them.
Phase-change memory uses heat to process and store data, replacing transistors with material that switches between two states to bridge RAM and long-term storage memory. These changes of state of the MCP material are large and rapid.
“What happens in these memory devices is that you melt the material and then you turn it off,” Aryana said. “That means the material goes up to about 600 [degrees] Celsius and then quickly drops back to room temperature on the order of nanoseconds, so it’s a very fast process.
The use of heat in phase change memory degrades the material and consumes a generous amount of energy. So the next step in developing effective technology is to make this process more efficient and sustainable.
“We seek to find material compositions and techniques in order to reduce the power consumption of devices, because one of the biggest limitations now associated with phase change memory is their high power consumption which stems from using heat to change phase,” Aryana said.
Consistent use of phase change memory affects material properties in the computer.
“[Phase-change memory] causes these materials to segregate inside our alloy,” Aryana said. “Over time, as you repeat this process over and over again, you will see that your material becomes separated. Now you have different elements inside a uniform alloy.
A major part of Aryana’s research is determining which material compositions are the most durable and minimize energy consumption. He discovered that an alloy known as silicon telluride has the lowest thermal conductivity, allowing it to insulate heat to miniscule thicknesses.
PhD student Md Shafkat Bin Hoque often collaborates with Aryana and also studies thermal management of electronic devices. Hoque studies how to remove heat from phase change systems. Hoque’s and Aryana’s research are important considerations in neuromorphic computing because they require computers to be faster, more efficient, and more durable.
“We are actively looking for ways to remove this heat from the transistors so they don’t ignite and operate efficiently for as long as possible,” Hoque said. “My goal in this research is to find different materials that can efficiently remove heat from hot spots.”
Although PCM research is destined to manifest itself in artificial intelligence, it is still in its infancy. According to Hoque, a lot of work remains to translate their work into creating a truly functional supercomputer.
“Supercomputers usually take a small part of the problem and then solve it,” Hoque said. “But if supercomputers get faster, they can solve the whole problem instead of a small part…and it’s possible to make them more energy-efficient and smaller in size to the point that they might become available on laptop or Mac.”
These discoveries have the potential to dramatically change the way we interact with artificial intelligence on a daily basis.
“It all depends on the technology and the devices, how fast that technology can process the data,” Aryana said. “If we can have a much faster processor that can do all of these calculations on the spot, you can make Alexa or Siri much faster because they can process information faster.”