Microsoft Research has revealed that it has been working on a computer for the past three years that uses photons and electrons, rather than transistors, to process data. The computer, called the Analog Iterative Machine (AIM), is designed to get around the limits of Moore's Law, which states that the number of transistors that can be placed on an integrated circuit doubles every two years.
The AIM uses a combination of analog optical and electronic technologies to perform calculations. Photons are used to carry information, while electrons are used to control the flow of photons. This allows the AIM to perform calculations much faster than traditional computers, which are limited by the speed of electrons.
In a blog post, Microsoft researchers say that the AIM could be used to solve a variety of problems, including image recognition, natural language processing, and drug discovery. They also say that the AIM could be used to create new types of artificial intelligence that are not possible with traditional computers.
“Analog optical computing thus involves constructing a physical system using a combination of analog technologies – both optical and electronic – governed by equations that capture the required computation. This can be very efficient for specific application classes where linear and non-linear operations are dominant. In optimization problems, finding the optimal solution is akin to discovering a needle in an inconceivably vast haystack. The team has developed a new algorithm that is highly efficient at such needle-finding tasks. Crucially, the algorithm's core operation involves performing hundreds of thousands or even millions of vector-matrix multiplications – the vectors represent the problem variables whose values need to be determined while the matrix encodes the problem itself.”
What are the Potential Benefits of the Analog Iterative Machine
- Speed: AIMs can perform calculations much faster than traditional digital computers. This is because photons can travel much faster than electrons, and AIMs can process data in parallel.
- Energy efficiency: AIMs require less power to operate than traditional digital computers. This is because photons require less power than electrons.
- Scalability: AIMs can be scaled up to much larger sizes than traditional digital computers. This means that they could be used to solve even larger and more complex problems.
Ongoing Development and Search for Applications
According to Microsoft, the AIM can achieve a speedup of up to 1000 times over state-of-the-art digital computers for some tasks, such as finding the optimal route for a delivery truck or training a neural network. The device can also simulate quantum systems with up to 100 qubits, which is beyond the reach of current quantum computers.
Microsoft says that AIM is still in the early stages of development and that it plans to explore its potential applications in various domains, such as logistics, healthcare, finance, and artificial intelligence. The company also hopes that the AIM will inspire new research directions in analog and optical computing.
Microsoft Research has been making gains in machine learning development in recent months. In March, the company revealed Azure Machine Learning Foundation Models from Hugging Face.
Users can leverage foundation models from various open source repositories (Hugging Face is just one) to help improve their foundation models and then deploy them. Capabilities include being able to use a pre-trained model for inference and deployment, as well as machine learning-supported tasks from customer data.