Concluding our series of articles on cutting-edge neuromorphic processor architectures, we decided to focus on domestic developments that often remain outside the public spotlight. In this article, we will discuss the “Altai” neuromorphic processor.

Did you know that the name Tianjic in Chinese carries a double meaning? On one hand, it signifies a "heavenly mechanism" or "sublime plan," and on the other, it symbolizes the union of different worlds. This reflects the essence of Tianjic: the integration of two artificial intelligence paradigms (ANN and SNN) into a single chip. In the first part of the article, we explored the history of Tianjic’s creation and its key technical solutions. Now it’s time to discover how the "heavenly" architecture of Tianjic performs in practice and the challenges it faces on the road to establishing the hardware foundations of AGI.

In our recent articles, we discussed SpiNNaker—a unique solution for simulating the operation of spiking neural networks. Today, we turn to another innovative approach: the Tianjic architecture, which brings together engineering solutions for both artificial neural networks (ANN) and spiking neural networks (SNN). Why focus on Tianjic? Unlike specialized solutions such as TrueNorth, Tianjic offers a universal hybrid approach capable of running diverse neural models on a single chip. This opens up new frontiers in the development of artificial general intelligence (AGI).

In the previous article, we explored the hardware architecture of SpiNNaker2 — from the fundamental Processing Element (PE) and its event-driven processing principles to scaling the system into a neuromorphic supercomputer. In this article, we continue our deep dive into SpiNNaker2 by focusing on its software ecosystem, examining key frameworks that simplify the development and testing of neural network models. We will also compare SpiNNaker2 with IBM’s TrueNorth and analyze real-world projects that showcase the potential of neuromorphic architectures in science, robotics, and machine learning.

In our previous article, we discussed TrueNorth — IBM's neuromorphic processor that proved the feasibility of energy-efficient data processing based on the principles of biological neural networks. However, the journey of neuromorphic computing doesn't end there. Other projects offer alternative approaches to creating a "computational brain." One such project is SpiNNaker — a scalable platform that combines parallelism with an event-driven signal processing paradigm. In this article, we’ll explore the evolution of SpiNNaker — from its early concepts to the modern SpiNNaker2 — and analyze the key architectural features that make this platform unique among neuromorphic systems.

Since the advent of the first computers, scientists have been striving to create machines capable of mimicking the workings of the human brain. One promising approach to achieving this goal is neuromorphic processors, which offer new methods of information processing. A striking example of such developments is TrueNorth — an experimental chip from IBM, designed based on the principles of biological neural networks. TrueNorth has demonstrated how a specialized neuromorphic architecture can enable highly efficient parallel processing with low power consumption, paving the way for new approaches in designing systems that handle large volumes of sensory data.

In our previous articles, we discussed classical processor architectures that laid the foundation for modern computers. However, technology does not stand still, and today we will look at a completely different approach to computing - neuromorphic architectures. These solutions, inspired by the principles of biological neural networks, open up new horizons for high-performance and energy-efficient systems.

As 2024 comes to a close, it has been a particularly special year for us — this is the year we opened the doors to our blog, creating a space for communication, knowledge sharing, and collective growth. We are sincerely grateful to each of you for becoming part of our community, supporting us, and inspiring us to achieve new milestones. In the coming year, we will continue to share interesting and valuable content with you. Stay tuned for updates!

In our previous article, we explored the CISC architecture and its impact on the development of computing technology. Today, we continue our exploration of alternative processor design approaches by discussing RISC (Reduced Instruction Set Computer). We will delve into its main features, examine its advantages and disadvantages, and look at how it is applied in modern systems.

In the world of computing technology, where speed and efficiency are becoming key factors of progress, processor architectural decisions play an important role. One of the significant approaches to processor design is CISC (Complex Instruction Set Computer) — an architecture whose roots trace back to the dawn of modern computers. In this article, we will delve into the history of CISC development, explore its strengths and weaknesses, and discuss the role of this architecture in contemporary computing technology.