🤖 IBM’s Brain-Inspired Computer Chip

🧠IBM’s Brain-Inspired Computer Chip

🇨🇳China Companies Invest $340M in OpenAI Challenger

🛠5 Trending Tools

💰Venture Capital Updates

💼Who’s Hiring?

The NorthPole processor chip eliminates the need to frequently access external memory, allowing users to perform tasks faster than existing architectures with less computing power.

NorthPole functions on neural networks (i.e., multi-layered arrays of simple computational units programmed to recognize patterns in data). The bottom layer processes data like pixels in an image. Next, each successive layer detects complex patterns and communicates them to the following layer. Eventually, the top layer produces an output built on the combined data analysis of each layer.

NorthPole contains 256 cores with onboard memory, removing the need for external memory access and improving processing speed. IBM estimated NorthPole could be 25x more efficient than current designs.

Damien Querlioz, a Nanoelectronics Researcher at the University of Paris-Saclay in Palaiseau, stated: “Its energy efficiency is mind-blowing. This computational design will shake the common thinking of computer architecture.”

Current computer chips require external RAM, also known as computer memory, to calculate layers of data inputs. Transferring data between these units reduces computing speed. This phenomenon is known as the Von Neumann Bottleneck: a limitation in computer architecture that occurs from the separation of processing and memory units.

NorthPole is mitigating the Von Neumann Bottleneck through cores wired together in a network layout inspired by the white-matter connections between parts of the human cerebral cortex. In simple terms, NorthPole places memory units close to the computing elements in the core.

Even NorthPole’s 224 megabytes of RAM cannot support Large Language Models (LLMs) like OpenAI’s ChatGPT, which requires several thousand megabytes of data. However, NorthPole could be helpful in speed-critical applications with pre-programmed neural networks, like self-driving cars.

The United States and China are competing fiercely to become world leaders in artificial intelligence (AI). This competition is referred to as the “AI War.” Both countries view AI as a critical technology for the future, potentially revolutionizing industries, militaries, and societies.

The United States has spawned notable companies like Google’s Bard, OpenAI’s ChatGPT, and Anthropic’s Claude 2. In response, China is funding Zhipu, a firm developing Large Language Models (LLMs) to rival OpenAI’s GPT-4.

The company reportedly offers a chatbot called Qingyan. This chatbot recently added an open-source bilingual conversational AI model labeled ChatGLM-6B, which is trained on six billion parameters and can perform inferences on a single consumer-grade graphics card.

Last week, the Biden administration imposed additional restrictions on the export of Nvidia chips to China, impeding the country’s ability to train LLMs. China’s deep-pocketed companies have been stockpiling $4B worth of A800 GPUs to fuel AI projects.

AI models with large numbers of parameters generally perform more advanced tasks. However, neural networks with limited parameters are more hardware-efficient.

Luzia secures $10M in funding to expand its WhatsApp-based chatbot.

Imbue lands $12M in follow-on funding to develop AI systems trained on massive datasets.

OpenAI (San Francisco, CA): Business Recruiting Manager

Hive (Seattle, WA): Systems Engineer, DevOps

Netflix (Remote): Software Engineer (L4), Consumer ML Model Compute and Serving Systems

⭐️⭐️⭐️⭐️⭐️Nailed it!

⭐️⭐️⭐️Average

⭐️Epic fail.

⚙️150 ChatGPT prompts for Copywriting

⚙️325 ChatGPT prompts for Email Marketing

📆Simple Project Management Board

⏱Time Tracker