Source: Tiger Says Chip Original Author: Tiger Says Chip This article introduces chips specifically designed for AI computation. In the context of the rapid development of artificial intelligence (AI), traditional general-purpose processors (such as CPUs) can no longer meet the growing computational demands. Therefore, AI chips have emerged as the core hardware driving the popularization … Read more
1. The World’s First Thermodynamic Computing Chip is Here Normal Computing has announced the successful tape-out of the world’s first thermodynamic computing chip, CN101. This news has sent shockwaves through the tech industry. The CN101 is an ASIC chip designed specifically for AI/HPC data centers, fundamentally different from traditional silicon computing methods. It leverages thermodynamics … Read more
A DSP chip, or Digital Signal Processor, is a microprocessor specifically designed for fast processing of digital signals. It efficiently performs tasks such as signal acquisition, conversion, filtering, analysis, and synthesis through a specialized hardware architecture and instruction set, making it widely used in scenarios requiring real-time signal processing. 1. Core Features of DSP Chips … Read more
Abstract Can AI chips be made just with money? Necessary but not sufficient. Only when you are successful enough can you have a name and reputation. Why are there only a handful of global CPU and GPU companies, yet the wave of artificial intelligence has led to the emergence of so many AI chip companies … Read more
Source: ZHUAN ZHI This article is approximately 1000 words long and is recommended for a 5-minute read. This article presents SparseLoRA, a method to accelerate the fine-tuning of large language models through contextual sparsity. Fine-tuning large language models (LLMs) is often both computationally intensive and memory-consuming. While parameter-efficient fine-tuning methods such as QLoRA and DoRA … Read more
In recent years, recommendation systems have played an increasingly important role across various industries. Among them, the embedding layer of recommendation systems has become a performance bottleneck due to the large volume of data and irregular memory access patterns. Existing works utilize the data locality of the embedding layer to cache frequently accessed embedding vectors … Read more