ISA Extension For Low-Precision NN Training On RISC-V Cores


New technical paper titled "MiniFloat-NN and ExSdotp: An ISA Extension and a Modular Open Hardware Unit for Low-Precision Training on RISC-V cores" from researchers at IIS, ETH Zurich; DEI, University of Bologna; and Axelera AI. Abstract "Low-precision formats have recently driven major breakthroughs in neural network (NN) training and inference by reducing the memory footprint of the N... » read more

Can Analog Make A Comeback?


We live in an analog world dominated by digital processing, but that could change. Domain specificity, and the desire for greater levels of optimization, may provide analog compute with some significant advantages — and the possibility of a comeback. For the last four decades, the advantages of digital scaling and flexibility have pushed the dividing line between analog and digital closer ... » read more

Enhancing Datasets For Artificial Intelligence Through Model-Based Methods


By Dirk Mayer and Ulf Wetzker Industrial plants and processes are now digitized and networked, and AI can be used to evaluate the data generated by those facilities to increase productivity and quality. Machine learning (ML) methods can be applied to: Product quality classification in complex production processes. Condition monitoring of technical systems, which is used, for examp... » read more

HBM2E Raises The Bar For Memory Bandwidth


AI/ML training capabilities are growing at a rate of 10X per year driving rapid improvements in every aspect of computing hardware and software. HBM2E memory is the ideal solution for the high bandwidth requirements of AI/ML training, but entails additional design considerations given its 2.5D architecture. Designers can realize the full benefits of HBM2E memory with the silicon-proven memory s... » read more

Using ML In EDA


Machine learning is becoming essential for designing chips due to the growing volume of data stemming from increasing density and complexity. Nick Ni, director of product marketing for AI at Xilinx, examines why machine learning is gaining traction at advanced nodes, where it’s being used today and how it will be used in the future, how quality of results compare with and without ML, and what... » read more

Tradeoffs Between Edge Vs. Cloud


Increasing amounts of processing are being done on the edge, but how the balance will change between what's computed in the cloud versus the edge remains unclear. The answer may depend as much on the value of data and other commercial reasons as on technical limitations. The pendulum has been swinging between doing all processing in the cloud to doing increasing amounts of processing at the ... » read more

Education Vs. Training


While writing my recent articles on the subject of training, a number of people pointed out that training and education are not the same thing. In a very simple sense, training is defined to be learning a skill or behavior that enables you to 'do' something, whereas education is the acquisition of knowledge from study or training. These definitions leave me cold and, in my mind, miss a very ... » read more

What Is Intern Reading Club?


As the summer winds down, interns are busy completing their assigned projects and preparing their end of summer presentations. These presentations have been a rite of passage for interns on the Pointwise team for many years and gives each intern a chance to show off what they learned and accomplished. And the rest of the team gets to hear all the details of what they've been working on. Anothe... » read more

Continuous Education For Engineers


Continuous education is essential for engineers, but many companies don't recognize the value or they are unwilling to provide the necessary resources. This should be a line of questioning before every new hire makes the decision about where they want to work, because it not only affects their future career, but also impacts the value they can provide to that company during the course of the... » read more

RaPiD: AI Accelerator for Ultra-low Precision Training and Inference


Abstract—"The growing prevalence and computational demands of Artificial Intelligence (AI) workloads has led to widespread use of hardware accelerators in their execution. Scaling the performance of AI accelerators across generations is pivotal to their success in commercial deployments. The intrinsic error-resilient nature of AI workloads present a unique opportunity for performance/energy i... » read more

← Older posts Newer posts →