Chip Industry Week in Review

By The SE Staff - 25 Apr, 2025 - Comments: 0

To listen to the podcast version, click here. TSMC unveiled an unusually detailed roadmap at this week's North America Technology Symposium, including future architectures for 3D-ICs for high-performance computing and small, extremely low-power chips for AR/VR glasses, and two implementations of system-on-wafer. Fig. 1: TSMC's future packaging and stacking roadmap. Source: TSMC The ... » read more

AI Drives Re-Engineering Of Nearly Everything In Chips

By Ed Sperling - 24 Apr, 2025 - Comments: 0

AI's ability to mine patterns across massive quantities of data is causing fundamental changes in how chips are used, how they are designed, and how they are packaged and built. These shifts are especially apparent in high-performance AI architectures being used inside of large data centers, where chiplets are being deployed to process, move, and store massive amounts of data. But they also ... » read more

Implementing AI Activation Functions

By Bryon Moyer - 10 Apr, 2025 - Comments: 1

Activation functions play a critical role in AI inference, helping to ferret out nonlinear behaviors in AI models. This makes them an integral part of any neural network, but nonlinear functions can be fussy to build in silicon. Is it better to have a CPU calculate them? Should hardware function units be laid down to execute them? Or would a lookup table (LUT) suffice? Most architectures inc... » read more

No Fooling With Voxel Pooling

By Steve Roddy - 13 Mar, 2025 - Comments: 0

A variety of new and complicated transformer models have emerged in the past 18 to 24 months as new “must have” networks in advanced automotive use cases. These novel architectures often introduce new network operators or novel ways of combining tensors – often from different types of sensors – in ways to enhance detection and recognition of objects in L3 / L4 / L5 ADAS and autonomous d... » read more

Vision Language Models Come Rushing In

By Steve Roddy - 17 Feb, 2025 - Comments: 0

Just when you thought the pace of change of AI models couldn’t get any faster, it accelerates yet again. In the popular news media, the introduction of DeepSeek in January 2025 created a moment that captured headlines in every newspaper and website heralding comparisons to the Sputnik moment of 1957. But rapid change is also happening in many quarters that are hidden from view of the Chat-App... » read more

Normalization Keeps AI Numbers In Check

By Bryon Moyer - 13 Feb, 2025 - Comments: 0

AI training and inference are all about running data through models — typically to make some kind of decision. But the paths that the calculations take aren’t always straightforward, and as a model processes its inputs, those calculations may go astray. Normalization is a process that can keep data in bounds, improving both training and inference. Foregoing normalization can result in at... » read more

Chip Industry Week In Review

By The SE Staff - 31 Jan, 2025 - Comments: 0

Chinese startup DeepSeek rattled the tech world and U.S. stock market with claims it spent just $5.6 million on compute power for its AI model compared to its billion-dollar rivals in the U.S. The announcement raised questions about U.S. investment strategies in AI infrastructure and led to an initial $600 billion selloff of NVIDIA stock. Since its launch, DeepSeek reportedly was hit by malicio... » read more

MACs Are Not Enough: Why “Offload” Fails

By Steve Roddy - 16 Jan, 2025 - Comments: 0

For the past half-decade, countless chip designers have approached the challenges of on-device machine learning inference with the simple idea of building a “MAC accelerator” – an array of high-performance multiply-accumulate circuits – paired with a legacy programmable core to tackle the ML inference compute problem. There are literally dozens of lookalike architectures in the market t... » read more

2025: So Many Possibilities

By Brian Bailey - 16 Jan, 2025 - Comments: 0

The stage is set for a year of innovation in the chip industry, unlike anything seen for decades, but what makes this period of advancement truly unique is the need to focus on physics and real design skills. Planar scaling of SoCs enabled design and verification tools and methodologies to mature on a relatively linear path, but the last few years have created an environment for more radical... » read more

What’s The Best Way To Sell An Inference Engine?

By Bryon Moyer - 16 Jan, 2025 - Comments: 0

The burgeoning AI market has seen innumerable startups funded on the strength of their ideas about building faster, lower-power, and/or lower-cost AI inference engines. Part of the go-to-market dynamic has involved deciding whether to offer a chip or IP — with some newcomers pivoting between chip and IP implementations of their ideas. The fact that some companies choose to sell chips while... » read more

← Older posts

tag: Quadric