Blog Review: Feb. 25

Chiplet architecture spec; LPDDR6; unmanufacturable PCB vias; AI graphics rendering.

popularity

Cadence’s Mick Posner introduces the Foundational Chiplet System Architecture, a specification that aims to deliver a vendor and CPU-neutral architecture, common system partition guidelines, and a shared vocabulary and set of standards for system-level and interface definitions between chiplets.

Synopsys’ Scott Knowlton explains why LPDDR6 represents a big step forward in memory management compared to its predecessors, with voltage scaling and command encoding schemes reworked to enable more granular power states and improved signal integrity.

Siemens’ Bill Ji shares some general guidelines to avoid creating unmanufacturable PCB via structures, including how to watch out for uncommon via structures that can prove problematic.

Imagination’s Ed Plowman finds that modern graphics rendering is characterized by compute-driven processes, neural inference, and an increasing reliance on AI-assisted image formation, with direct consequences for GPU design.

Arm’s Will Abbey highlights how low-cost mobile platforms and biometric ID technology help track medical records and improve the success of malaria vaccination campaigns in rural Ghanaian villages.

Keysight’s Maryvonne Chalony introduces a multiphysics simulation workflow for designing integrated lidar-on-chip systems that combines finite difference time domain (FDTD) for complex radiative structures with beam propagation method (BPM) for guided wave components.

The ESD Alliance’s Bob Smith chats with Aki Fujimura of D2S about the role of GPU acceleration in mask design, a predicted move toward curvilinear mask features, and his personal engineering journey.

Plus, check out the blogs featured in the latest Manufacturing, Packaging & Materials newsletter:

Amkor’s Dambi Jo shows why efficient heat dissipation is critical in fan-out packages.

Lam Research’s Sam Sarkar explains within-wafer variations and how to identify sensitive parameters, optimize recipes, and improve die yield without the time and cost of extensive wafer-based testing.

Synopsys’ Anders Blom and Igor Markov examine the benefits of machine-learned force fields, transforming the modeling of atomic interactions.

SEMI bloggers Jan-Bart Smits and David Harap dig into regional differences and predictions for photonics, warning that teams must make decisions now about technologies that won’t mature for years.



Leave a Reply


(Note: This name will be displayed publicly)