Re-Architecting Die-to-Die IO For AI


By Lakshmi Jain and Wei-Yu Ma As AI-driven workloads continue to push the boundaries of compute scale, power efficiency, and bandwidth density, conventional die-to-die interconnect technologies—such as SerDes-based links and wide parallel IO—are increasingly becoming limiting factors. These approaches struggle to meet the growing demands for higher bandwidth density and improved energy e... » read more

AI Models Transform Defect Inspection And Review, But Can Fail To Scale


Key Takeaways: AI plays a role in improving defect capture rate and distinguishing between yield-killing and nuisance defects. New developments in wafer edge inspection are proving essential to bonded wafer yields. 70% of AI initiatives stall after pilot implementation, but some pitfalls can be avoided. One of the brightest spots in AI use today is the industry’s ability t... » read more

The Evolution Of UCIe


Since it was released in March 2022, the Universal Chiplet Interconnect Express (UCIe) has grown from a basic way of connecting two dies together into a comprehensive specification that can ensure the handoff of data between various components in an advanced package, as well as validate the chiplets within that package. Mayank Bhatnagar, director of product marketing at Cadence, talks about the... » read more

Swapping Out Chiplets: I/Os Vs. Compute


Key Takeaways: Companies can save time and money by swapping out a compute, memory, or I/O chiplet to gain technology improvements, while keeping the other dies stable. Chip architects may choose to keep their I/Os stable and swap out compute to move from a 5nm process node to 3nm to achieve performance and power improvements, or swap out memory from LPDDR5X to LPDDR6. Swapping out... » read more

AI Accelerator Testing Depends On DFT Innovations


Key Takeaways: I/O and lane repair capabilities are becoming critical to improving yield. System-level testing catches marginal defects and rare defects such as silent data corruption errors. Synopsys and TSMC developed a multi-die demo vehicle capable of full test, monitor, debug, and repair capability across the system’s lifecycle. The proliferation of accelerators in AI... » read more

How To Streamline Your Advanced Package Interconnect Designs


Monolithic system-on-chip (SoC) designs was once a popular choice. However, they face significant constraints in the era of AI. By forcing all chip functions into a single die and process node, they reduce engineering, manufacturing, and design cost flexibility. In contrast, the multi-die nature of chiplets enables different SoC functions to be designed and verified independently and fabrica... » read more

Beating The Heat In 3D Packages


Key Takeaways: Thermal management is a central design constraint, requiring early, thorough planning. Accurate thermal simulation requires AI-driven adaptive meshing and real-world validation. Innovative STCO strategies can drastically reduce GPU peak temperature. As HPC and AI accelerators push power densities to 1kW and beyond, the heat generated by rapidly switching tran... » read more

Making Hybrid Bonding Better


Key Takeaways Fab processes are optimizing for cleanliness, planarity, and high bond quality. Nanotwinned copper and SiCN PVD enable lower anneal and deposition temperatures for HBM. A thin, protective layer helps preserve the Cu/dielectric during aggressive processes. The future of semiconductor manufacturing is no longer dependent just on shrinking features. Instead, chipm... » read more

One-on-One With proteanTecs CEO Shai Cohen


The acceleration of technology is unprecedented: AI data centers, edge build-out, robotics, photonics, quantum, multi-die assemblies. Semiconductor Engineering Editor in Chief Ed Sperling talks with proteanTecs CEO Shai Cohen about what's changing and what impact it will have. Click here to listen. » read more

Chiplets And 3D-ICs Add New Electrical And Mechanical Challenges


Key Takeaways • Chiplets and 3D-IC architectures add new thermal-mechanical stresses that can affect the reliability of entire systems. • As chiplets are assembled into packages, defectivity targets become more stringent for each component in a system. • Traditional silos are breaking down, forcing design teams to address issues such as materials choices that previously were handled by... » read more

← Older posts