Chiplet Interfaces Embrace Failures
Why lane swapping is essential to meet assembly yield.
By Anne Meixner, Semi Engineering | August 12th, 2025
Redundancy in chiplet interfaces is now a prerequisite for achieving sufficient yield in high-performance computing devices, which today are packed with tens of thousands of interconnects. And as the number and density of those interconnects increases, the prospects for yield only worsen.
For more than two decades, high-speed I/O interfaces have included reliability strategies to manage in-field system board failures. For example, the PCI Express 2.0 standard introduced in 2007 included 16 lanes for transactions. But if there was a lane failure then only 8 of those lanes were utilized, cutting the transaction rate in half. Commonly referred to as graceful degradation, this remains a strategy for computing chiplet interfaces such as UCIe.
To read the full article, click here
Related Chiplet
- 12nm EURYTION RFK1 - UCIe SP based Ka-Ku Band Chiplet Transceiver
- Interconnect Chiplet
- Bridglets
- Automotive AI Accelerator
- Direct Chiplet Interface
Related News
- Paving the way for the semiconductor future: The Chiplet Center of Excellence commences operations
- Metrology And Inspection For The Chiplet Era
- Weaving State-of-the-Art Chiplet and SoC Fabrics
- Femtosense Combines AI Chiplet with MCU for Audio SiP
Latest News
- Rigetti Computing Reports Second Quarter 2025 Financial Results; Announces General Availability of its 36-Qubit Multi-Chip Quantum Computer
- AMI Launches Developer Program for Arm Total Design to Rapidly Accelerate Arm-based Chiplet Development
- Wave Photonics launches its PDK Management Platform with CORNERSTONE as the first user
- Chiplet Interfaces Embrace Failures
- Open Compute Project Enables Silicon Diversity in AI Clusters