Protecting die-2-die interfaces...
There are several reasons to combine various chiplets in a single package, either next to each other (2.5D) or stacked on top of each other (3D) like the memory cube depicted below.
Example of a 3D stacked, co-packaged assembly. 4 layers of DRAM memory are stacked on top of a logic layer. Many of the interfaces between the logic layer and the memory devices stay within the package.
For instance, it enables chip designers to select the most optimal process for each type of function (interface, memory, compute, FPGA, analog circuits, MEMS, optical, …). It also allows to create chips with larger gate count when the reticle size is the limiting factor. And of course, yield improves a lot when the area of the die is reduced.
At Sofics we get a lot of questions about the required ESD robustness for the die-2-die (D2D) interfaces between chiplets in a package. People wonder how to select the right ESD standard (HBM, MM, CDM, TLP – more info) and what robustness level (1kV HBM?) they need to design for. Furthermore people wonder how they can validate the robustness once ESD devices are integrated?
What is the right ESD standard and ESD robustness level?
To read the full article, click here
Related Semiconductor IP
- NPU IP Core for Mobile
- NPU IP Core for Edge
- Specialized Video Processing NPU IP
- HYPERBUS™ Memory Controller
- AV1 Video Encoder IP
Related Blogs
- SSD Interfaces and Performance Effects
- Protecting Against IP Theft
- On-chip interfaces gain importance in next-gen FPGAs
- Verifying MIPI interfaces in SoCs
Latest Blogs
- Cadence Extends Support for Automotive Solutions on Arm Zena Compute Subsystems
- The Role of GPU in AI: Tech Impact & Imagination Technologies
- Time-of-Flight Decoding with Tensilica Vision DSPs - AI's Role in ToF Decoding
- Synopsys Expands Collaboration with Arm to Accelerate the Automotive Industry’s Transformation to Software-Defined Vehicles
- Deep Robotics and Arm Power the Future of Autonomous Mobility