Overcoming Timing Closure Issues in Wide Interface DDR, HBM and ONFI Subsystems
By Brian Gardner, True Circuits
ABSTRACT:
In wide chip interfaces like DDR, HBM and ONFI, it can be challenging to synthesize and connect high-frequency controllers to the PHY hard macros. Clock trees can be expansive, pushing tools to their limits, and often multiple clock domains are needed. Jitter can also be an issue on long paths. In this presentation Brian will show how True Circuits PLL and DLL IP is being used by multiple customers to build ONFI and HBM subsystems in advanced TSMC process nodes, and discuss the tradeoffs and timing budget concerns among different timing architectures. In addition, he will explain how source-synchronous signaling is used in our DDR PHY to ease timing closure, and to allow the memory controller to be synthesized for high-frequency operation, which reduces its size and lowers its latency. By using a soft IP "shim" between the memory controller and PHY, the memory controller only needs a single localized clock tree, reducing mismatch and jitter. The long routes between the soft shim and the PHY hard macros are source-synchronous, so data/strobe groups need only be roughly matched, something easily accomplished by place and route tools.
AUTHOR:
Brian Gardner is TCI's V.P. of Business Development. He is a long time semiconductor industry veteran, and has spent the last 11 years in the semiconductor IP market. He has held senior positions in general management, marketing and business development at Denali, Cadence, QLogic and Motorola. Brian is well known for his knowledge and experience in IP subsystems, particularly DDR and ONFI.
Related Semiconductor IP
Related White Papers
- Making Better Front-End Architectural Choices Avoids Back-End Timing Closure Issues
- Interface Timing Challenges and Solutions at Block Level
- Optimizing Floorplan for STA and Timing improvement in VLSI Design Flow
- I2C Interface Timing Specifications and Constraints
Latest White Papers
- Reimagining AI Infrastructure: The Power of Converged Back-end Networks
- 40G UCIe IP Advantages for AI Applications
- Recent progress in spin-orbit torque magnetic random-access memory
- What is JESD204C? A quick glance at the standard
- Open-Source Design of Heterogeneous SoCs for AI Acceleration: the PULP Platform Experience