Contact Us

UCIe Bridge IP connects an existing SoC protocol or interconnect to a Universal Chiplet Interconnect Express (UCIe) die-to-die connection. It allows traffic from interfaces such as AXI, CHI, CXL, PCIe, or CXS to move through a UCIe-based chiplet architecture without requiring the surrounding subsystem to be redesigned around a different protocol.

This is important because UCIe provides the standardized die-to-die connection, but the logic inside each chiplet may already communicate using established processor, accelerator, memory, coherent, streaming, or I/O protocols.

What Does UCIe Bridge IP Do?

UCIe Bridge IP sits between the protocol environment used by the subsystem and the UCIe infrastructure used to communicate across chiplets.

A simplified data path can be represented as:

SoC Protocol Fabric → UCIe Bridge IP → UCIe Controller IP → Die-to-Die Link

For example, an accelerator built around AXI may use an AXI-to-UCIe bridge, while a coherent processor subsystem may require CHI-to-UCIe connectivity. A CXL-based memory or accelerator architecture can use a CXL-to-UCIe bridge, while PCIe and CXS environments can use their corresponding bridges.

For a more detailed explanation of this architecture, see UCIe Bridge IP Explained: Connecting AXI, CHI, CXL, PCIe, and CXS Across Chiplets.

Why Is Bridge IP Needed in a UCIe Chiplet Design?

UCIe standardizes communication across the die-to-die boundary, but it does not mean every subsystem inside a chiplet uses the same protocol. UCIe already defines protocol mappings for PCIe and CXL, so a separate protocol bridge is not required in every implementation. The need for Bridge IP depends on the subsystem interface and how the UCIe protocol functionality is integrated.

Modern multi-die systems can include processors, AI accelerators, memory subsystems, high-speed I/O, coherent fabrics, and streaming data paths. Each may already use a protocol suited to its particular function.

Bridge IP allows those protocol domains to connect to the UCIe infrastructure while preserving the behavior required by the surrounding architecture.

This is especially useful in heterogeneous chiplet systems where different die may require different types of protocol connectivity.

Which Protocols Can SmartDV Bridge to UCIe?

SmartDV currently provides dedicated UCIe Bridge IP for five protocol environments:

The appropriate bridge depends on the protocol environment surrounding the chiplet connection. A complex multi-die system may also use more than one bridge type across different chiplets or subsystems.

For more detail on how these protocol families connect into a UCIe architecture, see Can AXI, CHI, CXL, PCIe, and CXS Be Used with UCIe?

What Is the Difference Between UCIe Bridge IP and UCIe Controller IP?

UCIe Bridge IP and UCIe Controller IP perform complementary but different functions.

UCIe Controller IP implements and manages the UCIe protocol stack used for the die-to-die connection. UCIe Bridge IP connects that UCIe infrastructure to the AXI, CHI, CXL, PCIe, CXS, or other protocol environment used by the surrounding subsystem.

For the concise FAQ explanation, see What Is the Difference Between UCIe Controller IP and UCIe Bridge IP?

SmartDV provides both UCIe 2.x Controller IP and UCIe 3.x Controller IP alongside its protocol-specific bridge portfolio.

If you are evaluating which controller generation fits the design, see What Is the Difference Between UCIe 2.x and UCIe 3.x Controller IP?

For a deeper explanation of how these components work together, read UCIe Controller IP vs. UCIe Bridge IP: What Does a Chiplet Design Need?

Can a Chiplet Architecture Use Multiple UCIe Bridges?

Yes. A heterogeneous chiplet architecture can contain several protocol domains, so different chiplets or subsystems may require different bridges.

For example, an AI or HPC system could include an AXI-based accelerator, a CHI-based coherent processor subsystem, a CXL-connected memory component, and PCIe-oriented I/O functionality. These components can use different protocol bridges while sharing UCIe as the common die-to-die interconnect.

This modular approach is explored further in Building a Modular UCIe IP Stack for AI and HPC Chiplet Architectures.

How Is UCIe Bridge IP Verified?

Verification should cover more than whether data can pass through the bridge. Teams need to validate the behavior of the protocol on both sides of the boundary and confirm that transactions remain correct as they move through the complete data path.

Important verification areas can include:

  • Protocol transaction behavior
  • Ordering and flow control
  • Flit formatting and data integrity
  • Link initialization and training
  • Reset and recovery behavior
  • Sideband communication
  • Error detection and retries
  • Stress and corner-case traffic
  • End-to-end transaction integrity

SmartDV’s UCIe Verification IP supports UCIe versions 1.0, 1.1, 2.0, and 3.0 and provides protocol-aware infrastructure for validating UCIe interface behavior.

Additional verification considerations are covered in Why UCIe Verification Is Critical for Chiplet-Based SoC Design.

Explore SmartDV UCIe Bridge IP

SmartDV provides UCIe Controller IP, AXI, CHI, CXL, PCIe, and CXS Bridge IP, and UCIe Verification IP for building and validating configurable multi-die architectures.

Explore SmartDV UCIe IP for Next-Generation Chiplet Designs to review the controller, bridge, and verification technologies available for your chiplet architecture.

» Contact SmartDV to discuss UCIe Bridge IP, Controller IP, Verification IP, or project-specific integration requirements.

See All Chiplets & Die-to-Die Interconnects