The M88MX6852 is an Memory eXpander Controller (MXC) chip compliant with the CXL® 3.1 Type 3 specification. The controller supports both CXL®.mem and CXL®.io protocols, delivering high-bandwidth, low-latency memory expansion and pooling solutions for data-intensive workloads such as machine learning and cloud computing infrastructure.
Built on a PCIe® 6.2 physical layer interface, the M88MX6852 acheives data transfer rates of up to 64 GT/s (x8 lanes) and integrates a CXL® subsystem, a DDR subsystem with DDR5 PHY, and a RISC-V subsystem. Compliant with the CXL® 3.1 specification, it also supports the JEDEC DDR5 standard and provides rich peripheral interfaces such as SMBus, I3C/I²C, and SPI for connectivity with CPUs, memory systems, and external Flash memory.
As a critical enabler of memory sharing and memory pooling, the M88MX6852 can be widely deployed in Add-in Cards (AICs), EDSFF memory modules, and other devices. It not only significantly enhances memory expansion performance and energy efficiency but also promotes the development of disaggregated memory architectures through standardized protocols, laying a solid foundation for next-generation high-efficiency computing infrastructure.
Features
- CXL® Type 3 Memory eXpander Controller
- Integrated CXL® controller, DDR controller and RISC-V micro-processor
- On-chip PVT sensor
- SMBus, I3C/I²C, SPI, JTAG and UART interfaces
- Package: 1211-ball FCBGA
- CXL® Controller
- Compliant with CXL® 1.1, CXL® 2.0 and CXL® 3.1 specifications, supporting CXL®.mem and CXL®.io protocols
- Designed based on PCIe® 6.2 interface, supporting up to 64 GT/s data rate (x8 lanes)
- Rich RAS features
- DDR Controller
- Compliant with JEDEC DDR5 standard
- Speed up to DDR5-8000
- Support DDR5 UDIMM/RDIMM and DDR5 DRAM on-board
- Low power consumption
- Optimized DDR I/O equalization and programmable I/O impedance
Application
- Add-in cards or EDSFF memory modules for memory expansion, memory pooling, etc.
- Data-intensive and compute-intensive applications, such as in-memory database, cloud infrastructure, and emerging data center workloads