AI and HPC workloads are outpacing what traditional interconnects can deliver. Memory is stranded, bandwidth is capped, and inference cost keeps climbing. Compute Express Link® solves this by making memory coherent, poolable, and shareable securely across CPUs, XPUs, and accelerators — at rack scale.
Synopsys CXL IP gets you there faster with a complete, silicon-proven solution spanning CXL 4.0, 3.x, 2.0, and 1.x — backed by the industry's most proven PCIe foundation, with 25+ years of PCIe leadership and thousands of PCIe design wins behind every link.
CXL is an open, cache-coherent interconnect built on PCIe. It lets CPUs, XPUs, and accelerators share memory with near-cache latency — the foundation for solving AI's memory wall.
CXL 4.0 doubles bandwidth to 128 GT/s and adds new Bundled Port capabilities can enable over 2TB/s bandwidths bundling 4 x16 links or over 4TB/s bandwidths bundling 8 x16 links.
CXL.mem enables rack-scale coherent memory pooling, accelerating KV cache offload 3–6× vs. SSD alternatives with sub-200 ns load-to-use latency.
Yes. CXL 4.0 includes IDE security, fabric protections, and enhanced RAS. Synopsys IDE Security Modules deliver AES-GCM based encryption and authentication with zero-cycle overhead for CXL .cache/.mem and TSP/TDISP for confidential computing/virtualization support.