SOLUTION
EMFASYS:
Elastic AI Memory Fabric System
- Industry's first RDMA Ethernet based AI Memory Fabric System
- Offloads GPU, HBM, and local head node DDR resource consumption
- Drives down the cost of LLM inference at fleet scale
- Enables seamless integration with LLM inference application frameworks
Under the Hood
Under the Hood
- Powered by Enfabrica's 3.2 Tbps ACF SuperNIC silicon
- Elastically connects up to 144 CXL memory lanes per system to resilient bundles of 400G / 800G RDMA Ethernet ports
- Pooled memory target up to 18 TB CXL DDR5 DRAM per system (expandable to 28 TB in the future)
- High memory bandwidth aggregation enabled by striping transactions across 18+ memory channels per system
- Read access times in microseconds
- Designed to interoperate with multiple GPU servers and initiator RDMA NICs
Delivers up to 50% lower cost per token per user
Delivers up to 50% lower cost per token per user
- Ideal for agentic, batched, expert-parallel, high-turn, and/or large-context inference workloads
- Equally deployable for AI training: activation storage offload, distributed checkpointing, optimizer state sharding
- Software-enabled, high-performance caching hierarchy hides transfer latency within compute pipelines
- Outperforms flash-based inference storage offload solutions with 100x lower latency and unlimited write/erase transactions
- 100% fabric-attached, headless memory free from head node CPU thread contention or locality constraints
Unlock Innovation
Unlock Innovation
Download to learn more!
ACF SuperNIC Silicon Product Brief
ACF SuperNIC System Product Brief
ACF SuperNIC Host Software Brief