Overview

Infrastructure applications including networking, storage, and server have a wide range of compute performance and power requirements.  The Arm CoreLink CCN family is well suited to scale across the performance spectrum by supporting from 1 to 48 cores along with an integrated, configurable Level 3 system cache. The L3 system cache allows IO and accelerators to allocate memory on chip, offering reduced latency and power by minimizing accesses to external memory. The CCN family is optimized for the latest AMBA 5 CHI processors and memory controllers, and AMBA 4 ACE-Lite/AXI4 for IO interfaces.

  • CoreLink CCN-512 Cache Coherent Network
  • CoreLink CCN-512

    • Full coherency for up to twelve processor clusters (48 cores)
    • Supporting 1-32MB of L3 system cache for highest compute density

  • CoreLink CCN-508 Cache Coherent Network
  • CoreLink CCN-508

    • Full coherency for up to eight processor clusters (32 cores)
    • Supporting 1-32MB of L3 system cache for high performance compute and IO

  • CoreLink CCN-504 Cache Coherent Network
  • CoreLink CCN-504

    • Full coherency for up to four processor clusters (16 cores)
    • Supporting 1-16MB of L3 system cache for the highest performance of 16 core systems

  • CoreLink CCN-502 Cache Coherent Network
  • CoreLink CCN-502

    • Full coherency for up to four processor clusters (16 cores)
    • Supporting up to 0-8MB of optional L3 system cache for small foot print cost sensitive applications

Product Comparison

Arm offers a range of CCN interconnects optimized for different size applications. Each product includes a wide configuration space allowing the SoC architect to optimize the interconnect to meet performance goals with the smallest possible area and power.

  CoreLink CCN-512 CoreLink CCN-508 CoreLink CCN-504
CoreLink CCN-502
Summary  Up to 12 cluster interconnect
(max processor bandwidth)
Up to 8 cluster interconnect
(high bandwidth) 
Up to 4 cluster interconnect
(high IO performance) 
Up to 4 cluster interconnect
(optimized for size) 
Performance Up to 225 GB/s
Up to 200 GB/s
Up to 150 GB/s
Up to 100 GB/s
Processors Up to 48 cores  Up to 32 cores  Up to 16 cores  Up to 16 cores 
DDR 1 to 4 channels  1 to 4 channels   1 to 2 channels   1 to 4 channels  
IO  Up to 24 AXI4/ACE-Lite Up to 24 AXI4/ACE-Lite  Up to 18 AXI4/ACE-Lite  Up to 9 AXI4/ACE-Lite 
Level 3 Cache  1-32 MB 1-32 MB  1-16 MB  0-8 MB 
System Size Large  Medium-Large  Medium  Small 
Example Applications Cloud applications, storage array network controller, Macro Base Station, and core networks.
Cloud applications, storage array network controller, Macro Base Station, and core networks.

Small enterprise solutions, cellular cell network, media content.
Home networks, home gateway, small cell base station, DSL modem, and web tier server.

CoreLink CCN Family Features

Optimized for Arm Cortex Processors

The CoreLink CCN family supports the latest Arm 64-bit processors include high efficiency Cortex-A53 and high performance Cortex-A57 and Cortex-A72. The CoreLink CCN family use the latest AMBA 5 CHI (Coherent Hub Interface) protocol which has been developed to support high frequency, non-blocking data transfers between multiple fully coherent processors. It is especially suited for networking infrastructure and server applications.

Integrated Low Latency Level 3 System Cache

The CoreLink CCN family includes an optional integrated level 3 (L3) system cache and snoop filter functions. The L3 cache, which is configurable in size, extends on-chip caching for demanding workloads and offers low latency on-chip system cache for allocation and sharing of data between processors, high-speed IO interfaces and accelerators. The snoop filter removes the need for broadcast coherency messaging, further reducing latency and power.

High Performance DDR3 and DDR4 Memory Interfaces

The CoreLink CCN family are optimized to work with the CoreLink DMC-520 Dynamic Memory Controller providing a high-bandwidth interface for off-chip memory, such as DDR3, DDR3L and DDR4 DRAM. CoreLink DMC-520 offers enterprise class RAS (Reliability, Availability and Serviceability) features such as ECC for x72 DRAM, TrustZone security and End to End Quality of Service (QoS).


Performance and Frequency

The CoreLink CCN family is designed to work closely with the latest Arm Cortex applications processors and can be implemented at clock speeds that match the processor. Interfaces to memory and processors can be configured for asynchronous operation to allow power management including dynamic voltage and frequency scaling.

CoreLink CCN System Examples

Server SoC with CoreLink CCN-512

Below is an example server configuration based on the CoreLink CCN-512.  It is designed to maximize compute density and well suited for web, cloud and big data analytic applications.  Some of the key features include:

  • Maximize compute density with 48x Cortex-A72, Arm's highest performance processor
  • Integrated 32MB of Level 3 System Cache for compute and IO
  • Up to 24 coherent IO ports for high bandwidth controllers such as 100Gb Ethernet and PCIe
  • 4 channels of x72 DDR4-3200 with CoreLink DMC-520

Networking SoC with CoreLink CCN-502

To meet the demands of increasing networking trafffic throughput and number of connected end points, system architects are looking to balance compute, storage and acceleration for a wide range of system design targets.  The CoreLink CCN-502 has been optimized for area and power making it the ideal interconnect for small to mid-range networking solutions.

  • Cortex-A72 processors provide the compute performance required of the Control Plane
  • Cortex-A53 processors provide efficient packet processing and IO throughput required of the Data Plane
  • Specialized DSPs, GPUs and other accelerators provide off-load engines for range of diverse needs such as security and layer 1 signal processing
  • Configurable CoreLink CCN-502 with flexible number of master interfaces, memory ports and optional Level 3 System Cache (0-8MB)