Home  /  Products  /  CEVA's DSP Cores  /  CEVA-XC Family  /  CEVA-X2

CEVA-X2Print this page

PHY Layer Control DSP for high end mobile baseband


Target applications for CEVA-X2

CEVA-X2 is primarily aimed at Physical Layer Control of high-end multi-carrier and multi-standard cellular baseband.
The CEVA-X2 has been specifically designed to tackle the huge PHY control complexity increase of LTE-Advanced Pro which includes Rel.13 multi-carrier and multi-standard handsets. PHY Control task challenges include:

  • Support capacity increase up to 1.2Gbps in Rel.13 through the aggregation of 5 carriers and the use of MIMO 4x4 and 256 QAM. Control performance scales with the number of carriers and the PHY controller also needs to perform measurements, calibration and other processing functions per carrier,
  • Support Dual-Connectivity to improve offload to Small-Cells and handle concurrent baseband connectivity to both macro and small cells,
  • Merge efficiently Multi-RAT modems (LTE-A, 3G, TD-SCDMA, 2G) into a single unified modem to support dual SIMs, concurrent data over LTE-A and voice over 2G or 3G until VoLTE is widely deployed. Such a feature very low latency task switch to schedule modems concurrently,
  • Control the whole modem datapath at the system level with its multiple DSPs and UL/DL coprocessors and HW accelerators, without Controller overhead. This requires sophisticated mechanisms such as Queue and Buffer Managers, QoS, Flow Control, to automatically schedule PHY components under ultra-low latency with minimal overhead,
  • VoLTE EVS requires 5 to 10 times the complexity (performance and memory footprint) of previous generation voice codecs,
  • Battery life needs to remain competitive despite the huge complexity increase.

CEVA-X2 Block Diagram


The CEVA-X2 offers a wealth of architectural features as follows

CEVA-X2 Key features

  • VLIW/SIMD architecture
  • 10 stage pipeline
  • 128 bit memory bandwidth
  • 4.0 Coremark/MHz
  • Four-MAC 16x16 per cycle
  • 64-SIMD Fixed-Point Operations
  • Up to two IEEE Single-Precision Floating-Point Operations
  • 1.5GHz operating frequency in 16nm FF
  • Instruction and Data Caches
  • Branch Target Buffer
  • CEVA-Connect Technology to schedule control and data planes of Hardware Accelerators

Controller features

  • Four scalar operations per cycle
  • Zero latency ISA
  • Static branch prediction
  • Optional dynamic branch prediction
  • 32-bit HW division and multiplication
  • Ultra-fast context switch
  • Supervisor and User modes
  • Semaphores

DSP features

  • Two SPU (Scalar Processing Unit)
    • Four 16x16 MAC per cycle
    • Two 32x32 MAC per cycle
  • 64-bit SIMD processing
  • 8/16/32/64 bit data type support
  • 8/16/32/64 bit ALU operations
    • Four 16-bit operations per cycle
    • Two 32-bit operations per cycle
    • Two 64-bit operations per cycle
  • Optional single-precision IEEE floating point in each SPU
  • Loop buffer

System features

  • Data Cache
    • 2-Way or 4-Way set-associative cache
    • Write-through and write-back policies
    • Non-blocking read and write on cache miss
    • Hardware and software pre-fetch capabilities
  • Instruction Cache
    • 2-Way or 4-Way set-associative cache
    • Non-blocking read on cache miss
    • Hardware and software pre-fetch capabilities
  • CEVA-Connect
    • Offload the processor with dedicated HW control and data planes
    • Hardware queue and buffer managers for controlling the data flow in the PHY
    • Intelligent scheduling with high QoS and very low latency
    • Data traffic management
    • Special interfaces to connect multiple accelerators / DSP
  • High throughput system interfaces

Exceptional power efficiency

  • Incorporates Power Scaling Unit (PSU 2.0)
  • Enhanced power-optimized pipeline
  • Optimized for battery-operated and stationary devices

Scalable and configurable baseband control DSP architecture for use in advanced multi-RAT LTE modems through different processors, configurations and optional modules

  • Scalable computation capabilities and memories
  • Configurable utilization of optional instruction sets
  • Includes tightly coupled memories (TCM), caches, AXI system interfaces, APB interface, advanced DMA controller, CEVA-Connect, message queues, emulation and profiling modules.
  • Ensures easy integration and optimal performance in Target SoCs

Smooth C-level software development and easy integration into the target SoC reduces risk and time-to-market

  • Advanced IDE
  • Optimizing C compiler
  • Cycle-accurate simulation and graphical profiling of the entire DSP sub-system
  • Macro assembler, linker, and GUI debugger
  • RTOS
  • Smooth migration path from off the-shelf ASSPs
  • MATLAB bi-directional connectivity (optional)
  • Open architecture and standardized APIs to allow Additional software components can be easily developed or licensed through CEVA’s partners