Cadence enthüllt Tensilica ConnX B20 DSP: 30× Faster für 5G & Automotive Radar

Cadence Design Systems hat Cadence Tensilica ConnX B20 DSP IP Gestartet – Fastest Member Well-Known ConnX Family. Built On Deeper Pipeline Architecture, ConnX B20 Delivers Superior Speed und Power Efficiency für Automotive und 5G Communications – Including Next-Gen Radar, Lidar, Vehicle-to-Everything (V2X), User-Equipment Infrastructure, IoT Use Cases. Compared With Widely Used ConnX BBE32EP DSP, B20 Can Accelerate Segments Communications Stack Up To 30× und Parts Radar/Lidar Pipeline Up To 10×.

Performance Drivers

Higher-Resolution Imaging Radar mit More Antennas für Autonomous Driving, Rise 5G Mobile Equipment, Consumer-Grade Radar All Demand Dramatically More Processing Power While Staying Energy-Efficient. New ConnX B20 und Companion B10 DSPs Meet Those Requirements – Offering Higher Performance und Lower Energy Consumption Across Broad Range Applications.

Technical Specifications

• Fully Compatible Existing ConnX DSP Family (BBE16/32/64EP)
• 512-Bit Vector Width, 128 MACs, Loads 1.024 Bits Data Each Cycle
• Operated at 1,4 GHz or Higher in 16 nm Process – Extends Family Performance Spectrum
• Higher Precision: 32-Bit Fixed-Point Ops (Optimized MAC, FFT, FIR) oder Single-/Half-Precision Vector Floating-Point (VFPU) Ops – Half-Precision Delivers 2× Throughput Single-Precision
• Extended VFPU Adds Complex Floating-Point Support – Doubles Real-Floating-Point Ops at Same Vector Width – Ideal Front-End Radar Processing Chain
• Dedicated Communications Option Accelerates Forward-Error Correction (FEC) für Software-Defined Radio at Lower Bit-Rates

With Native 32-Bit Support und Higher Clock Speed, ConnX B20 Delivers Up To 10× Faster Performance Radar/Lidar Processing vs. Popular ConnX BBE32EP. Clock-Speed Boost und FEC Acceleration Make It Up To 30× Faster für Parts Communication Chain. Both B20 und B10 Provide Customers Scalable Performance und Efficiency While Preserving Familiar ConnX Architecture und Development Environment. Fast Software Development Enabled by Advanced Compilers, Debugging/Simulation Tools, Optimized Math Library mit Ready-Made Application Examples.