Performance, latency, and energy efficiency are critical parameters for system solutions at the edge.
Edge computing solutions facilitate data processing near the source of data generation and serve as a decentralized extension of the cloud or data center networks. This eases the integration of locally generated data with lower latency and reduces bandwidth by filtering the relevant data at the edge.
What is an FPGA?
FPGA are highly flexible computing devices, that enable fully optimized hardware for specific compute challenges.
Unlike CPUs or GPUs, their architecture is re-programmable enabling efficient use of resources and power efficiency.
FPGAs have large amounts of on-chip memory that can be flexibly combined with compute and logic functionality to build high-performance data-flow-oriented products.
Acceleration with FPGA
CPU
- 10s of processing cores
- Pre-defined instruction set
- Serial execution of instructions
- Optimized for general purpose computing
GPU
- 1000s of processing cores
- Pre-defined instruction set
- Single Instruction Multiple Data-flow
- Highly effective parallel execution
FPGA
- Millions of programmable logic cells
- Fully flexible datapath widths
- Fine grained memory blocks
- Many high speed interfaces
