NeuralBoost
FIRST real DPU for all FPGAs
Engineered for the most demanding applications, NeuralBoost delivers failsafe, energy-efficient, and adaptable solutions for next-generation systems.
NeuralBoost is the groundbreaking Deep Processing Unit (DPU) engineered to bring advanced AI acceleration to any FPGA platform. Whether you’re developing edge computing solutions, embedded AI systems, or next-gen robotics, NeuralBoost delivers unmatched performance, ultra-low power consumption, and failsafe reliability. Unlike traditional solutions, NeuralBoost offers full compatibility with open-source tools and supports a wide range of neural networks – including CNNs, RNNs, and Transformers – without vendor lock-in.
Designed for industrial, medical, automotive, and aerospace markets, NeuralBoost transforms how you deploy machine learning on FPGAs, offering true scalability, flexibility, and future-proof innovation.
What Sets NeuralBoost Apart?
- Seamless AI Inference Across Models: NeuralBoost enables AI inference without model restrictions, supporting a variety of neural network architectures including CNNs, RNNs, and Transformers
- Open-Source Toolchain Compatibility: Built to integrate effortlessly with open-source frameworks like TensorFlow and PyTorch, eliminating the need for proprietary toolchains
- Universal FPGA Compatibility: Runs without restrictions on any FPGA or FPGA SoC device, ensuring broad applicability across hardware platforms
Features
- Custom Operating System: Tailored OS solutions to meet specific application needs.
- Real-Time/Determinism: Ensures predictable performance for time-sensitive applications.
- Bare-Metal Operation: Operates directly on hardware for maximum efficiency.
- Resource Configuration: Fine-grained control over hardware resources without the need for recompilation.
- Multiple Model Support: Handles various AI models simultaneously, enhancing versatility.
- Lightweight C/C++ APIs: Facilitate easy integration into existing software environments.
- MicroBlaze™ Compatibility: APIs to control the IP from AMD MicroBlaze™, enabling support for devices without ARM cores.
- Fine-Grained Recompilation: Allows targeting of different hardware resources for optimized performance.
- Automatic Layer Scheduling: Non-supported layers are automatically scheduled on the ARM NEON accelerator if an ARM core is available.
Compatibility
NeuralBoost is compatible with a wide range of FPGA devices:
- AMD Zynq™ 7000
- AMD Zynq™ UltraScale™
- AMD Artix™/Spartan™ UltraScale™
- AMD Virtex™ UltraScale™
- AMD Versal™ AI Core/AI Edge
- AMD Versal™ Prime/Premium
- Microchip PolarFire™/PolarFire™ SoC
If you want to know more about this product or pricing, contact us at staff@www.makarenalabs.com.
