Edgecortix adds AI to Bittware’s Intel FPGA cards
Fabless芯片设计师Edgecortix已与Bittware合作，加速了Intel Agilex F系列FPGA的机器学习应用程序。
动态Ne EdgeCortix嵌入其专有的ural Accelerator (DNA) Intellectual Property (IP) into BittWare’s IA-420F and IA-840F enterprise-class Agilex F-Series based FPGA Accelerator Cards, creating a high performance, low-latency inference solution using standard machine learning frameworks.
The EdgeCortix DNA is a flexible IP core for deep learning inference with up to 20 trillion operations per second (TOPS) of dedicated AI compute capability, low latency, and scalable inference engines, specially optimized for streaming (low batch sizes) and high-resolution data. DNA IP core in combination withEdgeCortix MERAcompiler software framework, enables seamless acceleration of today’s increasingly complex and compute intensive AI workloads on 2nd generation HyperFlex architecture-based Agilex FPGAs.
This is an initial phase for EdgeCortix in the BittWare Partner Programme.
“We are excited to bring the power of EdgeCortix’s Dynamic Neural Accelerator IP to Agilex FPGAs through the new BittWare Partner Program,” said Sakyasingha Dasgupta, Founder and CEO of EdgeCortix. “It’s a game changer on multiple levels in the AI acceleration market. Together with BittWare, we are delivering a complete solution from servers to edge boxes that can be used as ‘drop-in’ replacements for CPUs or GPUs.”
“From an end customer perspective, deploying the AI inference applications requires no FPGA programming experience, and the solutions are easily accessible to software and machine learning engineers using standard frameworks like PyTorch, TensorFlow and ONNX. From a performance perspective these new solutions deliver significantly lower inference latency on streaming data with up-to a 7X performance advantage compared to competing FPGA board offerings,” he said.
+MORE
