Benchmark Tests Show Dramatic Advantages in Latency and Power-Efficiency Achieved by Innovative Design Approach
Santa Clara, CA, USA, October 25, 2018 – NovuMind Inc., a leading innovator in full-stack Artificial Intelligence technologies, today released performance details of its NovuTensor AI chip, with class-leading power efficiency. NovuTensor, purpose-built for deep neural network inference computations, will expand the frontiers of AI by enabling mass deployment of AI inference models at the edge of the network.
NovuMind has produced its first silicon chips and is releasing benchmark test results showing performance superior to the most advanced GPUs currently available for AI inference applications. The benchmark tests were run with a variety of networks including ResNet18, ResNet34, ResNet50, ResNet70, VGG16, and Yolo2.
According to Dr. Ren Wu, founder and CEO of NovuMind, “Until now, GPUs have powered the advances in AI, particularly around the training of deep neural network models from large sets of data. Once models are trained, however, the challenge is to deploy them at scale. GPUs and other processors are expensive and consume large amounts of power. Their architectures are optimized for 2-dimensional matrix computation. While they perform well when processing large batches of data, these chips are not suited for real-time applications that require low latency. They also lack power efficiency and they tend to be very expensive. With the arrival of our NovuTensor chip, we are breaking these barriers and ushering in a new era where AI can be deployed at scale.”
Dr. Wu continued to say that the first NovuTensor chips have recently been delivered from the foundry and are operating in NovuMind’s lab. “We produced our first chip using a conservative 28nm semiconductor process, to validate the design. I am pleased to announce that the first chips work perfectly and our testing validates the superior performance we were expecting. We outperform the most advanced chips from the competition, even those costing thousands of dollars. As we migrate our design to more advanced semiconductor processes such as 16nm or 7nm, our advantage will extend even further.”
NovuTensor performance is achieved by a patented design that natively performs 3-dimensional tensor computations on chip, avoiding the overhead inherent in other chips which require tensors to be unfolded into 2-dimensional matrices. This fundamentally new approach, covered by US Patent No. 10,073,816, is optimized for convolution-based deep neural networks.
The following test results show how NovuTensor compares to a market leading GPU:
NovuMind will soon be announcing a family of products based on NovuTensor. Initial products will include PCI Express AI accelerator cards and a Developers Kit.
NovuMind helps companies put the power of AI into their products. The NovuTensor chip, along with a full stack of AI solutions support, enables OEMs to embed deep learning inference capability, even in applications where power or network access are limited. Led by AI industry pioneer Ren Wu, the NovuMind team has expertise in machine learning, chip design, and high-performance computing. The company’s headquarters and R&D are located in Santa Clara CA, with additional offices in Beijing and Guangzhou, China. For more information, please visit us at: www.novumind.com