NVIDIA Reducing the Size and Cost of AI
If the best way to measure the impact a technology is having on the industry-at-large is by observing the sheer volume of hardware/software being introduced to the market and the size of the burgeoning ecosystem around that technology, then AI is making a very big impact indeed. New AI chips are introduced on a weekly (if not daily) basis, all claiming to have set new standards for size, efficiency and performance. To quote a recent whitepaper on machine learning performance, “the myriad combinations of ML hardware and ML software make assessing ML-system performance in an architecture-neutral, representative, and reproducible manner challenging.” The question then becomes: how do you separate the wheat from the chaff? One way is by relying on brand name recognition – if you want something done right, go to the people who know what they’re doing. In that spirit, NVIDIA, a pioneer in fields like computer graphics and parallel computing, recently announced the Jetson Xavier NX, the newest member of NVIDIA’s Jetson family of chips for embedded AI and industrial IoT systems. The Jetson Xavier NX, according to the company, provides server-class performance while consuming as few as 10 watts of power, all in a form factor smaller than a credit card.
The Jetson Xavier NX is designed for embedded edge computing devices that require high performance but are otherwise restricted by factors like size and power requirements. The chip measures 70 x 45mm and provides up to 14 TOPS at 10 watts and up to 21 TOPS at 15 watts, making it an optimal solution for use cases such as handheld medical devices, high-resolution optical sensors, and even commercial and/or industrial drones. Like all members of the Jetson family, the Xavier NX operates on the CUDA-X AI software architecture, meaning outfits already working on embedded machines with other Jetson products won’t incur additional expense or significantly escalate their time to market. The Jetson Xavier NX is pin compatible with the Jetson Nano, making upgrading without significantly altering pre-existing hardware designs relatively simple.
“AI has become the enabling technology for modern robotics and embedded devices that will transform industries,” according to Deepu Talla, VP and General Manager of edge computing at NVIDIA. “Many of these devices, based on small form factors and lower power, were constrained from adding more AI features. Jetson Xavier NX lets our customers and partners dramatically increase AI capabilities without increasing the size or power consumption of the device.”
NVIDIA, primarily known for making GPUs that “revolutionized” the PC gaming industry, has been at the forefront of AI and machine learning in recent years – in fact, the company just excelled in the MLPerf (machine learning performance) Inference v0.5 tests, a set of five benchmarks intended to measure AI performance in three key areas: object detection, machine translation and image classification. Last week, the MLPerf consortium made the results of approx. 600 individual tests that 14 companies willingly submitted their AI-based processors too public. NVIDIA outperformed its competition, with its Turing GPUs delivering higher processing performance than its Intel and Google-designed counterparts. According to Ian Buck, VP of accelerated computing at NVIDIA: ““AI is at a tipping point as it moves swiftly from research to large-scale deployment for real applications. By combining the industry’s most advanced programmable accelerator, the CUDA-X suite of AI algorithms, and our deep expertise in AI computing, NVIDIA can help data centers deploy their large and growing body of complex AI models.”
The Jetson Xavier NX will be available in March 2020 and is priced at $399.00 US. Developers interested in getting a head start on their designs can purchase the Jetson AGX Xavier Development Kit which comes with a software patch to emulate the Xavier NX’s functionality.