Artificial intelligence (AI) and machine learning (ML) are two technologies that have taken the world by a storm. Today, these two technologies are not only being used in scientific research, but also they are being used in banks and financial institutions and healthcare services. Now, Japanese electronics and IT giant NEC has developed the world’s smallest supercomputer that can perform AI and ML related tasks with ease.
NEC has launched SX-Aurora TSUBASA AI platform, which brings supercomputer scalar- and vector-processing performance, along with a large memory subsystem, to a broad range of applications. NEC says that its compact and cost-effective PCIe card form factor fits in everything from workstation towers for a single card, to rack-mounted servers for multiple cards.
In the heart of NEC’s SX-Aurora TSUBASA AI platform is a supercomputer called the NEC Vector Engine Processor, which offers high performance and low power consumption. The Vector Engine Processor comes with six HBM2 memory modules all using chip-on-wafer-on-substrate technology, which leads to the memory bandwidth of 1.2 TB/s.
The NEC Vector Engine Processor works in combination with single instruction multiple data (SIMD) pipelining and the company’s Frovedis middleware. “…this AI Platform accelerates data analysis operations on top of the Apache Spark MLlib and DataFrame framework, while maintaining compatibility with standard programming languages such as C, C++ and Fortran,” NEC wrote on its product page.
The company says that its newly launched platform that ideal for large scale e-commerce recommendation engines, high-throughput financial and document transactions, fraud detection, enterprise governance, and authentication. NEC offers multiple variants of its SX-Aurora TSUBASA AI platform with the top configuration, A500-64, offering a maximum memory capacity of 384GB and maximum vector engine performance of 157.28 TFLOPS. The company hasn’t specified the price of the platform but you can contact the company to get the quotes.