Big Blue’s new servers mix Power8 processors with Nvidia’s NVLink technology
IBM has revealed three Power8 Linux servers designed to power artificial intelligence (AI), deep learning and cognitive computing applications, along with making data centres more efficient.
The new trio of servers feature IBM’s Power8 processors, which Big Blue claims allows for five times the data movement speed of any competing platform. The chips are combined with Nvidia’s NVLink technology to connect the IBM processor directly to Nvida’s range of GPUs designed for high performance computing.
In the case of IBM’s Power Systems S822LC server, two Power8 processors use NVLink to connect to four of Nvidia’s Tesla P100 GPU, which feature the graphics card giant’s latest Pascal architecture.
This setup gives the server the compute power to handle intensive applications that require a large throughput of data, such a cognitive computing tasks carried out by likes of IBM Watson.
Through the work the OpenPower community has undertaken, IBM claims its Power8 servers deliver 80 percent more power per dollar than x86 systems. These boasts may seem like hot air from IBM, but as more companies look to create and run software with AI capabilities and the ability to process huge amounts of data, having servers with significant bang for the buck becomes and attractive proposition.
“The user insights and the business value you can deliver with advanced analytics, machine learning and artificial intelligence is increasingly gated by performance. Accelerated computing that can really drive big data workloads will become foundational in the cognitive era,” said Doug Balog, general manager of power at IBM Systems.
OpenPower is increasingly making its presence felt, with Google and Rackspace both having worked together on OpenPower server design using IBM’s Power 9 processors.
Are you clued up with cloud? Show us your knowledge of cloud news from the last few months.