Advanced Computing in the Age of AI | Monday, June 24, 2024

IBM Adds V100 GPUs to IBM Cloud; Targets AI, HPC 

IBM today announced availability of Nvidia V100 GPUs in the IBM Cloud. The move keeps pace with Amazon and Microsoft Azure which also offer V100s and is more evidence of the effort by major cloud providers to rapidly beef up their AI and HPC capabilities. In making the announcement, IBM was quick to note it was “first to offer a comprehensive suite of GPUs including the P100, K80 and M60 on IBM Cloud bare metal and virtual servers.”

John Considine, GM of Cloud Infrastructure Services, IBM, wrote in a blogpost, “Starting today, you can equip individual IBM Cloud bare metal servers with up to two Nvidia Tesla V100 PCIe GPU accelerators – Nvidia’s latest, fastest and most advanced GPU architecture. The combination of IBM high-speed network connectivity and bare metal servers with the Tesla V100 GPUs provides higher throughput than traditional virtualized servers.”

“We’re focused on delivering new AI capabilities both in the cloud and on premises to help enterprises not only gain critical insights from their data, but also create new value with that data. We’ve been working closely with Nvidia to bring their latest GPU technology, Nvidia Tesla V100, to the cloud.”

For now, the V100 will be on x86-based servers in the cloud and only on bare metal servers. Plans for offering V100 on POWER9-based servers in the cloud were not disclosed. “IBM does not comment on the status of future plans. We are working side by side with the IBM Power Systems team to ensure that IBM Cloud will deliver access to the best of IBM technology to allow customers to run HPC and AI workloads,” said a spokesman.

The V100 is available in IBM’s POWER9 server line. Considine wrote, “To power on-premises workloads, IBM also offers the industry’s only CPU-to-GPU Nvidia NVLink connection on our latest POWER9 servers.” POWER9 is IBM latest processor (see HPCwire article,  IBM Begins Power9 Rollout with Backing from DOE, Google). The new processor is being used in Summit and Sierra, two pre-exascale supercomputers funded by DoE.

Like others, IBM is touting its cloud-based ability to handle AI workloads including the ability to speed training of “deep learning models and to create powerful cloud-native applications.” One NASA example cited is fascinating:

NASA Frontier Development Lab this past summer, a team of researchers and data scientists used machine learning techniques on the IBM Cloud to develop new processes for 3D modeling of asteroids from radar data. With an average of 35 new asteroids and near-Earth objects discovered each week, there is currently more data available than experts can keep up with, and existing 3D modeling processes can take several months. Using NVIDIA P100 GPUs on the IBM Cloud and IBM Cloud Object Storage, the team was able to generate asteroid shapes images an average of five to six times faster than previous processes allowed.”

Ian Buck, vice president and general manager of Accelerated Computing at Nvidia is quoted in the blog, “The new IBM Cloud offerings based on our Volta technology provide incredible processing speeds and the ability to scale up or down on demand for HPC and deep learning.” IBM also collaborates with technology partners such as Rescale and Bitfusion to enable instant access to GPUs on IBM Cloud.

Feature image: Inside IBM Cloud, Dallas