Compute Infrastructure

Introducing our new GPU family for IBM Cloud virtual servers

Share this post:

Over the last few months, we’ve been aggressively rolling out new features for IBM Cloud virtual servers. As we keep that momentum rolling in 2018, you’ve been asking for accelerated computing at scale—especially in an AI era.

 Accelerated computing in an AI era

Think about the new workloads and advancements in technology around machine learning, deep learning, and neural networks. They’re all synergized around business transformation and making sense of data. Improving weather forecasting and numeric modeling for climate research, genomic sequencing, image and text recognition, and structural simulations are just a few applications requiring compute accelerators.

Scale and rapid provisioning are inextricably tied with cloud and are non-negotiable—even when provisioning compute with accelerators to enable these workloads.

 Meet our new virtual server GPU family

We’re excited to announce our new GPU family—accelerated compute “ac1” flavors. They’re powered by NVIDIA Tesla P100 GPUs, initially available in DAL13 (with plans to expand globally) and come in the following flavors with block and local SSD storage:

These GPU flavors are built on NVIDIA’s Pascal architecture, accelerating both common HPC workloads such as Monte Carlo simulations and AI applications.

As GPU compute requirements dynamically change, you can expect the same experience in scaling up or down with rapid provisioning on virtual servers.

 Complementing GPU support across compute types

We currently support GPUs on bare metal and we were first to market with the P100s.

With GPU support on virtual servers, you unlock the raw power interoperability of bare metal, complement scale, and rapid provisioning.

The workloads you used to run on bare metal now extend to and burst for scale on virtual servers—allowing for more dynamism for compute-accelerated workloads.

There are many considerations when choosing compute options with GPUs. You’ve got steady state GPU workloads requiring isolation, customization, and power. You can scale workloads, such as Monte Carlo simulations that run for a few hours a day. It’s ultimately about providing choice for any workload you run across your organization.

P100 performance on IBM Cloud is another critical factor as we enable GPU support on virtual servers. It provides enhanced performance in deep learning workloads, no matter which compute type is provisioned.

In the near future, you can expect us to deliver more features on IBM Cloud virtual servers. Visit documentation to learn more  about our GPU family or to get started. You can also connect with me at @milan3patel to engage in cloud chatter.

Add Comment
No Comments

Leave a Reply

Your email address will not be published.Required fields are marked *

More Compute Infrastructure stories

Accelerate and streamline AI and HPC workloads with new NVIDIA GPUs on IBM Cloud

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 the performance and speed that enterprises need. That means AI models that once needed weeks of computing resources can now be trained in just a few hours.

Continue reading

Next Generation Intel® Xeon® E3-1270 v6 Processors now available on IBM Cloud

We're excited to announce the global availability of our bare metal servers powered by the new Intel® Xeon® E3-1270 v6 processor. With up to a 56 percent performance improvement compared to an Intel® second generation E3-1200 processor, the E3-1200 v6 processor-family delivers performance to manage today’s most demanding cloud workloads.

Continue reading

Get greater infrastructure visibility with our new post-install monitor tool

To maximize your user experience, we’ve upgraded our post-installation script system. Learn about the updated system.

Continue reading