Put those GPUs to work: vSphere Bitfusion delivers elastic infrastructure for AI/ML workloads

Promo Imagine, just for a moment, that you acquired a car just to use the radio. Pretty expensive, huh? Waste of resource, no? Now – and you don’t need to imagine this – you have scads of GPUs scattered throughout the organisation, with disparate teams throwing these hardware accelerators at disparate AI/ML workloads. Unfortunately, those accelerators are living in less than splendid isolation and may spend much of their time… not doing very much at all. This is inefficient and expensive.

So, why not share those GPUs in an organisation-wide elastic infrastructure? Not so simple, as the technology to pool GPUs is lacking or, at best, rudimentary. But the days of isolated islands of inefficiently used resources are coming to an end, according to VMware, courtesy of its upcoming vSphere Bitfusion release.

vSphere delivers elastic infrastructure for AI/ML workloads by creating pools of hardware accelerator resources. The best-known accelerators today are GPUs which vSphere can now use to create AI/ML cloud pools that can be used on-demand. GPUs can now be used efficiently across the network and driven to the highest levels of utilization possible. In other words, vSphere Bitfusion allows for the sharing of GPUs in a similar fashion to the way vSphere allowed the sharing of CPUs many years ago.

VMware bought Bitfusion last year for its ability to share hardware accelerators, like GPUs, across the network. The company has integrated the technology into vSphere and reports a positive reception from the analyst community and customers that have participated in either the proof of concept and/or beta phases of this new feature. VMware says vSphere Bitfusion will be generally available by the end of July.

On June 2, VMware and Dell Technologies will discuss vSphere Bitfusion at a live online event entitled "Enabling Real Artificial Intelligence with Dell Technologies and PowerEdge." For more info, register here.

Expect the companies to discuss their AI/ML solutions, which will integrate Bitfusion as another component of joint offerings, and validate designs.

And don’t worry if you are unable to make it on the day – the event is recorded. Just follow this permalink to check out the archived presentation and chat.

Additional Resources:

• VMware’s Hardware Accelerators with Bitfusion Page
• Bitfusion Keynote Demo from VMworld 2019 (Wednesday session, minute 48)
• VMware’s Machine Learning and AI Blog
• Bitfusion Acquisition Blog (from 2019)
• Blog Series on How to Use GPUs