Using virtualization to use CPU resources more efficiently in IT is such a familiar concept, it’s hard to think of a time before it was commonplace. The same cannot be said for using GPU resources, which in many organizations still operates at a one node per one user experience. To address this, VMware acquired Bitfusion in 2019 and is now rolling out their tech into vSphere 7. This allows users to request GPU resources using the bitfusion command, which can then aggregate and distribute resources over the network. Christopher Kusek heard some of the initial details at Tech Field Day Extra at VMworld 2019, and is excited to see them rolling out. He’s sees this as evidence of the maturation of ML and AI workloads in the enterprise, and Bitfusion will let organizations make best use of their existing GPU infrastructure.
Bitfusion at VMworld 2019 With TFDx
ML and AI workloads are only going to grow more in the enterprise in the coming years. At VMworld 2019, VMware showed how they are going to help meet the infrastructure needs of this shift. Using assets aquired from Bitfusion, VMware is able to virtualize GPUs, effectively doing for ML workloads what VSAN did for storage. Brandon Graves digs into the architecture that he saw at Tech Field Day Extra at VMworld US 2019 in this post.
