Home > Storage > PowerFlex > White Papers > Dell Validated Design for Virtual GPU with VMware and NVIDIA on PowerFlex > NVIDIA A100 Tensor Core GPU
The NVIDIA A100 Tensor Core GPU delivers unprecedented acceleration at every scale to power the highest-performing elastic data centers for AI, data analytics, and HPC. This GPU uses the NVIDIA Ampere Architecture. The third generation A100 provides higher performance than the prior generation and can be partitioned into seven GPU instances to dynamically adjust to the shifting demands.
The Tensor Core technology included in the Ampere architecture brings dramatic performance gains to AI workloads. The A100 GPU can achieve high acceleration for inference workloads. This technology provides a significant advantage for the data scientist and the organization. IT professionals also benefit from reduced operational complexity by using a single technology that is easy to onboard and manage across use cases.
The A100 GPU is a dual-slot 10.5 inches. PCI Express (PCIe) Gen4 card that is based on the NVIDIA Ampere architecture. It uses a passive heat sink for cooling. The A100 PCIe based GPU supports double precision (FP64), single precision (FP32), and half precision (FP16) compute tasks. It also supports unified virtual memory, and a page migration engine. The A100 GPU is available in 40 GB and 80 GB memory versions.
For more information, see NVIDIA A100 Tensor Core GPU documentation.
The Multi-Instance GPU (MIG) feature allows the A100 GPU to be portioned into discrete instances, each fully isolated with its own high-bandwidth memory, cache, and compute cores.
When combining MIG with the NVIDIA vGPU capabilities included with NVIDIA AI Enterprise software, enterprises can take advantage of the management, monitoring, and operational benefits of VMware server virtualization. VMware virtual machines (VMs) with MIG-backed vGPUs provide the flexibility to have a mixed-sized (heterogenous) partitioned GPU instance.
The MIG allows multiple vGPUs (and VMs) to run in parallel on a single A100 GPU. The MIG preserves the isolation of different MIG slices assigned to separate VMs. Administrators can partition the GPUs and allocate the required GPU capacity to individual data scientists. The data scientist can be assured of predictable performance due to the isolation and quality of service guaranteed by the vGPU.
For more information about MIG, see NVIDIA Multi-Instance GPU.