NVIDIA Virtual GPUs give you near bare metal performance in a virtualized environment, maximum utilization, management and monitoring, in a hypervisor-based virtualization environment for GPU-accelerated AI.
Deep Learning Training Performance Scaling with vCS on NVIDIA A100 Tensor Core GPUs
Developers, data scientists, researchers, and students need a massive amount of compute power for deep learning training. Our A100 Tensor Core GPU accelerates the workload, letting them do more faster. NVIDIA software, the Virtual Compute Server, delivers nearly the same performance as bare metal, even when scaling to large deep learning training models that use multiple GPUs.
Average Images per Second (normalized)
Deep Learning Inference Throughput Performance with MIG on NVIDIA A100 Tensor Core GPUs using vCS
Multi-instance GPU (MIG) is a technology, only found on the NVIDIA A100 Tensor Core GPU, that partitions the A100 GPU into as many as seven instances, each fully isolated with their own high-bandwidth memory, cache, and compute cores. MIG can be used with Virtual Compute Server, one VM per MIG instance. The performance is consistent when running an inference workload across multiple MIG instances on both bare metal and virtualized with vCS.
Average Images per Secon (normalized)