Venu Tiwari

Research and Development Engineer in VMware Cloud Foundation (VCF) at Broadcom

Venu Tiwari is an R&D engineer at Broadcom. She works in the VMware Cloud Foundation (VCF) Performance team, specializing in GPU virtualization performance automation with ML on vSphere/VCF. She has more than 11 years of industry experience related to virtualization and cloud infrastructure performance.

Author's Posts

Performance

Boost throughput with scaling VMs while keeping the GPUs to a minimum

This article describes some performance tests we conducted to explore the advantages of virtualizing NVIDIA GPUs with VMware vSphere for generative AI workloads. We tested LLAMA2-7b and LLAMA2-13b parameter models using various virtual GPU (vGPU) and multi-instance GPU (MIG) configurations on vSphere 8.0 U3 with NVIDIA vGPU driver v17.

Venu Tiwari

August 27, 2024