This section describes building a Proof of Concept (POC), sizing your VDI environment, general prerequisites, and general preparatory steps that must be addressed before deployment.
You should test your unique workloads to determine the best NVIDIA virtual GPU solution to meet your organizational needs and goals. The most successful customer deployments start with a proof of concept (POC) and are “tuned” throughout the deployment lifecycle. Beginning with a POC enables customers to understand the expectations and behavior of their users and optimize their deployment for the best user density while maintaining required performance levels. Continued monitoring is essential because user behavior can change throughout a project or an individual change within the organization. A user that was once a light graphics user (vApps, vPC) might become a heavy graphics user via professional visualization (RTX vWS) when they change teams and/or projects.
Consider the following during your POC:
A comprehensive review of all user groups, their workloads, applications utilized, and current and future projections should be considered
A vision for balancing user density with end-user experience measurements and analysis
Gather feedback from IT and end-users regarding infrastructure and productivity needs.
Based on your Proof of Concept (POC), we recommend sizing an appropriate environment for each user group you are trying to reach with your evaluation. NVIDIA provides in-depth sizing guides to assist with the process of optimally scaling your organization’s workloads.
Please refer to the appropriate sizing guides below to build your NVIDIA vGPU environment:
As an overview:
Scope your environment for the needs of each end-user
Run a proof of concept for each deployment type
Implement the NVIDIA recommended sizing methodology
Utilize benchmark testing to help validate your deployment
Utilize NVIDIA-specific and industry-wide performance tools for monitoring
Ensure performance and experience metrics are within acceptable thresholds
It is essential to resize your environment when switching from Maxwell GPUs to newer GPUs like the Pascal, Turing, and Ampere GPUs. NVIDIA GPU Architectures following the Maxwell Generation of GPUs may support ECC memory which incurs a 1/15 overhead cost on the GPU frame buffer. Additional information can be found here
The following elements are required to install and configure vGPU software on VMware vSphere:
NVIDIA-certified servers with NVIDIA cards see the web page NVIDIA vGPU Partners for a list of certified NVIDIA servers. It is helpful to cross-check this list with the VMware HCL to ensure compatibility for your deployment. The following specifications are recommended:
CPU for vPC/vApps:
Intel Xeon Gold 6338 @2.0 GHz or faster
AMD EPYC 7713 @2.0 GHz or faster
CPU for RTX vWS
Intel Xeon Gold 6354 @3.0 GHz or faster
AMD EPYC 7763 @3.0 GHz or faster
NoteWhen considering CPUs for your vGPU deployment, NVIDIA recommends the following:
vPC deployments should have a higher core count and lower clock speeds to prioritize density.
RTX vWS deployments should have a higher clock speed priority core count to priority performance.
High-speed RAM
Fast networking, e.g., best-in-class Mellanox ConnectX certified by NVIDIA
High-performance storage with Virtual SAN or high IOPS storage system
High-performance endpoints for testing access
If you use local storage, input/output operations per second (IOPS) play a significant role in performance. If you are using VMware for a Virtual SAN, see the VMware Virtual SAN requirements website for more details.
The following elements are required to install and configure vGPU software on VMware vSphere. For demonstration purposes, this guide uses ESXi release 8.0.2.
Review the Getting Started page of the VMware Horizon documentation. It provides a roadmap for implementing Horizon as a server with Citrix clients and links to a list of VMware education courses and other resources.
The requirements to install and configure vGPU software on VMware vSphere are the same for implementing VMware Horizon as a Server. Refer to the Choosing Your Hardware section for the requirements.
Ensure to use the appropriate NVIDIA GPU for your use case. Refer to the NVIDIA Virtual GPU Positioning Guide to better understand which GPU suits your deployment requirements. For additional guidance, contact your NVIDIA and VMware sales representatives.
VMware vSphere 8.0.2 build:
vSphere and vCenter Server are available from the VMware website at Product Evaluation Center for VMware vSphere 8.0.
As of vSphere 7.0, deploying a new vCenter Server or upgrading to vCenter Server 7.0 requires using the vCenter Server Appliance (VCSA).
VMware Horizon software:
If needed, you may register for a trial to obtain the license keys required for various elements to deploy and manage Horizon at the web page Welcome to My VMware.
NVIDIA vGPU software:
NVIDIA vGPU manager vSphere Installation Bundle (VIB)
The vGPU Manager VIB is loaded similarly to a driver in the vSphere hypervisor, and the vCenter Server then manages it.
For NVIDIA vGPU software builds latest releases, please refer to your NVIDIA Application Hub.
For additional information on these current releases, please consult NVIDIA vGPU product documentation for VMware vSphere.
Microsoft software:
Refer to the vGPU Software documentation for a supported Windows Guest OS list.
Your choice of one of the following CLI/SSH/SCP tools installed on your Windows-based toolbox PC:
MobaXterm (SSH and SPC) is available from the MobaXTerm website; This is the recommended tool.
Putty (SSH) and WinSPC (SPC), available from the Putty website and the WinSCP website.
Tight VNC Viewer (SSH) is available from the Tight VNC website.
Licenses:
From the VMWare Horizon website:
vSphere/ESXi vSphere Foundation is required to use vGPU on vSphere
VM vSphere
VMware Horizon
Microsoft licenses can be found at the service center here <https://www.microsoft.com/Licensing/servicecenter/default.aspx>`__ (volume licenses recommended)
Testing and Benchmarking tools (Optional but recommended):
GPU Profiler - GPUProfiler download
Lakeside Systrack with GPU monitoring
End-User satisfaction survey
A Virtual GPU Certified Server on which to install VMWare Horizon
VMware vSphere ESXi
VMware vCenter
Must be joined to a domain
Must be assigned a static IP address
VMware Horizon consists of many components working together in concert. Many environments are built on the vSphere virtual infrastructure. Refer to the VMware Horizon Documentation page for a complete list of components and features.
The versions of vCenter and ESXi supported by Horizon 8 can be found in the VMware Product Interoperability Matrix.
Before you install NVIDIA vGPU software:
Determine how vSphere will run on the physical hosts. Consider booting from a thumb drive.
Download and install any of the following for SSH and SCP:
An SSH tool (such as PuTTY)
WinSCP, which handles both SSH and SCP functions
MobaXterm, which handles both SSH and SCP functions
Tight VNCViewer (Remote console)
Configure the BIOS for your physical hosts as described below:
Hyperthreading - Enabled
Power Setting or System Profile - High Performance
CPU Performance (if applicable) - Enterprise or High Throughput
Memory Mapped I/O above 4-GB - Enabled (if applicable)
SR-IOV enabled
VT-d/IOMMU - Enabled
NVIDIA GPU architectures after the Maxwell architecture (Pascal, Turing, & Ampere) support VT-d and IOMMU.