PNY Technologies Inc.
0

GPU Virtualization Icon GPU Virtualization Icon

NVIDIA Professional Graphics For

GPU Virtualization

GPU Virtualization For Every Workload - From VDI to AI
GPU accelerated computing is on the rise. Enterprises are using AI-based applications to help drive better business decisions, boost productivity, and design better products that get to market faster. NVIDIA® virtual GPUs (vGPUs) enable enterprises to efficiently deploy GPUs in the data center to support every workload, from virtual client computing to the most compute-intensive workloads like AI, data science, and high-performance computing (HPC).

Why NVIDIA Virtual GPUs?

Exceptional User Experience

Exceptional User Experience

Ultimate user experience with the ability to support both compute and graphics workloads in hypervisor-based virtualization environments.

Predictable Performance

Predictable Performance

Consistent performance and quality of service, both on-premises and in the cloud.

Continuous Innovation

Continuous Innovation

Regular software releases ensure that you stay up to date with the latest features and enhancements.

Best User Density

Best User Density

Industry’s highest user-density solution with support for up to 32 virtual desktops per physical GPU. Lower TCO with more than eight vGPU profiles for the most flexibility to provision resources to match your users’ needs.

Optimal Management and Monitoring

Optimal Management and Monitoring

True, end-to-end management and monitoring with real-time insight into GPU performance and migration technology, as well as broad partner integrations that enable you to use the tools that you know and trust.

Broadest Support Ecosystem

Broadest Support Ecosystem

Support for all major hypervisors, over 140 certified servers, and an extensive portfolio of professional apps accelerated by GPUs.

NVIDIA Data Center GPUs

Products NVIDIA A100 PCIe NVIDIA A30 NVIDIA A40 NVIDIA A10 NVIDIA T4 NVIDIA A16
PNY Part Number NVA100TCGPU-KIT NVA30TCGPU-KIT NVA40TCGPU-KIT NVA10TCGPU-KIT TCSCT4-KIT NVA16TCGPU-KIT
Workload Description Highest Performance Compute Mainstream Compute Highest Performance Graphics Mainstream Graphics Small Footprint Low Power Small Footprint Low Power
  Recommended Number of GPUs per Server
Deep Learning (DL) Training and Data Analytics For the absolute fastest model training and analytics 4-8 GPUs 40GB: Bn+ parameter models (DLRM, GPT-2)          
DL Inference For batch and real-time inference 1-2 GPUs w/ multiinstance GPU (MIG) 40GB: large batch size constrained models (RNN-T) 2–4 GPUs with MIG   4-8 GPUs 4-8 GPUs  
High-Performance Computing (HPC) / AI For Higher Education Research and scientific computing centers 1-4 GPUs with MIG 2–4 GPUs with MIG        
Render Farms For batch and real-time rendering     4-8 GPUs 4-8 GPUs    
Graphics For the best graphics performance on professional VDI     2–4 GPUs for high-end virtual workstations* 2-8 GPUs for mid-range virtual workstations* 2-8 GPUs for entry-level virtual workstations* 2–4 GPUs for highest virtual desktop user density**
Cloud Gaming For 4K resolution / Android     4-8 GPUs (4K resolution) 4–8 GPUs (4K resolution) 1-2 GPUs (Android)  
Enterprise Acceleration For mixed workloads, including graphics, ML, DL, analytics, training, and inference 1-2 GPUs with MIG for compute workloads 1-2 GPUs with MIG for compute workloads 1-2 GPUs for graphics-intensive workloads* 1-2 GPUs for graphics-intensive* and compute workloads 1-4 GPUs for balanced workloads*  
Edge Acceleration For differing use cases and deployment locations 1-2 GPUs with MIG 1-2 GPUs with MIG 1-4 GPUs for graphics-intensive workloads & AR / VR* 1-8 GPUs for inference and video workloads 1-8 GPUs for inference and video workloads  

* NVIDIA RTX Virtual Workstation (vWS) software license required for virtual workstation workloads.
** NVIDIA Virtual PC (vPC) software license required for VDI workloads.

NETFLIX

Lost in Space

Three GPU-accelerated virtual machines working simultaneously on one RTX Server.

Netflix is a registered trademark of Netflix

INDUSTRY SOLUTIONS

Professionals in every industry are realizing the power of GPU-accelerated virtual desktops and workstations. NVIDIA virtual GPU technology drives increased productivity, manageability, and security across these key industries.

AEC Manufacturing Media & Entertainment Higher Education Government Healthcare Energy Exploration (Oil and Gas) Finance

vGPU AND THE NVIDIA QUADRO RTX 8000 AND RTX 6000

AUTHORIZED BY NVIDIA FOR VIRTUAL GPU DATA CENTER DEPLOYMENT

NVIDIA RTX™ features such as RT Cores for real-time ray tracing at cinematic quality, Tensor Cores that accelerate AI/DL/ML/MV applications and big data analytics, or AI enhanced design and visualization tools, make NVIDIA Quadro RTX 8000 or NVIDIA Quadro RTX 6000 active or passive professional graphics boards ideal candidates for implementing virtual GPU (vGPU) capabilities across your organization. Bring the latest NVIDIA professional gpu advances in graphics and AI to products ranging from smartphones to tablets and non-NVIDIA GPU equipped mobile or desktop PCs, with NVIDIA Quadro RTX 8000 and NVIDIA Quadro RTX 6000 active or passively cooled virtualization solutions.

NVIDIA now authorizes and supports the deployment of NVIDIA Quadro RTX 8000 and NVIDIA Quadro RTX 6000 active or passive professional graphics boards in server chassis for vGPU hardware infrastructure. Either choice offers outstanding GPU performance, but the NVIDIA Quadro RTX 8000 boards offer an unprecedented 48 GB of GPU memory, while the NVIDIA Quadro RTX 6000 offerings provide 24 GB. All utilize ultra-fast GDDR6 with optional ECC, and NVLink offers GPU memory pooling for two cards, providing 96 GB or 48 GB respectively, along with performance scaling since GPU core counts are effectively doubled. Since NVLink provides 100 GB/sec of bidirectional communications between two NVIDIA professional graphics boards, far in excess of what PCIe provides, NVIDIA RTX equipped vGPU hardware for data center deployment offers previously unrealizable levels of performance and paradigm shifting capabilities – all with NVIDIA GPU IT manageability.

SERVER VENDORS OFFERING UP TO 8X RTX 8000 OR RTX 6000 ACTIVELY COOLED SYSTEMS FOR vGPU DEPLOYMENT

Vendor Model Number | Part Number Type of GPU Supported Number of GPUs Supported
Advance HPC Mercury RM408 RTX 8000 or RTX 6000 Active 4x
Advance HPC SM 4029GP-TR2 RTX 8000 or RTX 6000 Active 8x
ASUS Computer ESC4000 G4 RTX 8000 or RTX 6000 Active 4x
ASUS Computer ESC8000 G4 RTX 8000 or RTX 6000 Active 8x
COLFAX CX41060S-XK7 RTX 8000 or RTX 6000 Active 8x
EXXACT Corporation Tensor EX TS2673917-RTX RTX 8000 or RTX 6000 Active 4x
EXXACT Corporation Tensor EX TS41337043-RTX RTX 8000 or RTX 6000 Active 8x
GIGABYTE W42G-PO8R RTX 8000 or RTX 6000 Active 4x
GPL Technologies SM 4029GP-TR2 RTX 8000 or RTX 6000 Active 8x
Images & Technologies IT-4200 RTX 8000 or RTX 6000 Active 8x
Penguin Computing Relion X2114GT RTX 8000 or RTX 6000 Active 4x
RAVE COMPUTER RAVE-AS RTX-4 RTX 8000 or RTX 6000 Active 4x
RAVE COMPUTER RAVE-AS RTX-8 RTX 8000 or RTX 6000 Active 8x
Supermicro SMC SS4019GP-TRT2 RTX 8000 or RTX 6000 Active 8x
Supermicro SMC SYS-709GP-TRT RTX 8000 or RTX 6000 Active 4x
Supermicro SMC SYS 1019GP-TT RTX 8000 or RTX 6000 Active 2x
TYAN Computer Thunder HX FT-B7105 RTX 8000 or RTX 6000 Active 4x
TYAN Computer Thunder HX FA77-B7119 RTX 8000 or RTX 6000 Active 10x

SERVER VENDORS OFFERING UP TO 10X RTX 8000 OR RTX 6000 PASSIVELY COOLED SYSTEMS FOR vGPU DEPLOYMENT

Vendor Model Number | Part Number Type of GPU Supported Number of GPUs Supported
ASRock Rack 3U8G-C621 RTX 8000 or RTX 6000 Active 8X
ASUS Computer ESC4000G4 RTX 8000 or RTX 6000 Active 4X
ASUS Computer ESC8000G4 RTX 8000 or RTX 6000 Active 8x
BOXX Technologies APEXX D8R RTX 8000 or RTX 6000 Active 8x
Inspur NF5468M5 RTX 8000 or RTX 6000 Active 8x w/NVLink
Supermicro SYS-1019GP RTX 8000 or RTX 6000 Active 2x
Supermicro SYS-1026GP RTX 8000 or RTX 6000 Active 3x
Supermicro SYS-1029GQ RTX 8000 or RTX 6000 Active 4x
Supermicro SYS-2029U/6029U RTX 8000 or RTX 6000 Active 4x
Supermicro SYS-7049-TRT RTX 8000 or RTX 6000 Active 4x
Supermicro SYS-2029GP-TR RTX 8000 or RTX 6000 Active 6x
Supermicro SYS-4029GP-TRT2 RTX 8000 or RTX 6000 Active 8x
Supermicro SYS-4029GP-TRT RTX 8000 or RTX 6000 Active 8x
TYAN Computer TN76-B7102 RTX 8000 or RTX 6000 Active 4x
TYAN Computer FT48T-B105 RTX 8000 or RTX 6000 Active 5x
TYAN Computer FT77D-B7109 RTX 8000 or RTX 6000 Active 8x
TYAN Computer FA77-B7719 RTX 8000 or RTX 6000 Active 10x

NVIDIA QUADRO RTX 8000 AND RTX 6000 ACTIVE v. PASSIVE PERFORMANCE

RTX 8000 & RTX 6000 Active RTX 8000 & RTX 6000 Passive Performance Delta
RTX-OPS 84 T 84 T None
Rays Cast 10 Giga Rays/sec 10 Giga Rays/sec None
Peak FP32 Performance 16.3 TFLOPS 14.9 TFLOPS Passive -8.60%
Peak FP16 Performance 32.5 TFLOPS 29.9 TFLOPS Passive -8.60%
Peak INT8 Performance 261.0 TOPS 238.9 TOPS Passive -8.46%
Deep Learning TFLOPS 130.5 Tensor TFLOPS 119.4 Tensor TFLOPS Passive -8.50%
Maximum Power Consumption TGP 260 W | TBP 295 W TGP 250 W Passive TGP 10 W Less
Power Connectors 8-pin + 6-pin PCIe 8-pin PCIe Passsive only requires 8-pin PCIe

THIS TABLE ILLUSTRATES RTX 8000 vGPU TYPES BY INTENDED USE CASE, DISPLAY INFORMATION, vGPUs SUPPORTED BY GPU AND BOARD, AND THE REQUIRED BOARD LICENSE EDITION

Virtual GPU Type Use Case GPU Memory (MB) Virtual Display Heads Maximum Resolution per Display Head Maximum vGPUs per GPU Maximum vGPUs per Board Required License Edition
RTX8000-48Q Virtual Workstations 49152 4 4096 x 2160 1 1 NVIDIA Quadro vDWS
RTX8000-24Q Virtual Workstations 24576 4 4096 x 2160 2 2 NVIDIA Quadro vDWS
RTX8000-16Q Virtual Workstations 16384 4 4096 x 2160 3 3 NVIDIA Quadro vDWS
RTX8000-12Q Virtual Workstations 12288 4 4096 x 2160 4 4 NVIDIA Quadro vDWS
RTX8000-8Q Virtual Workstations 8192 4 4096 x 2160 6 6 NVIDIA Quadro vDWS
RTX8000-6Q Virtual Workstations 6144 4 4096 x 2160 8 8 NVIDIA Quadro vDWS
RTX8000-4Q Virtual Workstations 4096 4 4096 x 2160 12 12 NVIDIA Quadro vDWS
RTX8000-3Q Virtual Workstations 3072 4 4096 x 2160 16 16 NVIDIA Quadro vDWS
RTX8000-2Q Virtual Workstations 2048 4 4096 x 2160 24 24 NVIDIA Quadro vDWS
RTX8000-1Q Virtual Workstations 1024 2 4096 x 2160 32 32 NVIDIA Quadro vDWS
RTX8000-48C Virtual Workstations 49152 1 4096 x 2160 1 1 vComputeServer or NVIDIA Quadro vDWS
RTX8000-24C Virtual Workstations 24576 1 4096 x 2160 2 2 vComputeServer or NVIDIA Quadro vDWS
RTX8000-16C Virtual Workstations 16384 1 4096 x 2160 3 3 vComputeServer or NVIDIA Quadro vDWS
RTX8000-12C Virtual Workstations 12288 1 4096 x 2160 4 4 vComputeServer or NVIDIA Quadro vDWS
RTX8000-8C Virtual Workstations 8192 1 4096 x 2160 6 6 vComputeServer or NVIDIA Quadro vDWS
RTX8000-6C Virtual Workstations 6144 1 4096 x 2160 8 8 vComputeServer or NVIDIA Quadro vDWS
RTX8000-4C Virtual Workstations 4096 1 4096 x 2160 12 12 vComputeServer or NVIDIA Quadro vDWS
Note Physical GPUs per board: 1 | C-series virtual GPU Types are supported only since release 9.1

THIS TABLE ILLUSTRATES RTX 6000 vGPU TYPES BY INTENDED USE CASE, DISPLAY INFORMATION, vGPUs SUPPORTED BY GPU AND BOARD, AND THE REQUIRED BOARD LICENSE EDITION

Virtual GPU Type Use Case GPU Memory (MB) Virtual Display Heads Maximum Resolution per Display Head Maximum vGPUs per GPU Maximum vGPUs per Board Required License Edition
RTX6000-24Q Virtual Workstations 24576 4 4096 x 2160 1 1 NVIDIA Quadro vDWS
RTX6000-12Q Virtual Workstations 12288 4 4096 x 2160 2 2 NVIDIA Quadro vDWS
RTX6000-8Q Virtual Workstations 8192 4 4096 x 2160 3 3 NVIDIA Quadro vDWS
RTX6000-6Q Virtual Workstations 6144 4 4096 x 2160 4 4 NVIDIA Quadro vDWS
RTX6000-4Q Virtual Workstations 4096 4 4096 x 2160 6 6 NVIDIA Quadro vDWS
RTX6000-3Q Virtual Workstations 3072 4 4096 x 2160 8 8 NVIDIA Quadro vDWS
RTX6000-2Q Virtual Workstations 2048 4 4096 x 2160 12 12 NVIDIA Quadro vDWS
RTX6000-1Q Virtual Workstations 1024 2 4096 x 2160 24 24 NVIDIA Quadro vDWS
RTX6000-24C Training Workloads 24576 1 4096 x 2160 1 1 vComputeServer or NVIDIA Quadro vDWS
RTX6000-12C Training Workloads 12288 1 4096 x 2160 2 2 vComputeServer or NVIDIA Quadro vDWS
RTX6000-8C Training Workloads 8192 1 4096 x 2160 3 3 vComputeServer or NVIDIA Quadro vDWS
RTX6000-6C Training Workloads 6144 1 4096 x 2160 4 4 vComputeServer or NVIDIA Quadro vDWS
RTX6000-4C Inference Workloads 4096 1 4096 x 2160 6 6 vComputeServer or NVIDIA Quadro vDWS
Note Physical GPUs per board: 1 | C-series virtual GPU Types are supported only since release 9.1

NVIDIA Quadro RTX 8000 and RTX 6000 active or passive vGPU solutions are ideal for use cases where virtual workstations are desired or required. Essential NVIDIA software includes NVIDIA Quadro vDWS, CUDA-X AI and OptiX. Hypervisors and other required ISV software are available from leading vendors including Microsoft, VMware and Citrix (among others). For more information on how NVIDIA Quadro RTX 8000 or RTX 6000 active or passively cooled professional graphics boards energize and enhance vGPU deployments, contact PNY at gopny@pny.com.

Close