Product of the Week: NVIDIA QUADRO RTX 6000

July 21, 2021

Story

Product of the Week: NVIDIA QUADRO RTX 6000

How can you design high-performance systems without a high-performance development platform?

The NVIDIA Quadro RTX 6000 is a workstation graphics and server acceleration card that delivers advanced 3D imaging and AI functionality to professional engineers, data scientists, and researchers who need to truly immerse themselves in the high-end applications they’re developing. This horsepower is brought to you by the NVIDIA Turing architecture, as the card connects two Quadro RTX 6000 GPUs over the 100 GBps NVIDIA NVLink interconnect into a unified platform that includes:

  • More than 4,600 CUDA cores for parallel processing
  • 576 Tensor cores for deep learning 
  • 72 RT cores for ray tracing
  • 1 Encode & 1 Decode Engine

In terms of maximum quantifiable performance, all of that compute yields 16.3 TFLOPS of single-precision FP32 and 130.5 TFLOPS of Tensor performance, 10 GigaRays per second, and 84 RTX-OPS (an NVIDIA-specific measure of ray tracing performance).

This is backed by 24 GB (up to 48 GB in multi-GPU configurations) of next-generation GDDR6 ECC SDRAM that supports huge AI datasets, 8K video content, and more, over a 384-bit memory interface. To enhance these memory capabilities even further, Quadro RTX 6000 cards are outfitted with NVIDIA GPUDirect. GPUDirect is a suite of technologies that provide network adaptors and storage drives with direct read and write access to GPU memory to improve overall system performance by reducing latency, eliminating memory copies, and reducing CPU overhead.



While focused on the deep learning, 3D graphics rendering, modeling, simulation, and other computationally demanding markets, the Quadro RTX 6000 also outfitted with a VirtualLink port for access to high-resolution, head-mounted virtual reality (VR) displays to let designers experience environments they create first-hand. 

Other I/O available on the dual-slot, full-height 4.4” x 10.5” RTX 6000 cards consists of:

  • 16x PCI Express 3.0 interfaces
  • 4x DisplayPorts
  • 1x USB-C
  • Stereo Connector with 3D Stereo Support and VGA via connector/adaptor

The NVIDIA Quadro RTX 6000 in Action

The four DisplayPort interfaces are backed by the HDCP 2.2 copy protection standard and combine with the graphics performance of the Turing GPUs to enable simultaneous displays at resolutions of 4096x2160 @ 120 Hz (four concurrently), 5120x2880 @ 60 Hz (four concurrently), or 7680x4320 @ 60 Hz (two concurrently).

For developers who leverage the graphics cards to render their projects across multiple screens, NVIDIA has packaged the devices with Mosaic multi-display technology that allows you to span applications across as many as 16 high-resolution curved panels or projectors as if they were a single monitor. No visual artifacts or application software modifications are required.

 
NVIDIA RTX Desktop Manager help maximize workflow productivity by organizing and managing applications running across such multi-monitor setups, while support for RTX Virtual Workstation (vWS) makes the physical resources of the Quadro RTX 6000 remotely available to large, distributed teams – many of whom are working from home.

When using the graphics card for their development, engineers will make use of various compute and graphics APIs that provide direct access to hardware resources. These include:

  • CUDA
  • DirectCompute
  • OpenCL
  • OpenGL 4.6
  • Vulkan 1.1
  • DirectX 12.0
  • Shader Model 5.1

The platforms have also been tested and certified for more than 100 apps in the medical, manufacturing, energy, and other industries thanks to advances in programmable shading, rasterization, and AI-enabled workflow acceleration that help independent software vendors (ISVs) develop and render more compelling, immersive software and solutions. 

Getting Started with the NVIDIA Quadro RTX 6000

The first step in getting started with the NVIDIA Quadro RTX 6000 is determining whether this level of performance is right for you. The table below compares the RTX 6000 with the consumer-grade GEFORCE cards, highlighting features such as active and passive cooling used for thermal management of the 295 W card, highly tuned and optimized drivers, and a three-year NVIDIA warranty.


After determining that a professional-grade server/workstation solution is right for you, the next phase towards capturing the Quadro RTX 6000’s visual computing performance is to decide if just one card is enough. If not, compatibility with NVIDIA Quadro Sync II daughter cards allows for the quicky integration and synchronization of up to four Quadro GPUs each. Two Quadro Sync II cards can be used in a workstation, permitting users to harness the processing capabilities of up to eight GPUs in a single system. All the Quadro Sync II cards require is access to a PCI Express slot.

The final step in acquiring this scalable, next-generation development platform is to source it through an NVIDIA partner, such as Arrow Electronics. You can contact Arrow online at www.arrow.com/arrow-services/nvidia#contact, or find out more about the Quadro RTX 6000 using the resources below.

Additional Resources: