The Quadro M6000 24GB, simply put, is the most powerful professional compute and graphics card available, designed specifically for high-volume datasets above 12GB per GPU. Nvidia claims it is intended for simulation-intense workloads, including in areas of seismic exploration, automotive design, large-scale visual effects and emerging virtual reality applications.
Last April, the company introduced the Quadro M6000 12GB, featuring half the VRAM as its successor, up to 7 teraflops of single-precision performance, and support for up to four 4K Ultra HD (3840x2160p) displays at 60Hz. Keep in mind that Maxwell is quite limited in terms of double-precision (FP64) calculations, as the design team has limited its Maxwell-based Quadro series to just 1/32 FP32. The exception to this is the Geforce GTX Titan Black, a top-tier professional grade GPU for both consumers and professionals which has driver-adjustable settings between 1/3 and 1/24 FP32 by switching to TCC mode.
The new 2016 Maxwell-based Quadro M6000 24GB model not only doubles the amount of VRAM, but also adds additional intermediate GPU clock points for smoother GPU frequency transitions. Nvidia claims this allows programs to increase and decrease in performance without interruptions to workflow. The new 24GB model also adds increased flexibility to software temperature control, allowing GPU temperature to stay below hardware slowdown threshold to prevent rendering interruptions. Lastly, Nvidia has added an “under-power” boot message to indicate if the card is not receiving enough wattage before system startup.
Thankfully, Nvidia is allowing the Quadro M6000 12GB and newer Quadro M6000 24GB to be mixed in the same system for rendering projects, but it does still depend on usage situation. For example, the M6000 12GB and M6000 24GB cannot be configured over SLI (multi-GPU) or Mosaic (multiple displays shown as a single GPU) as the cards use different frame buffer sizes. The cards will, however, run in the same system with the M5000, M4000, M2000, K2200, K620, K6000, Tesla K20 and Tesla K40.
“At Sony Pictures Imageworks, we regularly push the limits of our ability to display and interact with very complex scenes,” says Erik Strauss, Executive Director of Software Development at Sony Pictures Imageworks. “The Quadro M6000 24GB gives us a 10x performance boost with the throughput necessary to display these types of large scenes smoothly and interactively.”
“The new Nvidia Quadro M6000 24GB allows Redshift users to render scenes of stunning, Hollywood-level detail several times faster compared to any other rendering solution on the market,” said Panagiotis Zompolas, Founder and CTO of Redshift. “We’re super-excited to see what our users create using this GPU.”
Just as the Quadro M6000 12GB was introduced last spring at a $5,000 launch price, Nvidia’s newer 24GB model will launch at the same $5,000 price, bringing its predecessor down below $3,000 in some places.
Nvidia will be showcasing its workstation-grade Quadro M6000 24GB GPU next month at its 2016 GPU Technology Conference in San Jose, California. The card will be powering a majority of virtual reality (VR) partner demonstrations on the show floor, including those from Sterolabs, Lucasfilm Ltd., Redshift Rendering Technologies and Pixar Animation Studios, among others.