Six New Pascal-Based Quadro Cards Released, One With HBM2

Nvidia is rolling out six new cards in its workstation-oriented Quadro line at SolidWorks World this week. The company's GP100, P4000, P2000, P1000, P600, and P400 fill out a family of Pascal-based products previously populated by the high-end P6000 and P5000. Whereas AMD's Polaris-based Radeon Pro WX 7100, WX 5100, and WX 4100 cards address the entry-level and mid-range segments, Nvidia's offerings now span a much more diverse range.

Taken in its entirety, the family of Pascal-powered Quadros render older Kepler-based solutions obsolete, at the same time extending Nvidia's performance advantage in the professional space. The following slide illustrates how Nvidia sees the potential of its Quadro P series and where the company plans to sell it.

As we've seen on the desktop, Pascal raises the bar compared to the previous generation in terms of both absolute performance and efficiency. The chart below conveys some of the relative measurements, and they're pretty compelling. Although the real-world applicability of SPECviewperf 12.1 is up for debate, it does provide a solid comparison point.

Of course, these Pascal-based cards don't just improve performance; they also add new features, many of which we covered in our Nvidia GeForce GTX 1080 Pascal Review. Whether you're interested in Simultaneous Multi-Projection (SMP), VRWorks, Iray VR, Quadro Scalable Visualization Solutions (SVS), or the soon-to-be-available Quadro Sync II, each model offers a lot more functionality than the Quadros with Kepler and Maxwell GPUs on board.

Quadro GP100: A New High-End Monster

Nvidia's flagship carves out a position at the top of the hierarchy. It comes equipped with 16GB of HBM2 and should easily trounce the Quadro P6000. With great power, however, comes great responsibility, and in this case, that responsibility means paying a premium for unmatched  performance.

But where does such an expensive card make sense? Simultaneous visualization and compute workloads? No problem. Big data sets and real-time interaction with photorealistic content? Done. Of course, there's also the buzzword du jour: deep learning.

The Quadro GP100 can also be coupled with a second card using NVLink, which facilitates point-to-point communication between GPUs, or between a single/multi-GPU setup and a CPU at up to 80 GB/s. In the compute space, this high-bandwidth, energy-efficient path is second to none.

How Does Pascal Actually Perform?

We plan to test the new Quadros in as much depth as we recently did in our AMD Radeon Pro WX 7100 Review. Additionally, we will test and compare both the current Quadro cards and their AMD Radeon Pro counterparts directly using an Intel Xeon E5-2640 v4 CPU, 64GB of DDR4-2400 ECC memory, and MSI's X99A workstation motherboard. 

Expect the roundup to include Quadro cards ranging from the P6000 through the P1000, and all three Radeon Pros (the WX 7100, 5100, and 4100) in a range of real-world applications. Just be patient as we wrap it up; a half-baked story based on synthetic metrics does more harm than good, and we're working to serve up the most accurate analysis possible. That takes some time to get right.