Quadro P6000 And P5000 Review: NVIDIA's Most Powerful Pascal Graphics Cards
Evaluating NVIDIA's Powerful Quadro P6000 & P5000 Workstation Graphics Cards
NVIDIA’s Pascal architecture has been wildly successful in the consumer space. The various GPUs that power the GeForce GTX 10 series are all highly competitive at their respective price points, and the higher-end variants are currently unmatched by any single competing GPU. That may change when AMD launches GPUs based on its next-generation Vega architecture, but that won’t happen for a few more months, so cards like the GeForce GTX 1080 and TITAN X will continue to dominate benchmark charts at least for a little while longer.
NVIDIA has since retooled Pascal for the professional workstation market as well, with products that make even the GeForce GTX 1080 and TITAN X look quaint in comparison. We're speaking of the beastly Quadro P6000 and Quadro P5000 -- Pascal powered behemoths, packing up to 24GB of GDDR5X memory and GPUs that are more capable than their consumer-targeted counterparts. Though it is built around the same GP102 GPU, the Quadro P6000 is particularly interesting, because it is outfitted with a fully-functional Pascal GPU with all of its SMs enabled, which results in 3,840 active cores, versus 3,584 on the TITAN X. The P5000 has the same GP104 GPU as the GTX 1080, but packs in twice the amount of memory – 8GB vs 16GB.
NVIDIA's Quadro P6000’s and P5000’s main features and specifications are outlined in the table below. Take a look at what they offer at a high level, and then we’ll tunnel in a deeper on the pages ahead...
The NVIDIA Quadro P6000
The Quadro P6000 and P5000 are the most powerful desktop workstation graphics cards based on the GP102 and GP104 GPUs that exist today. The P6000 offers up to 12 TFLOPs of single-precision compute performance, while the P5000 comes in just a shade under 9 TFLOPS. Since both cards leverage Pascal, they support all of the bleeding-edge features of the architecture, like Simultaneous Multi-Projection, which allows the GPU to render to multiple viewports without having to resend the geometry from the application for each different perspective in a scene. They also support Lens Matched Shading, all of the other features that are part of NVIDIA’s VR Works, and advanced memory compression. We’ve covered the inner-workings of Pascal already, however, so we won’t dig into all of these things again here. For more details, we’d suggest checking out pages 2 and 3 of our GTX 1080 launch coverage.
The Quadro P6000 looks very similar to the previous-gen Quadro M6000, which was based on Maxwell. The cards feature the same black and bright-green coolers, dark PCB, and similar connector placement all around. The P6000 is a much more capable beast, though. As previously mentioned, the card is packing 3,840 CUDA cores, with 24GB of 9Gbps GDDR5X memory, linked to the GPU via a 384-bit interface. The GPU has a base clock of 1,417MHz and boost clock of 1,530MHz. At those frequencies, compute performance tops out at about 12 TLFOPs, with 432GB/s of peak memory bandwidth. The Quadro P6000 is also rated for max power consumption of 250W, which is right in-line with the previous-gen M6000, and requires a single 8-pin supplemental power feed.
The Quadro P5000 looks somewhat similar, but it has a different cooler assembly. It is also signature Quadro black and green, but the P5000 is a bit lighter weight. It is built around the GP104 GPU and has 2,560 active CUDA cores, with 16GB of 9Gbps GDDR5X memory, linked to the GPU via a 256-bit interface. The GPU can boost up to 1,733MHz, and as such the card offers up to 8.9 TFLOPs of compute performance with up to 288GB/s of memory bandwidth. Max power consumption is down to 180W on the P5000, but it too requires a single 8-pin supplemental power connector.
Both the Quadro P5000 and P6000 coolers are vented at the top (near the case bracket) and both feature SLI, SYNC, and Stereo connectors along the top edge. Overall dimensions for the cards are similar as well, at about 10.5” long and 4.4” high, with dual-slot coolers.
The display outputs on the cards consist of a quartet of full-sized DisplayPorts, and a dual-link DVI output. The DisplayPorts are 1.2 certified and DP 1.3/1.4 ready, which enables support for 4K displays at 120Hz, 5K displays at 60Hz, and 8K displays at 60Hz (using two cables and multi-stream transport). Up to four display outputs can be used simultaneously for multi-monitor or VR setups.
We should also mention, however, that these cards support NVIDIA Quadro Scalable Visual Solutions (SVS) and work with NVIDIA’s video synchronization modules. With add-ins like the upcoming Quadro Sync II (and multiple Quadro cards) users have the ability to drive up to 32 screens driven from a single system. When combined with the upcoming Quadro Sync II card, up to four P5000 or P6000 GPUs can be synchronized inside a single system to drive a 16-display configuration. And with two Quadro Sync II boards installed, a single system can connect up to eight P5000 or P6000 boards to synchronize 32 separate displays. That's a lot of rendering workspace.
Let's fire these cards up...
NVIDIA has since retooled Pascal for the professional workstation market as well, with products that make even the GeForce GTX 1080 and TITAN X look quaint in comparison. We're speaking of the beastly Quadro P6000 and Quadro P5000 -- Pascal powered behemoths, packing up to 24GB of GDDR5X memory and GPUs that are more capable than their consumer-targeted counterparts. Though it is built around the same GP102 GPU, the Quadro P6000 is particularly interesting, because it is outfitted with a fully-functional Pascal GPU with all of its SMs enabled, which results in 3,840 active cores, versus 3,584 on the TITAN X. The P5000 has the same GP104 GPU as the GTX 1080, but packs in twice the amount of memory – 8GB vs 16GB.
NVIDIA's Quadro P6000’s and P5000’s main features and specifications are outlined in the table below. Take a look at what they offer at a high level, and then we’ll tunnel in a deeper on the pages ahead...
The NVIDIA Quadro P6000
 |
|
| Quadro P6000 | Quadro P5000 | |
| Architecture | Pascal | Pascal |
| CUDA Parallel Processing cores | 3840 | 2560 |
| Peak Single Precision Performance | Up to 12 TFLOPs | Up to 8.9 TFLOPs |
| Frame Buffer Memory | 24 GB GDDR5X | 16GB GDDR5X |
| ECC Memory | Yes | Yes |
| Memory Interface | 384-bit | 256-bit |
| Memory Bandwidth | 432 GB/s | 288 GB/s |
| Max Power Consumption | 250 W | 180 W |
| Graphics Bus | PCI Express 3.0 x16 | PCI Express 3.0 x16 |
| Display Connectors | DP 1.2 (4), DVI-I (1), Optional Stereo (1) | DP 1.2 (4), DVI-I (1), Optional Stereo (1) |
| Maximum Supported Displays | 4 | 4 |
| Maximum Displays @ 4K 60Hz | 4 | 4 |
| Form Factor | 4.4” H x 10.5” L Dual Slot | 4.4” H x 10.5” L Dual Slot |
| Thermal Solution | Active | Active |
| NVIDIA 3D Vision and 3D Vision Pro | Support via 3 pin mini DIN | Support via 3 pin mini DIN |
| Video Sync Module | Quadro Sync II | NVIDIA Sync II |
| GPU Direct for Video | Yes | Yes |
The Quadro P6000 and P5000 are the most powerful desktop workstation graphics cards based on the GP102 and GP104 GPUs that exist today. The P6000 offers up to 12 TFLOPs of single-precision compute performance, while the P5000 comes in just a shade under 9 TFLOPS. Since both cards leverage Pascal, they support all of the bleeding-edge features of the architecture, like Simultaneous Multi-Projection, which allows the GPU to render to multiple viewports without having to resend the geometry from the application for each different perspective in a scene. They also support Lens Matched Shading, all of the other features that are part of NVIDIA’s VR Works, and advanced memory compression. We’ve covered the inner-workings of Pascal already, however, so we won’t dig into all of these things again here. For more details, we’d suggest checking out pages 2 and 3 of our GTX 1080 launch coverage.
The display outputs on the cards consist of a quartet of full-sized DisplayPorts, and a dual-link DVI output. The DisplayPorts are 1.2 certified and DP 1.3/1.4 ready, which enables support for 4K displays at 120Hz, 5K displays at 60Hz, and 8K displays at 60Hz (using two cables and multi-stream transport). Up to four display outputs can be used simultaneously for multi-monitor or VR setups.
Let's fire these cards up...
