Jigsaw24 Benchmarks: NVIDIA’s Quadro K5000 vs Quadro 4000

Now that the Quadro K5000 for Mac has finally shipped, we wanted put it through its paces. Our friends at NVIDIA kindly agreed to lend us one to test, so we broke out our Mac Pro, upgraded it to OS X 10.8.3 to get the latest GPU drivers (you’ll need to do this too if you want to use the card) and cracked on.

The test

We decided to test the K5000 against the Quadro 4000, as these two cards represent the only professional grade NVIDIA GPUs available for the Mac Pro, so we thought it would be a nice illustration of how the GPU power available to the Mac Pro has progressed. We were hoping that the K5000 would be faster across the board, as on paper it’s a far more powerful card.

We chose a range of tests designed to represent a cross-section of the GPU-centric media and entertainment workflows the K5000 has the potential to accelerate. All of our benchmarks were run on a clean copy of OS X 10.8.3 and repeated three times – we’re quoting the average score below.

Davinci Resolve 9

Firstly we tested Blackmagic Design DaVinci Resolve 9.1.1. Resolve is a colour grading application that makes heavy use of CUDA acceleration for both transcoding and colour grading.  Our (relatively simple) test involved transcoding a 2K DPX sequence into a 1920 x 1080 ProRes 422 Proxy file. So as not to cause a bottleneck that would skew the cards’ performance downward, the files were read/written to very fast storage.

Here the Quadro 4000 averaged 37fps transcoding speed and the K5000 averaged 53fps, which means that the K5000 offers roughly 40% more performance than the Quadro 4000 for this workflow. As almost all of Resolve is CUDA accelerated, this performance gain should translate to an overall increase in speed across DaVinci Resolve.

Adobe After Effects CS6

The next test used Adobe After Effects CS6. AE CS6 includes a new ray traced render engine, which is fully CUDA accelerated. This render engine allows very realistic rendering of things like shadows, reflections and glossy materials, and we planned to time how long it took each card to render a test scene setup.

The K5000 managed the render in 10 minutes and 5 seconds with the Quadro 4000 clocking 18 minutes and 3 seconds, making the K5000 almost twice as fast as the Quadro 4000 – again, a great upgrade for this workflow.

OpenGL benchmarks

Finally we wanted to gauge the general OpenGL performance of the cards to get an idea of how they would perform in 3D Applications such as Maxon CINEMA 4D and Autodesk Maya.

We used a benchmark called FurMark, which tests a card’s ability to draw complex 3D images on to the screen. Higher performance in this benchmark indicates that the card will be able to handle more complex geometry inside the viewport in apps like Maya or C4D, meaning a smoother editing experience with complex scenes.

The Quadro 4000 managed 24 frames per second and the K5000 managed 54 FPS. (To give some perspective we also ran this test on the Radeon 5770 that shipped with our Mac Pro and it scored 31 FPS.) Again, this indicates that the K5000 is over twice as powerful as the Quadro 4000 when it comes to manipulating geometry in the viewport of a 3D application.

The conclusion

Much as we love it, the Quadro 4000 is getting a little long in the tooth and the K5000 looks to be a solid upgrade performance-wise. The Quadro K5000 is in fact the most powerful officially supported CUDA GPU available for a Mac Pro – and as CUDA will not run on the Radeon cards that most Mac Pros ship with, it’s an essential upgrade for anyone who needs CUDA acceleration in their application.

The K5000 also offers significantly better OpenGL performance than the Radeon cards that ship with Mac Pros as standard, so is a solid upgrade for anyone doing complex 3D modelling or animation.

Buy your NVIDIA K5000 for Mac now

Want to know more? Give us a call on 03332 409 306 or email 3D@Jigsaw24.com. For all the latest news and reviews, follow @Jigsaw24Video on Twitter or ‘Like’ us on Facebook

Ben
Ben
Call us: 03332 409 306

Leave a Reply

Your email address will not be published. Required fields are marked *