NVIDIA has announced its brand new graphics card, the Quadro K6000, featuring 12GB of memory and 2,880 cores, while at the same time fully dispelling the notion that the company isn’t interested in the media and entertainment market.
The 12GB of on-board GDDR5 memory in the new flagship card is a significant increase, doubling the K5000 which was the previous top of the line Kepler card. Limited graphics card memory capacity has always been a critical issue in CGI world, and while this doesn’t solve the problem, providing access to twice the memory currently available is undeniably a significant improvement for many applications.
The company says the new card was designed based on what its clients in several different industries have been asking for from a GPU. “Our customers are trying to do one core thing: look at more data at one time and be able to interact with larger datasets,” NVIDIA Industry Executive, Media and Entertainment, Greg Estes told fxguide. “You could be an artist working in Maya, or a guy at Nissan trying to design a car, or a geophysicist looking at an oil field.”
The increase in memory will certainly improve the experience for 3D artists interactively working on large scale scenes, as fewer compute cycles will be spent getting data on and off the cards. It’s still short of the 48GB to 64GB numbers that is needed to approach doing final quality high res renders on the GPU. That being said, NVIDIA has a clear roadmap to fully address access to more memory on their cards in the future with their Volta generation, as we reported in our GPU Tech conference report earlier this year.
Estes says the anticipated increased performance from the Quadro K6000 is part of a push by NVIDIA to respond to new business models in, say, the visual effects industry. “Costs have to go down and you have to be able to produce stuff faster,” he notes, “but not by killing the artists or making them work for less. There has to be technology or forcing function that allows them to get higher quality content done faster and more economically.”
We’ve heard murmurs from artists about NVIDIA abandoning the vfx and entertainment industry, a conclusion leapt to for a variety of possible reasons. It could due to the fact that there hasn’t previously been a Kepler generation replacement for the Quadro 6000. Or the NVIDIA’s lack of inclusion in the upcoming Mac Pro system. Or the news that NVIDIA has ramped down SDI daughter cards for its boards.
The Quadro K6000 certainly looks squarely aimed at those in the visual effects and related fields. In fact, Estes says the Media and Entertainment (M&E) market makes up 25% of NVIDIA’s business. It doesn’t seem reasonable for a company to give up a market that makes up a quarter of their business.
Regarding the Mac Pro, Estes says NVIDIA is “very robustly in the mac market, and disappointed that we didn’t get the design with the new Mac Pro.” It is true that NVIDIA graphics are used in other Mac product lines, from the MacBook Pro to the iMac — and these are the lines that are selling for Apple.
As far as the lack of an NVIDIA SDI solution, “we’ve ramped down making our own SDI cards and invested in the software instead, so that the customers could use anybody’s SDI card,” says Estes. This could been problematic for some applications, such as Autodesk’s Flame, which rely on this daughter card for broadcast monitor support.
But for others, Estes is referring to NVIDIA’s GPUDirect for Video tech, which was first announced about two years ago. The API enables third party vendors like Blackmagic Design, AJA, Matrox, and others to get data on and off the GPU at sub-frame transfer times without introducing too much delay or latency.
“If we’re going to make it so you can have a choice of six different vendors, all of which cost less than what we do and make sure it all works with all software – how does that mean we’re abandoning the market?,” he adds. “NVIDIA is super-committed to media and entertainment – and the Quadro K6000 is a key example of that.”
You can find out more about the Quadro K6000 at NVIDIA’s SIGGRAPH 2013 booth (#803) this week. And several industry experts will also be at NVIDIA’s booth, including reps from Weta Digital, Dneg, Pixar, Zoic Studios, Adobe, Autodesk and Chaos Group.
No pricing has been announced, but we can likely expect it it in line with the other product offerings in the Quadro line. Some of the main specs:
- GPU Memory: 12 GB GDDR5
- Memory Interface: 384-bit
- Memory Bandwidth: 288 GB/s
- CUDA Cores: 2880
- System Interface: PCI Express 3.0 x16
- Max Power Consumption: 225 W
- Thermal Solution: Ultra-quiet active fansink
- Form Factor: 4.4”H × 10.5”L, Dual Slot, Full Height
- Display Connectors: DVI-I DL + DVI-D DL + 2x DP1.2 + Stereo
- Max Simultaneous Displays: 4
- Max DP 1.2 Resolution: 3840 × 2160 at 60 Hz
- Max DVI DL Resolution: 2560 × 1600 at 60 Hz
- Max DVI SL Resolution: 1920 × 1200 at 60 Hz
- Max VGA Resolution: 2048 × 1536 at 85 Hz
- Graphics APIs: Shader Model 5.0, OpenGL 4.3, DirectX 11
- Compute APIs: CUDA, DirectCompute, OpenCL
Here’s the full press release from NVIDIA:
=====
NVIDIA Unveils New Flagship GPU for Visual Computing
Quadro K6000 Tackles Most Complex Visual Computing Problems From Tapping Vast New Energy Sources to Animating Monster Fur
ANAHEIM, Calif. SIGGRAPH July 23, 2013NVIDIA today unveiled the visual computing new flagship technology the NVIDIA® Quadro® K6000 GPU, the fastest and most capable GPU ever built.
NVIDIA today also launched a new line of professional graphics GPUs for mobile workstations, delivering the highest levels of performance and graphics memory ever available on mobile platforms.
The Quadro K6000 GPU delivers five-times higher compute performance and nearly double the graphics capability of its predecessor, the NVIDIA Quadro 6000 GPU largest and fastest graphics memory.
Combining breakthrough performance and advanced capabilities in a power-efficient design, the Quadro K6000 GPU enables leading organizations such as Pixar, Nissan, Apache Corporation and WSI division to tackle visualization and analysis workloads of unprecedented size and scope.
The Kepler features are key to our next generation of real-time lighting and geometry handling. We were thrilled to get an early look at the K6000. The added memory and other features allow our artists to see much more of the final scene in a real-time, interactive form, and allow many more artis
-Guido Quaroni, Pixar vice president of Software R&D
12 GB of memory, I am now able to load nearly complete vehicle models into RTT Deltagen and have stunning photorealism almost instantly. Instead of spending significant time simplifying the models to fit into previous hardware, we can now spend more time reviewing and iterating designs up front which helps avoid costly
-Dennis Malone, associate engineer, Nissan North America
Energy Exploration Apache
The Quadro K5000, the Quadro K6000 tripled the performance when tion. With jobs running in mere minutes, we can run more simulations and get better insight into where to drill. In this business, drilling in the wrong place is a multi-million dollar mistake, and the Quadro K6000 gives us the edge to make better decisions.
-Klaas Koster, manager, seismic interpretation, Apache Corporation
Unprecedented Performance
The Quadro K6000 GPU is based s fastest, most efficient GPU architecture. Key performance features and capabilities include:
- 12GB ultra-fast GDDR5 graphics memory lets designers and animators model and render characters and scenes at unprecedented scale, complexity and richness
- 2,880 streaming multiprocessor (SMX) cores deliver faster visualization and compute horsepower than previous-generation products
- Supports four simultaneous displays and up to 4k resolution with DisplayPort 1.2
Ultra-low latency video I/O and support for large-scale visualizations
The NVIDIA Quadro K6000 GPU is the highest performance, most capable GPU ever created for the professional graphics marketsenior vice president, Professional Solutions Group It will significantly change the game for animators, digital designers and
New Mobile Workstation GPUs
NVIDIA today also revealed a new flagship professional graphics GPU for workstation notebooks, the NVIDIA Quadro K5100M GPU. Delivering the highest levels of performance and graphics memory available on notebook platforms, the Quadro K5100M anchors a new line of
NVIDIA Unveils New Flagship Professional Graphics Processor Page 3
workstation notebook graphics that includes the Quadro K4100M, K3100M, K2100M, K1100M, K610M, and K510M GPUs.
Quadro GPUs are designed, built and tested by NVIDIA to provide the superb reliability, compatibility and dependability that professionals require. They are certified and recommended by more than 150 leading software application providers worldwide.
Availability
The NVIDIA Quadro K6000 will be available beginning this fall from Dell, Lenovo and other major workstation providers; from systems integrators, including BOXX Technologies and Supermicro; and from authorized distribution partners, including PNY Technologies in North America and Europe, ELSA and Ryoyo in Japan, and Leadtek in Asia Pacific.
The new Quadro mobile workstation graphics product line will also be available beginning this fall from major mobile workstation OEMs.
For more information about Quadro GPUs, graphics/visualization and market news, see the
NVIDIA Quadro website. And follow us on YouTube and on Twitter @NVIDIAQuadro.
Very nice! I will say though, the GTX Titan can still nearly match the Quadro K6000 in performance aside from a heavy file load into memory. Titan CUDA cores = 2,688, Memory Bandwidth = 288.4 GB/sec, and GPU Memory: 6 GB GDDR5. This rending benchmark from BOXX shows the Titan besting the Quadro K6000 by 1 Ms/Sec http://www.youtube.com/watch?v=kAjvnGMAtz8 Titan’s potential negative is that it’s still only outputting 8-bit color afaik, so it might not be the best card for a high-end color correction/grading workflow.
Update: You’ll notice now that the YouTube video I posted previously no longer works. That’s because NVIDIA complained to BOXX and made them take it down… interesting uncomfortableness huh? 🙂
I am sure the video posted was using alpha drivers for the K6000. Not a fair comparison.
There is nothing that can touch the K6000. 12 gigs of ram let’s you do amazing things. It’s the most important graphics card to come to market for a long time. I can’t stress this enough.
Re: “There is nothing that can touch the K6000.” – Really? Where are your benchmarks to back-up that emotional statement? There’s no doubt more RAM will help, but in what special 3D scenes or situations does it really help? And how often are those special situations going to occur? What percentage of the time will 12 GB really be useful? And if that percentage is very small, then is all that extra RAM truly worth the very high price that it carries? Maybe a 6 GB Titan can cover 99% of most scenes for most artists..? I am interested in facts and figures, not emotional statements.
For the type of work Scott does, I imagine. Here’s an example:
http://www.fxguide.com/fxguidetv/fxguidetv-165-scott-metzger-on-mari-and-hdr/
Scott really pushes Mari and other apps with the work he does. Good news is we’ll actually be featuring another of his projects in an upcoming fxguidetv episode we recently filmed.
Well the 12gb of Ram is really useful when painting large environments. Mari will go into texture bias mode depending on ram/how much you are painting. Mari will then stream the textures fitting what it can allowing room for paint operations/baking. Having virtual textures set to 16K in Mari allows extremely high resolution preview while painting. You can also project huge images from the paint buffer. I was hitting around 10-11gigs of video ram on average while projecting 16bit half float exrs.
I was using a K5000 prior and nothing came close to the K6000. I was able to load up 100 million polygons into Maya’s Viewport 2.0 which is really important to processing scan data from lidar as well.