NVIDIA announced their new flagship Quadro, the M6000. Like the Titan X which was introduced earlier this week, the new card contains a Maxwell generation MG200 core.
The on-paper specs of the card are identical to that of the Titan X and this is the first time in recent memory that a matching Quadro card has come out so quickly after a GeForce card, which is a big and welcome change.
As far as performance, there should be a significant speed boost compared to the previous generation K6000 cards. We haven't seen any actual benchmarks at this point in time, but in our briefing, NVIDIA's Greg Estes mentioned that the LucasFilm team saw significant increases and this was echoed by Luz Latta, Principal Engineer at Lucasfilm. "With the NVIDIA Quadro M6000 GPU, we saw overall gains of 55% in heavy a compute and memory access ray-tracing application using layered shadow maps," says Latta.
A key factor for the vfx industry is the concept of "unified virtual memory", which is part of the Maxwell architecture. This makes it possible for the GPU operations to read and see the CPU memory and the CPU to see the GPU memory. This is very different from before, loading and offloading of data and textures was much more cumbersome. There is still a performance hit when moving to the card, but the management is much easier. The next generation of cards, Pascal, will address this memory issue even further with 3D memory on the graphics card.
One area which sees a step back in performance is double-precision calculations. FP64 performance is considerably lower than the previous generation, due to the fact that NVIDIA has dedicated more processing power to graphics applications. For most artists it is a bit of an edge case and won't have a hit in most vfx applications, but it will have an impact upon simulations and other highly complex mathematical calculations.
Pricing has not been announced at this point in time. With the Titan X coming in at $999 USD, it will be interesting to see how much the premium for the Quadro flagship will be.
NVIDIA announced a new Quadro VCA today, which effectively contains eight of the new M6000. Technically, they are not exactly the same cards (due to BIOS/software changes), but the performance should be virtually identical.
Linux support is a relatively recent improvement in the VCA software that will pay dividends for the vfx industry. Before this, many facilities around the world were still using the VCA to virtualize desktops on artist monitors. But with so many facilities using linux, this could speed the adoption.
iRay 2015 Announced
NVIDIA's iRay physically-based raytracer sees significant use in the design and visualization industry (architecture, automotive, etc), in many ways due to its extremely high quality texture rendering. Its use is certain to expand with the announcement that iRay will be available for more applications, including Autodesk Maya & Revit, Rhino, Daz 3D, and even MAXON's Cinema 4D later this year.
This means that they are providing access to the Iray Material Definition Language (MDL) across applications, so an artist could use a material in Maya and have it match exactly in Cinema 4D. NVIDIA also plans to provide additional and new iRay features which have been part of the software, but not been previously accessible to users.
We've been a free service since 1999 and now rely on the generous contributions of readers like you. If you'd like to help support our work, please join the hundreds of others and become an fxinsider member.