This site may earn affiliate commissions from the links on this page. Terms of apply.

OctaneRender-Feature

One of the major differences betwixt AMD and Nvidia is their share of the professional graphics market. Nvidia dominates this space and its profit margins, and while AMD has had some loftier contour wins with Apple, it hasn't cut deeply into Nvidia's market place share. Role of the reason Nvidia has a lock on both workstation and loftier operation calculating is CUDA, its programming language for GPU compute. Now one company, Otoy, is claiming to have broken that lock.

Otoy is the owner and programmer of Octane Render, a existent-time unbiased rendering engine that supports 3D rendering software suites like 3ds Max, Maya, Cinema4D, and Lightwave. It'south also available as its own standalone software suite. It was the showtime unbiased rendering suite to back up GPU-only rendering and a high-contour early win for Nvidia's CUDA — which is part of why it's surprising to see the company branching out to support other architectures in this fashion.

Didn't AMD just denote this?

There's some timing oddities here that I'grand not sure how to explain. Final year, AMD announced its Boltzman Initiative. Part of that initiative is a software layer that allows AMD GPUs to execute CUDA code through the use of a compatibility layer.

Open-Source-Compute1

AMD's Boltzmann Initiative

Here'south how VentureBeat describes Otoy's new compatibility layer: "In a nutshell, Otoy reverse-engineered Nvidia's general purpose graphics processing unit of measurement (GPGPU) software, known as CUDA, to run on non-Nvidia hardware. That means that programs written in the CUDA language are no longer exclusive to Nvidia graphics chips."

Co-ordinate to Otoy's CEO, Jules Urbach, the bespeak of developing this CUDA translation layer is so that the company's high-finish Octane Return software can run as hands on AMD GPUs as their Intel counterparts. "Nosotros accept been able to practice this without changing a line of CUDA code, and it runs on AMD fries," Urbach said. "You tin now plan once and have CUDA everywhere. AMD has never really been able to provide an alternative."

Octane Render

Octane Render in action. Image from FXGuide

AMD's Boltzmann Initiative would seem to provide the alternative that Urbach is referencing, and it appears to accomplish the aforementioned goal. It's not clear how the two programs differ from each other, though Otoy does mention wanting to run software on a wider variety of platforms, operating systems, and technologies. AMD'south Boltzmann Initiative, of class, is designed solely for AMD'south own GPUs.

As for performance, Urbach states that "It runs on the other cards at the same speed every bit it runs on Nvidia cards." But over again, that'southward something AMD has unsaid about its ain Boltzmann Initiative — when we asked the company how AMD GPUs compared to NV cards running CUDA, the RTG division implied that unless the CUDA code had been mitt-optimized for a specific CUDA architecture, it should run as chop-chop on AMD hardware as on an Nvidia counterpart GPU.

Urbach claims that the long-term goal is to allow CUDA to target Vulkan, DirectX, and OpenGL (along with Android, PS4, and WebGL three), and that Otoy wants to exist able to run CUDA applications on platforms like iOS, where Apple's Metallic is the ascendant low-overhead API.

Supposedly Otoy is working on turning Octane Render into a plugin that the UE4 engine can use, but Octane Render isn't used for real-time rendering. Adapting a version of it to work inside a game engine would be extremely challenging. Information technology's non at all clear why Otoy would want to translate native CUDA into many of the APIs that Urbach lists — no games that I'thousand aware of leverage OpenCL or CUDA for any kind of tasks, and neither AMD nor Nvidia have talked about using either language for this purpose.

We've reached out to Octane Render and volition update this story if nosotros hear more details. At the very least, it looks similar AMD'south push to convert CUDA code into something that can run on more GPUs has caught the attending and imagination of other vendors.