0
NVIDIA's GPU IP Drives into MediaTek's Dimension Auto SoCs

NVIDIA's GPU IP Drives into MediaTek's Dimension Auto SoCs

MediaTek this week introduced a new lineup of Dimensity AutoCockpit System-on-Chips, covering the entire market spectrum from entry-level to premium. And while automotive chip announcements aren't usually the most exciting of things, this one is going to be an exception to the rule because of the graphics IP MediaTek is tapping for chips: NVIDIA's. This means that the upcoming Dimensity AutoCockpit chips will be the first chips released by a third-party (non-NVIDIA) vendor that will be based on NVIDIA's GeForce graphics technology.

of NVIDIA First attempt to license your GPU IP to third parties The year is 2013, when the company proposed to license its Kepler GPU IP and thus compete with Arm and Imagination Technologies. An effort that, at the time, fell flat on his face. But a decade later and a fresh attempt to license some of NVIDIA's IP, and it looks like NVIDIA has finally succeeded. Overall, MediaTek's new Dimensity Auto Cockpit System-on-chips will rely on NVIDIA's GPU IP, Drive OS, and CUDA, marking a landmark development for both companies.

Family of Mediatech Next-generation Dimensity Auto-Cockpit processors There are four separate system-on-chips, including the CX-1 for range-topping vehicles, CY-1, CM-1, and CV-1 for entry-level cars. These are highly integrated SoCs that pack an Armv9-A-based general-purpose CPU core as well as NVIDIA's next-generation graphics processing unit IP. NVIDIA's GPU can run AI workloads for power infotainment systems as well as support for the GPU IP driver, as it fully supports graphics technologies like real-time ray tracing and DLSS 3 image upscaling.

Dimensity auto-cockpit processors are monolithic SoCs with built-in multi-camera HDR ISP, say Hardware Lux. This ISP supports front-facing, in-cabin, and bird's-eye view cameras for a variety of security applications. Additionally, these processors include an audio DSP that supports various audio assistants.

The announcement from MediaTek doesn't reveal which generation of NVIDIA's graphics IP they're adopting — only that it's a “next-generation” design. Given the certification requirements involved, automotive SoC announcements tend to be more conservative, so it remains to be seen how this graphics IP will be “next-gen” from the current generation Ada Lovelace architecture.

The new MediaTek SoCs will be fully supported by NVIDIA's Drive OS, which is already widely used by automakers. This will allow automakers to unify their software stacks and use a single set of software for all their cars powered by MediaTek's Dimensity. Additionally, since NVIDIA's Drive OS fully supports CUDA, TensorRT, and Nsight, MediaTek's Dimensity SoCs will be able to take advantage of AI applications developed for the green company's platform.

“Generative AI is transforming the automotive industry just as it has in the mobile market with more personalized and intuitive computing experiences,” said Jerry Yu, corporate senior vice president and general manager of MediaTek's CCM business group. It has created a revolution.” “The Dimensity Auto Cockpit portfolio will usher in a new wave of AI-powered in-vehicle entertainment, and our unified hardware and software platform will make it easy for automakers to scale AI capabilities across their entire lineup.” Is.”

Of course, licensing the graphics IP and platform IP to a third party is a milestone for NVIDIA in general, and for its automotive efforts in particular. Leveraging DriveOS and CUDA beyond NVIDIA's own hardware platform is a big deal for a business unit that NVIDIA has long considered poised for significant growth, but faces tough competition and slowdowns thanks to conservative automakers. have faced. Meanwhile, it remains to be seen how MediaTek's new Dimensity AutoCockpit processors will stack up against NVIDIA's previously announced ones. Thor SOC and affiliates DriveThor Platformwhich integrates a Blackwell-based GPU that delivers 800 TFLOPS of 8-bit floating-point AI performance.

About the Author

Leave a Reply