MediaTek might be the first company to confirm the use of NVIDIA’s next-gen RTX & AI GPU IP for its upcoming Dimensity SOCs for automobiles.
NVIDIA’s Next-Gen Blackwell “RTX & AI” GPU IP Could’ve Been Utilized By MediaTek For It’s Dimensity Automobile SOCs
At Computex 2023, NVIDIA announced its partnership with MediaTek to drive the next generation of AI & accelerated computing in Automobiles. This would be achieved by incorporating NVIDIA’s GPU IP in upcoming Dimensity SOCs which are responsible for the Auto Cockpit. It looks like the GPU IP used by Dimensity will be based on the next-gen “RTX & AI” architecture as confirmed by both companies during GTC 2024.
In its official PR, MediaTek specifically highlights that four automotive SOCs, Auto Cockpit CX-1, CY-1, CM-1 and CV-1, will be harnessing this new GPU IP and additionally support the NVIDIA Drive OS. The platform is expected to employ NVIDIA’s “next-gen” RTX GPU architecture, which will be utilized for AI-focused & graphics-intensive apps on MediaTek Auto Cockpit, including a graphical interface filled with specific AI capabilities to aid drivers through an LLM assistant.
To support deep-learning capabilities, the Dimensity Auto Cockpit chipsets integrate a state-of-the-art ARM v9-A system and NVIDIA’s next-gen GPU accelerated AI computing and NVIDIA RTX graphics. The AI-enabled Dimensity Auto Cockpit platform runs large language models (LLMs) in the car, allowing vehicles to support chatbots, rich content delivery to multiple displays, driver alertness detection, and other cutting-edge AI-based safety and entertainment applications. Running applications locally offers improved security, plus speed and latency benefits. To further protect user data, the Dimensity Auto Cockpit incorporates hardware-based security features with support for the latest automotive safety standards.
Dimensity Auto Cockpit takes in-cabin entertainment to the next level. It integrates an NVIDIA RTX GPU, which supports ray tracing for realistic visuals and lighting effects in games, plus AI upscaling and frame generation for fast, fluid action.
via MediaTek
Some of the features of the SOCs include:
- Designed from the ground up for the automotive market
- Flagship 3nm manufacturing process
- State-of-the-art Armv9-A CPU with enhanced performance, security, and deep learning capabilities
- NVIDIA RTX graphics, including ray tracing and DLSS3
- Powerful, hardware-accelerated NVIDIA DLA runs LLMs at the edge, providing Generative AI capabilities
- Highly integrated SoC lowers BOM costs – processors, displays, cameras, audio streams, connectivity
- Advanced HDR ISP supporting the next generation of camera sensors.
- Multiple audio DSPs for speech and audio processing
The utilization of NVIDIA’s next-gen GPU architecture will be targeted at giving users “realistic” scenario displays, including ray-tracing images to ensure proper visuals. Moreover, MediaTek has confirmed that the GPU tile would be monolithic, featuring NVIDIA’s IP, which is different from the “chiplet” design utilized by the Blackwell AI/Data Center GPUs.
Considering that this is a GPU IP being used which is contrary to using the Drive Thor platform which also employs a monolithic chip based on the same Blackwell GPU and ARM CPU architecture, it looks like this is the first time we’ve seen confirmation of a Blackwell variant for RTX-capable GPUs which gives us an idea that gaming parts will be unveiled in the second half of 2024.
This move comes after NVIDIA announced the firm’s intention to go under the custom chip segment after CEO Jensen Huang disclosed that Team Green sees an opportunity in this segment. It was previously reported that Team Green had formed a new business unit focusing on the R&D of custom chips, and while granting MediaTek a GPU IP certainly doesn’t mean that the company has done a custom chip deal with them, it does show that the firm is interested in this segment, and we might see several similar partnerships in the future as well.
News Source: HardwareLuxx