At this year’s AI Cockpit Innovation Experience, Intel announced its entry into the automotive, discrete GPU (dGPU) market for vehicle cockpits.
VP and general manager of Intel Automotive Jack Weast presents Intel Arc Graphics for Automotive.
With the Intel Arc A760A at the core of this solution, Intel Arc Graphics for Automotive supports up to four 4K displays and includes features such as single root input/output virtualization (SR-IOV), which enhances performance by up to 40% in virtualized environments.
Introducing Intel Arc Graphics
Intel engineered the Intel Arc A760A with 28 Xe cores and an equal number of Ray Tracing Units, which deliver advanced real-time ray tracing capabilities essential for rendering highly realistic 3D environments. Additionally, the GPU integrates 448 matrix engines for AI computations, enabling complex tasks such as deep learning inference directly within the vehicle and 448 vector engines for parallel processing to enhance the overall computational throughput.
Intel also designed the system with scalability in mind, integrating Intel’s AI-enabled SoCs to create a synergistic environment where computational power and AI capabilities are multiplied.
Intel Arc Graphics discrete GPU.
To enable substantial bandwidth for data transfer between the GPU and other system components, the A760A is equipped with 16 GB of GDDR6 memory on a 256-bit interface and operates on a PCI Express Gen4 x16 configuration. The system supports a peak performance of up to 14 TFLOPS FP32 and up to 229 TOPS for AI operations.
It offers support for OpenGL, Vulkan, and advanced media codecs like AVC, HEVC, VP9, and AV1.
Intel’s Vision for the AI Cockpit
Powered by this new hardware, Intel also discussed its vision for future AI Cockpit Experiences, designed to provide an immersive and interactive environment within the automotive sector.
Intel’s AI Cockpit leverages a combination of hardware and software optimizations. It includes support for 4K high-refresh-rate displays, fully 3D user interfaces, and multimodal interaction, which incorporates voice, gesture, touch, and eye movement recognition. Intel believes the multimodal approach will enable more accurate and intuitive interaction by processing and fusing data from different inputs.
The Intel AI Cockpit supports large language models (LLMs) that enable cognitive functionalities within the vehicle. This includes features such as voice dialogues, precise intent recognition, and on-device data processing that ensures privacy by keeping sensitive information within the vehicle. The system also supports complex AI applications like the Zhipu AI Car Assistant and the Baichuan NPC AI Companion, which deliver nuanced vehicle control, interactive manuals, and lifelike conversational experiences.
Aligning with Intel’s Strategic Objectives
Intel’s move into the automotive discrete GPU market aligns with its broader strategic goals of expanding its footprint in high-growth, data-centric industries. By entering this space, Intel hopes to diversify its portfolio and remain at the forefront of all things compute.
All images used courtesy of Intel.
