- Mediatek unveils Dimensity 9300+, enhancing on-device AI processing at the edge.
- Dimensity 9300+ boasts accelerated clock speeds and broad support for LLMs.
- Features innovative NeuroPilot Speculative Decode Acceleration technology for superior AI performance.
- All-Big-Core design with Arm Cortex-X4 cores and 3rd generation TSMC 4nm process.
- Integrates hardware ray tracing engine and Arm Immortalis-G720 GPU for immersive gaming experiences.
- Incorporates HyperEngine technologies for optimized power efficiency and network connectivity.
- Equipped with Imagiq 990 ISP for exceptional photo and video capture, even in low-light conditions.
Main AI News:
In a bold move towards advancing generative AI processing at the edge, MediaTek Inc. introduces the Dimensity 9300+, the latest addition to its esteemed Dimensity chipset series. Boasting accelerated clock speeds and tailored functionalities, the Dimensity 9300+ promises to revolutionize on-device AI processing, offering extensive support for Large Language Models (LLMs) and a plethora of performance enhancements over its predecessor, the Dimensity 9300.
“The Dimensity 9300+ heralds a new era in the realm of generative AI applications, fortified by its robust LLM support and on-device LoRA Fusion technology,” stated JC Hsu, Corporate Senior Vice President at MediaTek. “This chipset not only amplifies AI experiences but also delivers unparalleled performance, ensuring swift LLM inference and token processing for an immersive user journey.”
Harnessing an innovative All-Big-Core architecture powered by the cutting-edge 3rd generation TSMC 4nm process, the Dimensity 9300+ flaunts a formidable configuration. With one Arm Cortex-X4 core sprinting at speeds of up to 3.4GHz, complemented by three Cortex-X4 cores and four Cortex-A720 cores, this chipset sets a new benchmark in AI processing prowess. Furthermore, the Dimensity 9300+ introduces MediaTek’s groundbreaking NeuroPilot Speculative Decode Acceleration technology, elevating generative AI capabilities to unprecedented heights. Bolstered by the formidable APU 790 AI engine, this chipset not only accommodates LLMs with 1B, 7B, and 13B parameters, scalable up to 33B, but also executes LLM tasks with remarkable efficiency. Thanks to NeuroPilot Speculative Decode Acceleration, the Dimensity 9300+ achieves a staggering processing rate of 22 tokens per second for LLMs with seven billion parameters, outperforming competitors by over twofold.
Diving deeper into its feature set, the Dimensity 9300+ integrates a state-of-the-art 2nd generation hardware ray tracing engine alongside an Arm Immortalis-G720 GPU, ensuring gamers indulge in captivating ray tracing experiences at a seamless 60 FPS, coupled with console-grade global illumination effects. Leveraging MediaTek’s cutting-edge HyperEngine technologies, this chipset raises the bar for gaming performance. MediaTek Adaptive Gaming Technology (MAGT) optimizes power consumption during gameplay, extending battery life and preserving device thermals. Moreover, the incorporation of HyperEngine’s new Network Observation System (NOS) enhances dual-network concurrency, bolstering WiFi/Cellular connectivity while leveraging predictive network technology to conserve power and data usage. Enabled NOS functionality yields up to 10% power savings and a remarkable 25% reduction in cellular data consumption.
Equipped with MediaTek’s Imagiq 990 ISP, boasting 18-bit RAW processing capabilities, the Dimensity 9300+ empowers users to capture stunning photos and videos, even in low-light conditions. The Imagiq 990’s AI Semantic Analysis Video Engine facilitates advanced AI videography features, including real-time video capture with 16 scene segmentation, effectively reducing noise, enhancing brightness, and enriching colors. Additionally, the Dimensity 9300+ integrates MediaTek’s MiraVision 990, featuring advanced AI depth engine technologies to elevate smartphone content to new heights of visual excellence.
Conclusion:
MediaTek’s Dimensity 9300+ chipset marks a significant advancement in edge computing capabilities, particularly in the realm of generative AI processing. With accelerated clock speeds, broader LLM support, and innovative technologies such as NeuroPilot Speculative Decode Acceleration, this chipset sets a new standard for on-device AI performance. Moreover, its integration of advanced gaming features, network optimization, and imaging capabilities further solidifies its position as a leader in the market, catering to the evolving demands of consumers and developers alike.
