The Dimensity 9300+ offers increased clock speeds and has been designed to accelerate on-device generative AI processing, offering broader support for LLMs, and other performance enhancements over the Dimensity 9300.
“The Dimensity 9300+ will help us continue to build a rich ecosystem of generative AI applications thanks to the chip’s broad LLM support and on-device LoRA Fusion,” said JC Hsu, Corporate Senior Vice President at MediaTek. “To enhance these AI experiences, the Dimensity 9300+ offers impressive performance and enhancements to speed up LLM inference, running tokens much faster for a better user experience.”
Leveraging an All-Big-Core design with 3rd generation TSMC 4nm process, Dimensity 9300+ has one Arm Cortex-X4 core operating at up to 3.4 GHz, plus three Cortex-X4 cores and four Cortex-A720 cores.
The device also boasts increased AI processing over the previous generation thanks to MediaTek’s new NeuroPilot Speculative Decode Acceleration technology in the company’s latest generative AI engine. The APU 790 AI engine allows the chip to not only support LLMs with 1B, 7B, and 13B parameters, with scalability up to 33B, but also run LLMs quickly and efficiently.
According to MediaTek, with NeuroPilot Speculative Decode Acceleration, the Dimensity 9300+ can run LLMs with seven billion parameters at 22 tokens per second, more than 2X the rate of other mass market solutions currently on the market.
Dimensity 9300+ integrates a 2nd generation hardware raytracing engine with an Arm Immortalis-G720 GPU, giving gamers fast raytracing experiences at a smooth 60 FPS, along with console-grade global illumination effects. The chip also leverages MediaTek’s latest HyperEngine technologies to take gaming to the next level.
MediaTek Adaptive Gaming Technology (MAGT) offers increased power efficiency when enabled in popular game titles, helping to extend battery life and keep devices cool.
Additionally, the chip integrates HyperEngine’s new Network Observation System (NOS) to improve WiFi/Cellular dual-network concurrency and leverage precise network prediction technology. When enabled, MediaTek NOS offers up to 10% power savings and up to 25% cellular data savings.
With MediaTek’s Imagiq 990 ISP that offers 18-bit RAW processing, Dimensity 9300+ will enable users to take enhanced photos and videos, even in low-light environments. Imagiq 990 has a built-in AI Semantic Analysis Video Engine to support advanced AI videography features including 16 scene segmentation with real-time video capture to reduce noise, increase brightness, and enhance colours.
Additionally, the Dimensity 9300+ integrates MediaTek’s MiraVision 990, which has advanced AI depth engine technologies to enhance smartphone content.
Dimensity 9300+ is intended for a wide range of AI applications, offering:
- Support for on-device LoRA Fusion and NeuroPilot LoRA Fusion 2.0 so developers can quickly bring to market new generative AI applications with text, images and music.
- Support for the latest LLMs including 01.AI Yi-Nano, Alibaba Cloud Qwen, Baichuan AI, ERNIE-3.5-SE, Google Gemini Nano, and Meta Llama 2 and Llama 3.
- ExecuTorch Delegation for ecosystem deployed on-device inferencing.
The Dimensity 9300+ includes a 5G R16 modem designed with AI situational awareness. The modem supports 4CC-CA Sub-6GHz with up to 7Gbps downlink.
