Qualcomm has launched two new Snapdragon processors built on TSMC’s advanced 2 nm process, marking a major leap in on‑device AI. The Snapdragon 8 Elite Gen 6 and the more powerful Snapdragon 8 Elite Extreme Gen 6 both feature Qualcomm’s custom Oryon CPU architecture, breaking the 5 GHz barrier and introducing new levels of performance and efficiency.
The Extreme variant stands out by enabling smartphones to run a 30‑billion‑parameter mixture‑of‑experts (MoE) model locally. This capability allows advanced generative AI tasks—such as multimodal reasoning, image creation, and natural language processing—to be executed directly on the device, reducing latency, enhancing privacy, and cutting reliance on cloud infrastructure.
Key Details of the New Snapdragon Chips
Snapdragon 8 Elite Extreme Gen 6
CPU: Oryon cluster with 2 Prime cores clocked at up to 5.0 GHz and 6 Performance cores up to 4.0 GHz.
Cache: 16 MB Oryon Flex Cache for dynamic workload sharing.
GPU: Adreno GPU with Neural Fusion AI matrix cores, delivering 44% higher graphics performance and 40% better efficiency vs Gen 5.
AI Engine: Enhanced Hexagon NPU with transformer‑optimized accelerators.
Video: Supports 8K 60fps capture and advanced imaging features.
AI Capability: Can run 30B‑parameter mixture‑of‑experts models locally, enabling generative AI tasks without cloud reliance.
Snapdragon 8 Elite Gen 6
CPU: Same Oryon architecture, also breaking the 5 GHz barrier.
GPU: Re‑architected Adreno GPU with AI‑driven graphics, though less memory bandwidth than Extreme.
AI Engine: Hexagon NPU with 85% more performance and 20% better efficiency compared to Gen 5.
Target: Designed for broader flagship adoption, balancing performance and cost.
Comparison Table
| Feature | Snapdragon 8 Elite Extreme Gen 6 | Snapdragon 8 Elite Gen 6 |
|---|---|---|
| Process Node | 2 nm | 2 nm |
| CPU | 2 Prime @ 5 GHz + 6 Perf @ 4 GHz | 2 Prime @ 5 GHz + 4 Perf @ 4 GHz |
| Cache | 16 MB Flex Cache | Shared Flex Cache |
| GPU | Adreno w/ Neural Fusion + 18 MB HPM | Adreno w/ Neural Fusion |
| AI Capability | 30B MoE model locally | Up to 200M parameter models |
| Video | 8K 60fps capture | Advanced imaging, lower tier |
| Performance Uplift | +13% CPU, −37% power vs Gen 5 | Similar architecture, tuned lower |
What does different CPU Cores do ?
Prime Cores (up to 5.0 GHz): Handle the most demanding tasks, delivering peak performance for AI inference, gaming, and heavy multitasking.
Performance Cores (up to 4.0 GHz): Balance speed and efficiency, managing sustained workloads like video rendering and productivity apps.
Efficiency Cores: Optimized for low‑power tasks such as background processes, notifications, and lightweight apps, extending battery life.
Together, these cores form a heterogeneous cluster that dynamically allocates workloads, ensuring maximum responsiveness without draining power.
Qualcomm has decided to go without any efficiency cores and focus whole on Prime and performamce cores
More From us …
Discover more from WireUnwired Research
Subscribe to get the latest posts sent to your email.
