3rd-Gen Qualcomm Oryon CPU Explained: The Key Upgrades You Should Know

Qualcomm’s third-generation Oryon CPU is the company’s latest custom architecture for Windows laptops, and this version focuses on widening the core, increasing cache, and adding CPU-side AI acceleration.  Also Read: Using Android 14, 15, Or 16? CERT-In Says Your Phone Might Be At Risk

The biggest upgrade sits in the Prime cores. Qualcomm keeps the 6-core Prime cluster layout, but almost everything around it has been expanded. A 9-wide decoder replaces the earlier setup, letting the CPU dispatch more micro-ops per cycle. As per Qualcomm, the cores can now reach up to 5.0GHz boost, Qualcomm suggested.  Also Read: B by Lenskart AI Smart Glasses To Open Developer Access In India, Launch Expected By December

Fetch, branch prediction and decode pipelines get improvements too, including a 1-cycle branch target buffer, 2-cycle conditional predictor, and 16-instruction fetch per cycle. Qualcomm’s comparison chart shows that this generation targets up to 39% higher single-core performance while using up to 43% less power at the same level as before. Also Read: OnePlus 15 Launch Date Confirmed, Coming October 27 With OnePlus Ace 6: What To Expect

3rd-Gen Qualcomm Oryon CPU With Much Larger L2 Cache

One of the most visible changes is the cache size. The third-gen Oryon features 44MB of L2 cache in total: 16MB per Prime cluster + 12MB on the Performance cluster.

The tech giant says that the cache sits physically close to the cores for ultra-low latency access, and that the L2 can handle 220+ in-flight transactions, which helps with heavy multitasking and memory-bound workloads.

3rd-Gen Qualcomm Oryon CPU: Matrix Engine, Performance Cores, Security, More

Each cluster includes the Qualcomm Matrix Engine, a CPU-side accelerator for machine-learning operations. This block handles vector and matrix math using 8×8 and 4×8 ML grids, and uses 512-bit vector registers. It also runs in its own clock domain, meaning AI workloads don’t interfere with CPU frequency.

Talking about the performance, the Performance cluster – also 6 cores – is built for lower power and long-duration tasks. 

• 3.6GHz base clock
• 12MB L2 cache
• Similar out-of-order design as Prime, but less wide, with fewer execution pipes and a smaller TLB.

Qualcomm says these cores can operate below 2W, which is important for background and steady workloads.

The third-gen Oryon also updates security at the core level. The Memory Management Unit also expands support up to 1GB page sizes, adds 16 in-flight table-walk requests, and includes stronger TLB prefetching.