with AMD processors that integrate a 50 TOPS NPU, belonging to the new generation Ryzen AI 300 Series (Strix Point): no more is needed since the PanterLake will be very expensive and we will have to see what surprises ZEN 6 has in store for us.
Interesting...
However Team Blue will need to actually DELIVER on Performance, cost & availability to gain any inroads or market share away from AMD.
Sadly for Intel "performance" needs BOTH mature drivers AND capable hardware. Nether of which Team Blue has a record of delivering.
Then as others will likely mention, Intel has a habit of pricing their products much higher than AMD even if they don't perform as well. IMHO for Intel to make any market inroads they will need to be competitive on price.
The odds of Team Blue "fumbling" the launch is much greater than zero.
Intel is already using 3nm, 18A is not much better than TSMC 3nm. Increasing the number of shaders won't help if there's no progress in the manufacturing process and bandwidth.
Quote from: Fierce.XT on July 29, 2025, 21:49:33Intel is already using 3nm, 18A is not much better than TSMC 3nm. Increasing the number of shaders won't help if there's no progress in the manufacturing process and bandwidth.
You know nothing about 18A, and here you are making baseless claims for an unreleased cpu. 18A is far more advanced than the poor performing N3B, regarded as one of the worst process node jumps ever for minimal improvements over N4P. 18A has back side power delivery, uses GaaFET, moves the memory controller back to the cpu tile for reduced latency, has 8-12% IPC uplift over Arrow Lake, will clock higher, use less power, and has far better iGPU, far better NPU.