Thanks, just checked Qualcomm's description of dragonwing.Qualcomm have their in-house Hexagon NPU which grew out of a DSP:
https://www.qualcomm.com/products/technology/processors/hexagon
The star of the Qualcomm® AI Engine
The Hexagon NPU, designed from the ground up for accelerating AI inference at low power, features the industry’s most advanced NPU architecture—evolving along with the development of new AI use cases, models, and requirements.
... but it's pretty feeble as they share the AI load between the NPU, CPU, and GPU with increasing load.
What is the benefit of heterogeneous computing for AI?
Heterogeneous computing for AI allows new AI instructions or hardware accelerators to be inserted into each processor. This processor diversity allows the architecture to continuously evolve for heterogenous computing features, in turn maximizing performance and power efficiency. This is beneficial as most other chipset vendors typically license several third-party processors, which cannot guarantee if they will fit or work under the same constraints.
The Qualcomm AI Engine, featured in our Snapdragon platforms and many of our other products, is at the core of our on-device AI and heterogenous computing advantage. With the CPU, GPU, NPU and Qualcomm Sensing Hub all working together and the result of many years of full-stack AI optimization, the Qualcomm AI Engine provides best-in-class on-device AI performance at extremely low power to support use cases today and in the future.
https://www.qualcomm.com/content/dam/qcomm-martech/dm-assets/images/uploads/npus.png
View attachment 78946
It says "powered by leading-edge on- device, AI, computing and connectivity"
It does mention Neuromorphic AI at all but 'leading-edge on-device confuses things a little.
Dragonwing | Qualcomm
Powered by leading-edge on-device AI, computing, and connectivity, Dragonwing products and solutions unlock potential, so today’s industries and businesses can scale to new heights.