As the vRAN / Open RAN deployments are slowly ramping up, there are a lot of questions on the best compute architecture to use, especially for L-1 processing. The objective is to utilize the most power-efficient, cost-effective, and scalable option. The popular choice seems to be using optimized silicon instead of a generic CPU.
In this episode, I talk to Joel Brand of Marvell Technology about why silicon optimized for radio workloads such as DSP-based accelerator is better. Joe also shared details about their Octeon10 Fusion processor announced at MWC 2023 and their partnership with Nokia, Samsung, and others.
My article on the subject:
What is the best compute architecture for vRAN/open RAN?
