NVIDIA's CUDA has expanded its support to RISC-V processors, and this marks a significant milestone towards the enablement of the platform, especially in the AI domain.
RISC-V Architecture Will Now Support NVIDIA CUDA Workloads Through Code Porting
The AI markets have been relying on data center CPUs based on x86 and ARM architectures for quite some years now. Intel and AMD dominate with x86 offerings, while NVIDIA and other Big Tech companies have their specialized ARM solutions, but besides that, we have no other architecture in the race for now. This seems to change, as based on a development revealed by RISC-V themselves, it is disclosed that NVIDIA's CUDA software stack will now be allowed to run on a RISC-V processor through a porting mechanism, potentially fueling market adoption.
Well, NVIDIA's CUDA is the "big boss" when it comes to AI computation, and more importantly, many industries revolve around it, so with RISC-V getting a dedicated CUDA port platform, this will allow for a more extensive adoption. Several benefits are associated with the RISC-V architecture, but for the AI industry in particular, a big one comes from having no licensing fees. Developers and companies can use, modify, implement, and distribute the open-source ISA with no royalties, allowing for a more extensive adoption with small-scale customers and startups.
Apart from this, RISC-V is claimed to be scalable-friendly. Through a minimal instruction set, the chip design and verification process becomes much easier, speeding up development and testing. RISC-V has a much larger prospect, especially in platforms where edge AI is required, considering that ARM and x86 are currently dominating large-scale clusters. The AI segment hasn't adopted RISC-V in a wider term, but one firm does stand out, which is Jim Keller's Tenstorrent.
Tenstorrent aims to provide powerful yet cost-effective AI chips through its Wormhole AI chips, notably the Wormhole n150 and the Wormhole n300. Apart from this, RISC-V has a much broader application with Chinese developers, since the platform's open-source nature attracts the region, and with NVIDIA's CUDA support, we will see a much higher interest in the AI segment.
Follow Wccftech on Google to get more of our news coverage in your feeds.
