Arrcus upgrades ACE-AI solution for distributed AI applications at the edge
Arrcus, a company specializing in hyperscale networking software, has announced upgrades to its ACE-AI platform to meet the increasing demand for unified networking solutions for distributed AI workloads.
As AI tasks spread across various locations due to economic and application needs, Arrcus ACE-AI facilitates networking and delivery of applications at the edge with fast connectivity, according to the company.
The platform caters to the evolving federated learning model for AI, allowing collaborative training of models with decentralized data while ensuring smooth execution of tasks in hyperscale environments at the edge for diverse applications.
Arrcus recognizes the necessity of a unified networking fabric that can connect these distributed workloads irrespective of their locations. The modern data center landscape requires high throughput and low latency, both of which Arrcus says its ACE-AI platform aims to deliver while minimizing CPU overhead.
Shekar Ayyar, chairman and CEO of Arrcus, highlights the importance of flexibility and intelligence in networking solutions for AI workloads, emphasizing the platform’s ability to connect and orchestrate distributed AI tasks effectively.
The evolving nature of artificial intelligence, high-performance computing, and storage workloads presents challenges for data center networking. Arrcus claims to address these challenges by introducing features like RoCEv2, PFC, ECN, ETS, AR, Dynamic Load Balancing, and Global Load Balancing, which contribute to building a robust Ethernet fabric.
According to the company, another critical challenge in achieving high-performance networking for AI is the limitation of traditional TCP/IP stacks due to high CPU overhead at such speeds. Arrcus aims to tackle this by incorporating RDMA technology, which offloads transport communication tasks from the CPU to hardware, thereby providing direct memory access for applications.
Additionally, Arrcus announces support for industry-leading switching platforms like Tomahawk5, Jericho3, and Ramon3 from Broadcom, in partnership with device manufacturers such as Ufispace and Edgecore.
“Broadcom is very excited to collaborate with Arrcus to deliver industry-leading switching solutions that are optimized to meet the performance demands of next-generation AI workloads. Together, Arrcus and Broadcom are enabling customers to build high-performance, scalable, and intelligent data center networks,” says Ram Velaga, senior vice president and general manager, core switching group at Broadcom.
Arrcus offers high-performance networking suitable for AI applications, multi-cloud cost optimization, automated 5G network slicing for AI, and end-to-end network visibility through its ArcIQ capabilities.
Read more:
Arrcus launches FlexMCN multi-cloud solution on Equinix edge infrastructure
What is an SBOM and why is it critical for IoT security?
Article Topics
AI | Arrcus | connectivity | edge
Comments