Eridu

Thanks to everyone on the Eridu team who helped us have a great time celebrating Diwali, the festival of lights, this week! 🪔

We shared food, sweets and conversation while decorating diyas, taking a break from the hard problems of silicon and systems to mark the occasion together.

To all who celebrated, we hope your Diwali was filled with light, joy and peace. ✨

Celebrating the traditions that matter to our team around the world is an important part of Eridu's culture. Learn more about what it's like to work at Eridu at eridu.ai/careers.

Discussions about AI data centers often focus on the number of GPUs, power requirements and costs. But an underappreciated resource is just as critical to the productivity and success of those data centers: the network.

As GPU performance, cluster sizes, and workload demands have raced ahead, data center networking has struggled to keep pace. The consequences of that reality impact the productivity of the entire investment: lower tokens per second, fewer tokens per watt and higher costs per token. Network complexity also adds to the risk of delays before the data center can even begin operations.

In a new article in @dcdnews, Eridu co-founder and Chief Product Officer @omarhassen examines why networking is becoming a hard limit on AI infrastructure productivity. He discusses:

▶️ Why the network has become a critical constraint in AI data centers

▶️ Why the network needs to be part of data center design from early stages

▶️ How high-radix switching can help by flattening networks, reducing complexity and improving performance and latency

Read the article for a closer look: datacenterdynamics.com/en/opinions/th…

#AINetworking #AIInfrastructure #AIDataCenters

AI scale-up infrastructure needs to move beyond the rack. Today's scale-up architectures are reaching their limits, requiring a new approach.

AI workloads are growing and evolving in ways that are straining the network--more model parameters and active experts, more data movement for communication collectives, less predictable traffic, and greater demand for heterogeneous compute are just a few examples. But today's single-rack scale-up architectures constrain how far domains can grow while meeting the bandwidth and latency requirements of AI workloads.

That's why Eridu helped launch a new workstream within the Open Compute Project focused on disaggregated, multi-rack scale-up architectures.

The initiative is focused on an open, standards-based architecture that extends scale-up across racks while maintaining a single tier of high-performance switching. The goal is to support larger scale-up domains, heterogeneous xPUs, and greater choice in compute and infrastructure components.

We outline the architectural challenges, what is needed to deliver 5000+ GPU scale-up domains, and the principles guiding this initiative at eridu.ai/news/why-ai-ne….

#AINetworking #AIInfrastructure #ScaleUp

Network bottlenecks and congestion can be scary. But our team of superheroes, villains and other characters has come together to drive that fright away.

Solving the challenges of networks that are not able to keep up with the demands of AI requires more than incremental improvements to existing silicon, software and systems. It requires redesigning the network switch from a clean sheet, building for petabit scale from the start.

To do that, we've assembled a world-class team with deep experience and expertise. We are continuing to expand our team to accelerate our work--from ASIC design and verification to physical design, systems design and software engineering.

If you want a chance to work on revolutionary networking technology and make an impact, check out our current job openings at eridu.ai/careers. Whether you come in costume or not, Eridu is looking for more characters to help reengineer network switching for AI.

Hope everyone had a great Halloween!

#AIcareers

We're looking forward to next month's TomorrowX Summit!

Eridu CEO Drew Perkins will be joining other thought leaders in business and technology to discuss the trends and disruptions that are shaping the tech investment landscape. The evolution of AI and the infrastructure needed to support it is going to be an important and interesting discussion at the event.