May 23, 2024
Ampere and Qualcomm Launch Arm-Based AI Server

Ampere and Qualcomm Launch Arm-Based AI Server

Ampere and Qualcomm may not seem like the most obvious partners since both offer Arm-based chips for data center servers, though Qualcomm’s primary market is still mobile. However, the two companies have announced a collaboration to deliver an AI-focused server that utilizes Ampere’s CPUs and Qualcomm’s Cloud AI100 Ultra AI inferencing chips for running — not training — AI models.

Like every other chip manufacturer, Ampere aims to capitalize on the AI boom. The company’s primary focus has always been on developing fast and power-efficient server chips. While Ampere can integrate some AI features using Arm IP, AI is not its core strength. To bridge this gap, Ampere partnered with Qualcomm and SuperMicro to integrate their solutions, as explained by Arm CTO Jeff Wittich.

“The idea here is that while I’ll show you some great performance for Ampere CPUs running AI inferencing on just the CPUs, if you want to scale out to even bigger models — multi-100 billion parameter models, for instance — just like all the other workloads, AI isn’t one size fits all,” Wittich told TechCrunch. “We’ve been working with Qualcomm on this solution, combining our super efficient Ampere CPUs to do a lot of the general purpose tasks that you’re running in conjunction with inferencing, and then using their really efficient cards, we’ve got a server-level solution.”

Collaboration and Efficiency in AI Solutions

Wittich emphasized that Ampere’s partnership with Qualcomm is about creating best-of-breed solutions. He highlighted the “really good collaboration that we’ve had with Qualcomm here,” noting shared interests in building efficient solutions across different market segments. Ampere focuses on highly efficient server CPUs, while Qualcomm develops efficient solutions for various parts of the market.

This Qualcomm partnership aligns with Ampere’s annual roadmap update. A significant highlight of this roadmap is the new 256-core AmpereOne chip, which is built using an advanced 3nm process. Although these new chips are not generally available yet, Wittich mentioned they are ready at the fab and should be rolled out later this year.

The new AmpereOne chips feature 12-channel DDR5 RAM, allowing Ampere’s data center customers to optimize memory access for their users’ needs. The sales pitch for these chips goes beyond performance, emphasizing power consumption and cost efficiency in data centers. This is particularly relevant for AI inferencing, where Ampere compares its performance to Nvidia’s A10 GPUs.

Expanding Partnerships and Future Prospects

In addition to the Qualcomm collaboration, Ampere announced another partnership with NETINT to create a joint solution that pairs Ampere’s CPUs with NETINT’s video processing chips. This new server will be capable of transcoding 360 live video channels in parallel while using OpenAI’s Whisper speech-to-text model to subtitle 40 streams simultaneously.

Ampere CEO Renee James highlighted the company’s long-term vision in today’s announcement: “We started down this path six years ago because it is clear it is the right path. Low power used to be synonymous with low performance. Ampere has proven that isn’t true. We have pioneered the efficiency frontier of computing and delivered performance beyond legacy CPUs in an efficient computing envelope.”

It’s important to note that Ampere is not discontinuing any of its existing chips in favor of the new ones. Wittich stressed that even the older chips still have significant use cases. Ampere’s strategy focuses on providing efficient and powerful solutions that meet the evolving demands of data centers and AI applications, reinforcing its commitment to innovation and performance.

Image by Brian Penny from Pixabay

Related posts

Women-Led Web3 Startups Secure Only 6.43% of $27.85 Billion Investments

Henry Clarke

UK’s NCSC Report: AI Progress Could Counterbalance Rising Cyber Threats

Bran Lopez

AI Developer Anthropic Fires Back at Universal Music Group’s Copyright Claims

Henry Clarke

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Please enter CoinGecko Free Api Key to get this plugin works.