“Ampere CPUs have enough host power to give you at or better throughput than the incumbent[s], so don’t take benchmark[s] at face value. Run your own workloads [and] see what happens to your total cost analysis.”
This quote, from Vikrant Soman, Cloud Infrastructure Engineer at Uber, is just one of the many customer testimonials heard during Ampere’s first Developer Summit, held on September 26th, 2024. This half-day virtual event brought together developers and innovators from across the globe to dive deep into cloud architectural migration and explore how Ampere(R) Cloud Native Processors are shaping the future of compute.
The Summit kicked off with a welcome and opening keynote from Pete Baker of Ampere Computing. Eight partner presenters joined us for four in-depth technical sessions, highlighting the challenges and benefits of migrating real-world applications to Ampere hardware cloud instances.
All the sessions were recorded and are now available on Ampere’s YouTube channel and Developer Playlist. We invite you to view each of the 30-minute sessions, where you can learn about:
- Concerns and opportunities with mixed architecture builds and deployments, with insights on tools and methods to improve the migration experience when using Ampere platforms – Keynote by Pete Baker of Ampere.
- Why cloud native processors matter across power efficiency, performance, and sustainability – Hosted by Sean Varley, Ampere’s Chief Evangelist, with contributions from Kate Goldenring of Fermyon and Dor Laor, CEO of ScyllaDB. They discuss how Ampere’s high core density and energy-efficient processors are ideal for modern applications, including using Kubernetes and distributed databases.
“ <Zeiss> was able to reduce their costs by 60% while also maintaining performance. And part of that cost reduction was once they switched to Webassembly, they could add in more Ampere Armnodes to their cluster.” Kate Goldenring – Fermyon
- Which workloads and applications work best on Ampere CPUs? – Presented by Naren Nayak, VP of Customer Engineering at Ampere, alongside Vikrant Soman from Uber and Dmitry Polyakovsky from Oracle. This session explored the types of workloads that see the biggest benefits on Ampere CPUs, including AI, stateless cloud workloads, and high-performance cache services.
“It’s no longer like Arm is operating in a vacuum and the ecosystem is sort of like a step function behind. No, it’s right shoulder-to-shoulder with how the x86 ISA works.” Vikrant Soman – Uber Vikrant Soman – Uber
- How cloud native CPUs enable optimal compute at a substantially lower cost and enhance AI inference – Hosted by Victor Jakubiuk of Ampere and featuring Amar Gowda, Senior Principal Product Manager at Oracle and Steve Notley, Director of Field Engineering at Wallaroo.AI. In this session, Victor, Steve, and Amar explored how Ampere processors can optimize cost, performance, improve onboarding for learning models, and reduce the cost of running AI inference workloads in the cloud.
“By deploying this wallaroo.ai package, llama.cpp quantized model, taking advantage of Ampere’s accelerators on OCI A1 machines were able to see double the inference speed relative to equivalent x.86 deployments.” Steve Notley – Wallaroo.AI
- Insights on migrating from legacy architecture in a mixed architecture world – Led by Dave Neary, Head of Developer Relations at Ampere, with guest speakers Maximilian Wittich from Stackable and Andreas Winther Lykke from Uber. The session shared nuances in adding a new architecture, including practical advice for rebuilding and testing software on that architecture, packaging it, and managing daily operations in production.
“We of course faced issues, but they were relatively easy to fix, and I think in a matter of weeks we had actually built like, 3000 services.”
Andreas Winther Lykke – Uber
Built for sustainable cloud computing, Ampere’s Cloud Native Processors deliver predictable high performance, platform scalability, and power efficiency. It’s good for the developer, good for your wallet, and good for our planet.
We encourage you to learn more and become part of the conversation using the following resources:
Craig HardyCraig Hardy is a Senior Technical Program Manager at Ampere Computing with over 30 years of high-tech experience in finance, operations, marketing, and software ecosystems for the client, data center, and cloud computing. He is energized by simplifying complex issues into straight forward execution steps. Outside of work, Craig spent nearly a decade owning and operating a local bakery. He and his family live in Portland, Oregon.
