Ampere just published their developer newsletter and I thought that I would post the URL and then keep updating this post with the latest version.
We were quite busy in September, and it looks like October and November will be just as busy, with us attending events, planning for upcoming events, many more blogs, a new program, and a secret we can’t disclose yet.
Google Cloud Next
Ampere’s Developer Relations team was in San Francisco for the Google Cloud Next conference. We started off with an announcement of the Google’s C3A instances. These are Ampere’s new AmpereOne powered instances. They are currently in private preview and hopefully will be in GA soon. “This new Arm-based VM delivers up to 40% better*price-performance over our comparable current-generation x86-based VMs.” – Google press release
We hosted a happy hour up the street from Cloud Next on the first night and it was great to see around 100 people there having fun. For the main part of the event, we got to see several talks about how Google sees the future of the Cloud. As a generalized impression of the event, it seems that people has moved on from “this is what cloud is” to “this is how you optimize your cloud spend” and moving to Cloud Native Processors is an easy way to do this. Most of the sessions were recorded, so you can check it out here.
Oracle Cloud World
Another event and another announcement. Oracle announced that OCI will be doing a private preview of their A2 instances with GA coming early in the New Year (who will be first to go to GA?). Large parts of the keynotes were talking about the Ampere instances (A1 and A2) are currently in use by many companies. One of the more telling parts of the keynote was captured by Patrick Moorhead on his Twitter account, showing how much energy is saved by switching from legacy x86 chips.
The amount of savings that 8x8 experiences on a per rack basis.
Next up- Ampere at the OCP Summit
Ampere will be out in force at the OCP Summit in San Jose from October 17-19 - and to kick off the event in style, we are hosting an evening event on October 16th on the theme of sustainable and open AI solutions. Join us for drinks, snacks, and a conversation about moving the cloud forward with sustainable, high-performance solutions for the cloud in the era of generative AI.
The event is Monday, October 16, 2023, from 6PM- 8PM PT at the Westin Hotel’s Grand Ballroom and you can RSVP here.
And while at the main event, there are many talks by Ampere people, including Jeff Wittich, CPO, Sean Varley VP of Product Evangelism, Peter Pouliot, Developer Advocate, and others. They will be discussing a wide range of topics from AI in the cloud, to sustainability, to firmware… It should be an exciting 4 days!
KubeCon and the Co-located events
Are you going to KubeCon? We are. We are sponsoring AppDeveloperCon, which is part of the co-located events of KubeCon/CloudNativeCon in Chicago. The co-located events are the day before, Monday November 6th, with the main event starting on the 7th. So please come and see us. If you see a member of the DevRel team and you take a selfie with them, you can earn a special, limited community edition badge. Or if you don’t want a picture with us, at least say hi. In the community pages, please tell us what you are excited to see. Oh, and we are hoping to announce something at CNCF, just not sure if we can get it through everyone’s lawyers in time.
Developer Center Highlights
At Google Cloud Next and Oracle Cloud World, the word of the day was AI and it was everywhere. The basic message from vendors on the show floor was, “If you are going to do AI, which you must, you have to use our product!!” But there wasn’t much talk about what it means to use AI and why you want to do it. And more importantly, very little talk on how to do it correctly. This very much reminded us of the early days of Cloud computing, with very little talk about how to do Cloud correctly, just that you had to do it now! While ten years later, the talk is how to get the most out of your cloud.
For AI, we are just starting that first phase. But Ampere has been talking about AI for our entire existence. And some of the biggest questions for AI are, “What is the best processor for AI? Should you do everything on the fastest GPUs possible? But they are very expensive. What about legacy x86 processors? Or Ampere/AArch64?” Well, legacy x86 processors are cheaper than GPU’s, but are power hungry and wasteful, thus, not really cheaper. The trick for AI, like most things in life, is to use the right tool for the right job. Tony Rigoni, head of AI Business Development walks you through what is the best processor(s) for your AI deployment in his blog post.
After reading the post, check out our other AI solutions and blogs. Learn why FP16 boosts AI inference performance or check out our Optimized PyTorch and TensorFlow. (See we aren’t new to the AI world )
The community continues to grow and grow. We added 10% new members in September. There are a wide variety of topics created in the last month, from do we need more 101 content to using the Ampere Developer Platform as a gaming platform. And of course, lots of technical questions on how to get the most out of your processors. And for something completely different, check out the latest Jeff Geerling short video on how Cinebench 2024 performs on a 96 core Ampere Developer Platform. (spoiler- incredibly fast, even though Cinebench doesn’t know how to use more than 64 cores).
As we are growing, it is getting tougher and tougher for us to find all the content that people are writing about AArch64 Servers. Please help us out by posting links to your favorite blogs that you find. And if you have written something, please post it and let the community help amplify it! Our goal is to promote our community members and to promote and grow the wider AArch64/Arm64 community.
We want you to be able to get out there and speak at events, so we have been collecting CFP’s for events coming up. Check it out here and let us know where you are speaking.