AMD Advancing AI 2024

Table of Contents

Introduction

This article will outline my experience at the AMD Advancing AI 2024 event, held on October 10, 2024. I was invited and sponsored by AMD to attend, as part of a contest I participated in – the Hackster x AMD Pervasive AI Developer Contest. While I did not win a prize for my contest submission, I was happy that I was able to participate, and learned a lot.

Background

I joined the Pervasive AI Developer Contest in November 2023. My project was a further exploration of concepts from a Hackathon, using AI to help troubleshoot Kubernetes clusters. I used tools such as K8sGPT, LocalAI, and Open WebUI to put together a self-hosted stack that could scan and diagnose issues inside Kubernetes clusters. It would report problems and potential solutions in plain English. For more details on my project, including source code, see this article.

I was invited to attend the conference on October 10, 2024; where the winners of the contest would be announced. AMD sponsored my attendance by covering flights, hotels, and food. I flew out to San Francisco on October 8, and returned on October 11.

Pre-Conference Events

On October 9th, there were mainly events for press and analysts. I ran into Wendell from Level1Techs at the hotel, and we chatted a bit about my project on the way over to the Moscone Center to register and pick up our badges. I think he had just arrived from the airport, so was a bit on the rough side.

I spent a bit of time in the event area working on a few things, and watched everything getting set up.

During the evening, there was a dinner at The View Lounge at the Marriott Marquis, on the 39th floor of the hotel, which had some amazing views of the sunset over the city:

Event Keynote

Here is a full stream of the entire keynote:

At the start of the official event on October 10, 2024, the keynote started with Dr. Lisa Su announcing the 5th Gen AMD EPYC “Turin” CPU, designed for Cloud, Enterprise and AI. This article from Phoronix has much more details on the specs for the various models:

Wendell has done a great video on the potential of this chip, some benchmarking, and it’s pretty exciting what it’s capable of:

Additional details and product announcements were made around the the AMD Instinct™ MI325X Accelerator GPU. There was also a lot of announcements on AMD ROCm Software improvements, including improvements to the open software stack.

From the datacenter networking side, the AMD Pensando™ Pollara 400 Ultra Ethernet AI NIC was announced. Using the Ultra Ethernet standard, it provides 400Gbps of bandwidth to interconnect GPU enabled servers on the back end to transfer data for processing.

The AMD Pensando™ Salina DPU (Data Processing Unit) was also covered, designed for connecting the front-end network for client access.

In addition to the product announcements, there was also a number of AMD Partners presenting, such as Meta, Google Cloud, Microsoft, Dell, HP, Lenovo and SuperMicro. This included various product launches for datacenter servers supporting the new AMD platforms:

There was also the announcment of the AMD Ryzen™ AI PRO 300 Series AI chip for laptops and PC. A lot of this revolved around Microsoft’s announcements regarding Copilot, the re-launch of Microsoft Recall, and other ways Ryzen AI can be leveraged.

AI Hub – Expo

In the AI Hub Expo area, there were a number of booths for sponsors and hardware vendors, as well as showcase areas for some of the announced AMD technologies.

Some of the contest winners had their projects set up to demonstrate. For example, for the AMD Radeon Pro W7900 GPU category, the first place winner had their Infinite Sands Project on display. The project allowed you to shape physical sand, then used AI models to generate artwork based on the designs in the sand. The design could be captured as an image, and also projected back onto the sand itself for a three-dimensional art piece.

AMD Illuminate Sessions

After lunch, there were a number of developer focused sessions, with speakers from organizations such as OpenAI, Meta, Microsoft, and UC Berkley. These were technical sessions that focused on AI related topics.

There were presentations on projects and open-source initiatives like OpenAI Triton, the Meta Llama Stack, SGLang, and vLLM; as well as a talk from Microsoft on their experience with the AMD Instinct™ MI300 Series Accelerators. I’ll have some photos of some of the slides in the photo albums below, but there was a lot of interesting insights into these projects.

Pervasive AI Contest Winners

After the presentations, Dr. Lisa Su gave closing remarks for the day, and then presented the first place winners for the Pervasive AI Developer Contest. I’ll have a full photo album for all of the winners below. She was patient enough to stay around after the awards were presented for some group photos and selfies.

AMD had their own photographer for the event, and I’ll try to get copies or links to the photos once they are available.

After the AMD Luminaries sessions and contest announcements, there was a dinner and networking event in the AI Expo area.

Ramine Roane (AMD – Corporate VP, AI) and Prithvi Mattur (AMD – Senior Program Manager, Data Center, Cloud & AI) presented the other awards in the contest. I’ll have a full photo album below. All of the winners were announced a few days later through the Hackster contest page.

Conclusion

AMD Advancing AI 2024 was a great experience. I was very happy to be able to participate. I was able to meet a lot of industry leaders, analysts, AMD employees, tech journalists, and other contestants.

I very much appreciate AMD inviting and sponsoring my attendance, and for all the support during the planning and coordination. There were a number of AMD representatives that were helpful in organizing all of the contestant attendance, as well as the event itself.

I learned a lot of about developments in the AI/ML space and initiatives to grow adoption. Hopefully the photos and videos coming out of the event will be helpful to you as you grow your adoption with this technology.