AMD is here to talk about Cloud native solutions, and started with an understanding of what it truly is. With it being application deployment at a faster speed. With a slide like this I don’t care what the company is building, I want to know more about what they what to say:

Within the cloud there is a tenant operator distinction where the tenant is a consumer of the cloud and the operator is who runs the cloud. One thing that AMD is seeing is operators wanting to move their running solutions within a DPU(Data processing unit) and not within the main solutions of the computer.

Probably the best slide I’ve seen on a DPU and what it looks like:

While AMD continued to discuss their value and what they are able to do with multiple use cases, I can’t help but wonder why this matters. Why does AMD want to tell us about their usage in the cloud, when hyperscalers decide how the hardware is used. Shouldn’t chip manufacturers be talking to the actual providers within the cloud, and not the customers?

My perspective may be a little askew as it comes from an AWS perspective. When talking to AMD or Intel at RE:Invent I always like to ask what they bring to the customer at those events, and I always hear the same thing, “NEW AMI’s!”. Excuse my insane eye roll, but the increase of different images or shapes, or however you want to define them is just unnecessary and annoying. It should be basically simple. VMware does it all the time with how it defines the amount of vCPU and RAM to each virtual machine that is built. My petty annoyances aside, if you were to do a quick google search on AMD and *Insert Hyperscaler* you would find out that the way AMD is utilized is widely varied between the different clouds you are using. For instance Oracle(OCI) utilizes AMD with high cpu and utilization while allowing it to build additional resources, which is a big deal when you think about how many GPU, and HPC resources can be utilized rather than upscaling to a single level and keeping it at a lower ratio of usage. Azure is using AMD for HPC and other solutions as well including dedicated solutions to really take advantage of AMDs chips.

The truth is, if you look at CPU vendors and you only know one cloud provider then you may be missing a very large amount of what that chip manufacturer is doing, and how that will impact the world. I don’t think it’s an embellishment to state the impact on the world. With the ways we are creating the silicon, the different power consumption due to different cores being utilized, and on top of all of this the idea of Moores law being dead and what that means. So things get more expensive, that’s agreed upon, but what about power consumption? Do the other chip manufacturers that are making chips with lower power consumption and with an understanding of how they actually should be built take a backseat because of the flash and power of high consumption solutions that literally no one uses!?

In conclusion of this rant of a post. We as the customer need to do our due diligence to pay attention to what chip manufacturers are doing. Team green(Nvidia) definitely gets the limelight(pun intended) when discussing power and the insane about of throughput it can provide, but at what cost?? Team blue(Intel) is building chips but is constantly put up against team red(AMD) and AMD is doing things correctly thinking through how everything should be done to provide caring for the world and the chips being produced, and continue to be put on the back burner outside of the high CPU requirements. Though we heard more and more about the chips that AMD created, and it didn’t really talk through how it impacts the cloud directly, I couldn’t help but go off the rails and discuss how these things truly matter as we progress in this journey we call life.