AMD Confirms Zen 4 EPYC Codename, and Elaborates on Frontier Supercomputer CPU


RSS feed bot

After the Computex Keynote today on stage, where AMD revealed its new Ryzen family of processors coming on 7/7, we had a chance to speak with AMD’s SVP and GM of the Datacenter and Embedded Solutions Group, Forrest Norrod. Specifically, we asked him about the ‘Road to Rome’, which AMD stated will be coming in Q3 this year. We have a full transcription of the interview planned, but a couple of key points came out of our discussion.

First, Forrest revealed/confirmed the name of the Zen 4 based EPYC processor that will go beyond Rome and Milan. That name is Genoa, which is follows the Italian naming pattern. Forrest also said that the Zen 5 product follows that pattern, but failed to elaborate what that name is. I was told that the goal with each generation is to make sure the low hanging fruit for performance on each design is taken, and that there is a continual drive for compute performance. Forrest wouldn’t comment on the timeframe for Genoa, but did state that Milan is a mid-2020 platform, and so Genoa is likely 2021/2022.

AMD EPYC CPU Codenames
Gen Year Name Cores
1st 2017 Naples 32 x Zen 1
2nd 2019 Rome 64 x Zen 2
3rd 2020 Milan ? x Zen 3
4th ? Genoa ? x Zen 4
5th ? ? ? x Zen 5​

The other snippet of information I wanted to break out from the interview is about the new Frontier supercomputer that was recently announced. This machine, built with AMD CPUs and GPUs, has a goal of being the most powerful supercomputer in 2021, measuring around 1.5 ExaFLOPS, making it a truly exascale machine. It has been largely assumed that this would be a combination of Milan CPUs and Radeon Instinct GPUs, and it would be connected CPU-to-GPU and GPU-to-GPU by Infinity Fabric rather than PCIe or CCIX. Forrest explained that the CPU is not Milan – it is actually a fully custom design CPU specifically for this project.

Forrest clarified that this custom CPU is not in the same way that Intel defines custom – i.e. it’s not simply the same silicon with adjustments in core counts / frequency / cache. The CPU for Frontier will be a fully custom design, built with CPU-to-GPU IF links in mind, without any excess. When asked if IF is going to be a connectivity in other platforms, Forrest would only confirm that it’s the connectivity for Frontier. Though, for what it's worth, AMD announced back at the Frontier unveil that the CPU would ultimately become available for other enterprise customers as well.

As mentioned, I’ll be trying to transcribe this interview ASAP, along with our Lisa Su roundtable. Stay tuned for more of our Computex coverage.

Want to keep up to date with all of our Computex 2019 Coverage?



Follow AnandTech's breaking news here!

Continue reading...