WilliamLam.com

ESX Passthrough of AMD Ryzen Integrated Graphics (iGPU)

by // 1 Comment

I recently had a conversation with Wenchao (creator of AMD Zen4/Zen5 IPMI Thermal Driver for ESX Fling and Realtek Network Driver for ESX Fling) about the incredible work he has done to support the VMware/Broadcom community. His latest contributions has really closed the capability gap when running ESX on popular AMD Ryzen platforms like the Minisforum MS-A2, which been an ideal system for running VMware Cloud Foundation (VCF) 9.1 in a lab environment.

The only remaining capability we have not yet been able to take advantage of is the integrated graphics (iGPU) on modern AMD Ryzen systems, which have become surprisingly powerful for running AI models.

While this has been a topic I have researched quite extensively and even spent time debugging with Engineering, it is always great to get a new perspective on things. I kid you not, a few days later Wenchao came back with a solution that had a few caveats which I was able to help iron out with some of my prior experience in this area. 😆

[Read more...]

AMD Zen4/Zen5 IPMI Thermal Driver for ESX Fling

by // 11 Comments

Happy Friday! 🥳

A couple of weeks back, Wenchao (creator of the Realtek Network Driver for ESX Fling) reached out to me to share an exciting development he had been working on.

Unlike traditional enterprise hardware, which typically includes an Intelligent Platform Management Interface (IPMI) for exposing various hardware statistics, consumer systems like the popular Minisforum MS-A2 lack this capability out of the box.

Wenchao has developed a pseudo-IPMI driver for AMD Zen 4 and Zen 5 platforms that surfaces the CPU thermal and I am excited to share this will be released as a new Fling!


In addition to viewing this information via the vSphere UI, users can also retrieve the raw metrics via ESXCLI:

esxcli hardware ipmi sdr list


The driver currently supports the following:

  • ESX 9.1
  • ESX 9.0.x
  • ESX 8.0 Update 3

[Read more...]

Quick Tip: High CPU Utilization on ESX due to Slow Entropy from AMD Zen 4 CPUs

by // 3 Comments

I had been troubleshooting a stubborn CPU utilization issue with a workload that, over time would also overrun the CPU on my physical ESX host. The assumption was that the workload was causing the issue, but after several rounds of collecting various ESX performance statistics, there was nothing conclusive that the workload was the culprit.

An interesting observation from some of the ESXi VMkernel Engineering team was that my VMkernel log contained a large number entropy errors:

NRandomHwrng: 246: Out of entropy, refreshing

The engineering team suspected that these entropy issues could actually be the root cause of the issues I had observed, especially as they have seen something simliar in another case when entropy requests fail.

[Read more...]