Let's talk about the curious case of the 2990WX performance regressions vs the 2950X where the 2990WX performs worse *on windows* than Linux.
I am on 4.18 but the video shows 4.18 4.15. Hmm…
$ uname -a
Linux localhost.localdomain 4.18.7-200.fc28.x86_64 #1 SMP Mon Sep 10 15:44:45 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
I wonder if anyone has tried Windows Server 2016. It’s not inconceivable that M. included a dumb scheduler in Windows 10 that isn’t aware of NUMA and is intended to benefit interactive use; whereas the Windows server scheduler is a touch more sophisticated.
Given that there are different memory limits, it’s not inconceivable although my only ‘evidence’ is that server desktop always seems just a touch sluggish. And I’m no Windows expert even if I have been known to use it from time to time (with much swearing).
It is a shame about the new AGESA bugs. Directly related to virtualization too.
I don’t think my debugging level is advanced enough for these early stages if I was sent a system. I would know GUI based workarounds, but not so much the actual code fixing.
For Cinebench on Ryzen1700X, Server2016 gets beaten by Win7 (4 runs on each, highest recorded).
The docs from Microsoft say the only change is the defaults. So if you go into system somewhere there is a radio button to optimize for background processes or interactive processes. And that’s it.
$ uname -r
4.19.0-rc3
The 2990WX runs well on this kernel. I have a copy of Server 2016 but I start to twitch at the thought of installing it…
What if you also disable core 0 on other AMD CPUs, or on Intel ?
I have a question about Lasso… does it really help with performance that much? I do take a frame rate hit while iRacing and streaming, wondering if I can mitigate that. Also maybe do something with ESO’s horrid 1 core from hell coding