AMD 5800X3D performance

Your problems all arise because your 3080 is basically idle, you are obviously using low graphic settings on an undersized monitor and then using DLSS quality mode for which it is way too powerful. In turn you are trying to counter this with basement level LODs in an effort to get gpu limited and consequently the CPU has not enough to do.

I suggest you set up a custom 3840 x 2160 desktop in NVCP which will then be available to use in MSFS and you should change that DLSS ‘quality’ setting to DLAA. Then raise your graphic settings to ultra and raise LOD as high as you can without stuttering.

Get that excellent hardware working as hard it surely wants to.

3090 Ti shows 45-65% core utilization
Sim running in 2K
27" native 2K monitor
TOD 200
LOD 130
Most everything on High/Ultra
I’ve tried it with TAA, and different flavors of DLSS - no real change in CPU usage.

Microsoft recommends high end hardware needs 4k but you can also use multi monitor. Get that 3090 working at 100% and maybe you can run T-LOD 400, O-LOD 200 and traffic 100% … it’s entirely your choice.

Don’t forget you are running an all core overclock, those unused cores don’t slow down when they have nothing to do and that’s is showing up in your HWinfo. A per core negative curve would be better and run cooler. From a base 3.8GHz my 5800X can actually boost to 5.1GHz, you won’t get that but you should be able to get 4.8 or 4.9

I don’t know what to believe. Here’s another screenshot. Massive discrepancy between Task Manager, Open Hardware Monitor, and CPUID HW Monitor.

You can see my GPU load, which is pretty normal for my in game settings.

1 Like

Yep it seems OHW is representing usages wrong … throw it in the bin.

1 Like

What does “throw it in the bin” mean?
I’m not familiar with that term.

High end hardware NEEDS 4K??? Never heard that before. You need high end hardware to run 4K at High/Ultra settings. Hardware capable of running 4k will run 2K even better.

Also, those cores (according to Task Manager) are definitely doing something. MSFS might not be using them, but why would I want to cripple them if doing so wouldn’t affect MSFS one bit?
Not to mention that there are other processes and services that the sim uses beside the .exe file.

2K resolution, Ultra/High graphics settings, GPU usage @ 50% (less heat, less fan speed.)
My CPU never gets above 60°C, GPU never gets above 65°C.
I’m getting a solid 60fps with good frame times (74fps / 13ms in my latest screenshot.)

I think everything is pretty well dialed in as is.
I’m just curious about why the hardware monitors are so wacky.

Something isn’t right there, are your bios a beta? The windows resource manager looks about the most accurate.

For MSFS Microsoft officially recommends
High end 4k
Mid range 1440p
Low end 1080p

Maybe NEEDS is a bit strong but most performance issues I see are caused by an unbalanced system.

As a matter of fact it is a Beta BIOS. It’s been very stable, but maybe you hit a nail on the head.

1 Like

Task Manager’s output there looks realistic for a machine with SMT enabled running MSFS. The others look obviously wrong, but similar to each other.

Note that in general, you’ll see only a couple of “big usage” threads in open country, which tend to get flipped betwen two logical cores on SMT – hence it’s common to see four half-loaded cores.

In big cities you’ll see more scenery objects being processed and the other cores start to come up in usage as they do bulk data processing.

I’ve been building computers for many clients for over 20 years. I see (mostly stupid) “Minimum System Requirements” all the time.

I’ve never seen “Don’t build a better system than you need for this application, because it will cause problems.”

I think Microsoft is saying that you need a high end system to run 4K, while a mid range system can run 2K. If all you want is 1080p, a lesser system will suffice.

I have read discussions about people matching high end GPU’s with lower performance CPU’s, and vice versa, and having problems. That kind of makes sense, especially if they’re using an older generation motherboard.

“Oooooo, look at me! I just bought an RTX 4090 to go in my 5-year old $1,000 computer, and I’m…Hey, wait a minute, this isn’t the performance they told me I’d get…” :joy:

1 Like

I think @DensestSnail693 may have nailed it with his BIOS question.

TBH, I don’t remember what the HW Monitor numbers were before I installed MSI’s latest Beta BIOS 2 weeks ago.

1 Like

He meant uninstall it (throw it in the bin, the trashcan). Open Hardware Monitor wasn’t updated from end 2020, it was also my favorite tool before…
Weird CPUID have the same report bug, I trusted it (before today)…

The only hardware monitoring software you need is HWInfo. The rest can be safely thrown in the trash. Task manager is decent for visualizing thread load, but that’s about it - I wouldn’t trust any of the actual numbers.

I use HWInfo, but it’s bugged on my machine with core frequency report. Core usage seems good, not frequency. Task manager have also a buggy CPU frequency report (E.g. 6,47Ghz when I have OC 5Ghz all core).
All those bugs (task manager CPU frequency, HWinfo individual core frequency, etc.) appeared with the launch of Windows 22H2 few months ago, and I’m not the only one seeing those bugs. All was perfect before, and I read somewhere MS acknowledged the problem.

1 Like

How is it bugged, exactly?

In this case I think CPUID have not had time to test with his beta bios and therefore are left in the dark. Although the X3D shares the same base architecture as the 5800X it’s sensory probes will be mapped differently and will use different algorithms.

Here you’ll see the differences with a 4.9Ghz OC and when idle (pictures from few months ago to show the issue).

I realize I’m using an old version, I’ll update and check.

Edit: HWInfo64 fixed with last version 7.40, good :slight_smile: (Task Manager still buggy)

I was about to say this and then saw the edit. The author of HWinfo updates almost weekly, which is one reason why I like it far better than others.

Task manager doesn’t report the actual frequency. I forgot what exactly it reports, but it’s not the real-time frequency as reported by HWInfo. I wouldn’t even bother looking at it.

1 Like

Why not using AMD Ryzen Master, because that utility gave reliable values of the AMD 5800X3D CPU or other AMD CPU’s.

1 Like