Ryzen 7 5800X3D - Brilliant MSFS performance

[dsmbell 36]

Hi, been reading this with interest as have recently bought a new PC with 5800X3D and 3080, I have tested my FPS with SMT on and off and it makes no difference on my PC. Interesting that for some it does but for me no difference. Can't understand why?

[Simple B 141]

17 minutes ago, dsmbell said:

Hi, been reading this with interest as have recently bought a new PC with 5800X3D and 3080, I have tested my FPS with SMT on and off and it makes no difference on my PC. Interesting that for some it does but for me no difference. Can't understand why?

Tried SMT off last night, lost about 10fps at altitude, no change in fps on ground. X3d with 6800 non XT 4K TAA ultra graphics DX11 Lods 225 and 200.

Obviously loaded CPU up cores more, with 3 to 4 of the 8 maxing out regularly.

I'll be turning SMT back to auto tonight.

Edited August 5, 2022 by Simple B

[Kanade 15]

The higher the display resolution, the more dominant the GPU load generally becomes.
HT OFF "appears" to double the load because the number of CPUs is apparently halved.
Either way, MSFS2020 will not scale performance properly with too many multi-core CPUs.
Below are the load factors for each core for the same scenario with HT ON and HT OFF.
HT ON
https://forums.flightsimulator.com/uploads/default/original/4X/9/4/b/94b9c52f70d81adfc81732e7ffa83431b3fc9e51.jpeg
HT OFF
https://forums.flightsimulator.com/uploads/default/original/4X/2/8/2/2820fdf536867411856f83fbe5418850bbee802c.jpeg
Assuming the same workload (FPS), the load should appear higher with HT OFF.

SU10 has now been versioned three times since the benchmark was taken, and the load trend has changed each time.
At the time the benchmark was taken it was 1.27.09 and now it is 1.27.13. SU9 may have different trends again.
 

[Virtual-Chris 682]

4 hours ago, dsmbell said:

Hi, been reading this with interest as have recently bought a new PC with 5800X3D and 3080, I have tested my FPS with SMT on and off and it makes no difference on my PC. Interesting that for some it does but for me no difference. Can't understand why?

 

3 hours ago, Simple B said:

Tried SMT off last night, lost about 10fps at altitude, no change in fps on ground. X3d with 6800 non XT 4K TAA ultra graphics DX11 Lods 225 and 200.

Obviously loaded CPU up cores more, with 3 to 4 of the 8 maxing out regularly.

I'll be turning SMT back to auto tonight.

If I understand it correctly, according to @Kanade the gains are very dependent on load. The point of turning off SMT is to ensure that no cores are starved for cache and creating a cache miss - fewer cores competing for that cache means less chance of data starvation.  If you're flying in an area that's not maximizing the cache (under light load) presumably you won't notice any difference. Make sure you test flying around a photogrammetry location as that puts the biggest load on your CPU (that I'm aware of).  In normal flying, I'm GPU bound. Around photogrammetry I'm CPU (main thread) bound.

Edited August 5, 2022 by Virtual-Chris

[Kanade 15]

Supplementary explanation.
The location I used for my measurements is Tokyo, a photogrammetric city included in the WU01; the WU01 is already installed; the WU01 is a "heavy" course, with a lot of buildings.
Also, the flight path is from Haneda Airport to Shinjuku.
It has many buildings and belongs to a relatively "heavy" course in the whole of Japan.
In the first report, the area with low FPS of 80-100 seconds belongs to Shibuya-Shinjuku, and this part is difficult to speed up even with 5800X3D.
If the results do not change with SMTON/OFF, it means that there is no thread contention caused by SMT.

[Kanade 15]

@Virtual-Chris
Yes, I agree with that understanding, SMTOFF may give better performance in scenarios where thread contention in SMT occurs.
In the CPU utilization graph I mentioned earlier, the area where the CPU does not reach 100% during SMTON and 100% during SMTOFF is the Shibuya-Shinjuku area.
Even X3D is overflowing processing objects and causing cache miss hits.
It is difficult to analyze cache hit/miss hit for an external program (MSFS) that does not have its own code, because it is difficult to do so unless the CPU has a specific instruction set. (Some XeonSPs have these instructions).

As additional information, the latency between SMT logical processors of the same processor is L2-dependent, which is a shared resource within the core (about 7ns for Zen3 architecture, which is about 3 times faster and bandwidth compared to 20ns for L3. This is very similar to the case of L3 and memory).

[Kanade 15]

Supplementation.
Add a link about the high load part.
At high loads, the CPU package is drawing more power and the FPS is reduced.
Since the package temperature remains constant and the clock temperature is also trapezoidal, we can infer that the Shibuya-Shinjuku area is reaching its performance limits.

https://forums.flightsimulator.com/uploads/default/original/4X/b/d/4/bd4d559f337c647c40e69ad197b4f35e32310b45.png
 

[Virtual-Chris 682]

So I spent a lot of time tonight tweaking performance and doing test flights to monitor FPS changes. It turns out my NV control panel was all on default (much to my surprise) and after optimizing those settings, I got about 10% FPS boost.

I then re-tested the difference between SMT enabled and disabled and found no difference. I’m a bit surprised because last time I tested, I got 10% better without SMT.  Ugh. I need to retest perhaps in a different area. Although Vancouver has photogrammetry, it’s not that extensive… as say Tokyo. Not sure. Anyway, SMT is back on for now. 

I will try additional testing tomorrow. 

Edited August 18, 2022 by Virtual-Chris

[MrFuzzy 2,118]

49 minutes ago, Virtual-Chris said:

So I spent a lot of time tonight tweaking performance and doing test flights to monitor FPS changes. It turns out my NV control panel was all on default (much to my surprise) and after optimizing those settings, I got about 10% FPS boost.

I then re-tested the difference between SMT enabled and disabled and found no difference. I’m a bit surprised because last time I tested, I got 10% better without SMT.  Ugh. I need to retest perhaps in a different area. Although Vancouver has photogrammetry, it’s not that extensive… as say Tokyo. Not sure. Anyway, SMT is back on for now. 

I will try additional testing tomorrow. 

I never had benefits with HT disabled, only more stuttering (despite a 10 core CPU).

The truth is that MSFS is a veeeery bad benchmark, you can run it in the same conditions 10 times and always get different values.

Also, if all these "tweaks" were worth anything, there would be already a Windows Gaming Edition on the market, with all the youtubers' snake oil embedded. In most cases youtubers just pick a click-bait title, spend just the right amount of time to make the video monetizable to explain the "tweak" and NEVER show before-after results in a controlled environment. 

Do you want more performance? Get a better CPU with a faster single thread performance (Ryzen 5800X3D or Intel 12th Series), 32 GB of fast RAM, a fast GPU, a NVME SSD. Keep your OS clean, minimize background processes, do not bloat your community folder with every addon you find on flightsim.to. Find the right balance between graphics and performance. 

In other words, do what is straightforward to get more performance, do not look into esoteric stuff... it's just a waste of time. 

[MarkW 1,490]

On 4/21/2022 at 2:56 PM, Maxis said:

Level 43 Here on a Wahoo Kickr Core under a 2018 Scott Foil + 42 inch Tv

But ill leave it at that before the mods start asking what's this Zwift thing is about and does it fly ? LOL 

I have tried Zwift several times but just was not for me.  TrainerRoad user here on a Wahoo Kickr Core.  

Edited August 18, 2022 by MarkW

[Virtual-Chris 682]

8 hours ago, MrFuzzy said:

I never had benefits with HT disabled, only more stuttering (despite a 10 core CPU).

The truth is that MSFS is a veeeery bad benchmark, you can run it in the same conditions 10 times and always get different values.

Also, if all these "tweaks" were worth anything, there would be already a Windows Gaming Edition on the market, with all the youtubers' snake oil embedded. In most cases youtubers just pick a click-bait title, spend just the right amount of time to make the video monetizable to explain the "tweak" and NEVER show before-after results in a controlled environment. 

Do you want more performance? Get a better CPU with a faster single thread performance (Ryzen 5800X3D or Intel 12th Series), 32 GB of fast RAM, a fast GPU, a NVME SSD. Keep your OS clean, minimize background processes, do not bloat your community folder with every addon you find on flightsim.to. Find the right balance between graphics and performance. 

In other words, do what is straightforward to get more performance, do not look into esoteric stuff... it's just a waste of time. 

A few things… I agree many tweaks are snake oil but how you setup vsync and gsync can make a noticeable difference on a high end 3000 series card. This is what I changed last night. I also was able to get my 3090 OC working again which is not snake oil. 

No one will argue that HT on/off makes any difference on Intel but if you take note, we’re in a thread in the 5800x3D here which has a unique cache capability. Over the last several pages here, testing has proven that turning SMT off can yield significant benefits under high loads. I saw it myself, but now don’t see it.

You might say that’s just the sim being inconsistent but I disagree that MSFS can’t be benchmarked. If you turn off live weather and AI traffic, test with clear skies at noon with the same plane, same airport, same flight, and pre-fly with the cache on, you can get very consistent results. But change anything, and yeah, your results might vary. 

The sim does have a very interesting dependency on CPU performance that can quickly and constantly shift the limiting factor around between CPU and GPU which I guess you could say proves the point that it can’t be benchmarked, but by paying attention to main thread vs GPU utilization (in addition to just FPS) you can start to get a feel for where improvements and settings make a difference and where they don’t. 

[Kanade 15]

I agree that benchmarking is difficult because each individual has a different PC environment, settings, and time (and physical network location).
I think we all just benchmark because we love flight sims and want to make them better.

I see people on the forums saying that the SMT is no longer effective with the recent SU10 update, so it is possible that they changed something.
DX12 allows the programmer to set any timing for drawing to the screen, so the behavior is different from DX11, where it is left up to the device driver.
Since MSFS is an application that can be used in both DX11 and DX12, it may be that the timing for the main thread to submit to the drawing thread has been adjusted.
 

[Waldo Pepper 1,952]

 AMD just released a driver for Win 11 22H2 users. 

https://www.amd.com/en/support/chipsets/amd-socket-am4/x570

https://www.amd.com/en/support/kb/release-notes/rn-ryzen-chipset-4-08-09-2337

Quote

AMD Ryzen™ Chipset Driver Release Notes 4.08.09.2337

 

Article Number

RN-RYZEN-CHIPSET-4-08-09-2337

Release Highlights

• Window 11 22H2 OS support added

 

[Virtual-Chris 682]

8 minutes ago, Waldo Pepper said:

 AMD just released a driver for Win 11 22H2 users. 

https://www.amd.com/en/support/chipsets/amd-socket-am4/x570

https://www.amd.com/en/support/kb/release-notes/rn-ryzen-chipset-4-08-09-2337

 

Thanks... I wonder what that will do. Things seemed to be running fine already.

[Virtual-Chris 682]

So, the details for the new AMD CPUs have been out for a few days, and they all have less L3 cache than the 5800x3D.  I wonder if any of these new CPUs will unseat the 5800x3D as the best CPU for MSFS.  The new CPUs have much higher clocks, so perhaps that will drive some added performance and they are noted to have a 13% increase in IPC performance which is substantial.  If you spend $800 USD you can get a 7950 with 80MB of cache and high clocks (and of course the added IPC uplift), which will very likely unseat the 5800x3D but then it's a LOT more expensive and it consumes a ton of power which might be a challenge to cool.  It will be interesting to see if the similarly priced 7700x with much smaller L3 can compete with the 5800x3D.