MSFS2024 - 9800x3D or 14900k?

[LGKR]

I have not found any videos or articles comparing performance in MSFS2024 on the 9800x3D or 14900K.

Have anyone found this?

I am in the process of upgrading my hardware, but with this new msfs 2024 I am unsure what to do.

[KL Oo]

I've not seen any benchmarks for 2024 yet, however a quick test I did the other night with my 7950X3D and switching cores from the V-Cache to high frequency is here. Doesn't appear to be any difference in this test in terms of FPS but I do think more testing is needed.

If V-Cache doesn't help 2024 as much (or at all) then that may influence your decision.

 

 

[LRBS]

3 hours ago, LGKR said:

I have not found any videos or articles comparing performance in MSFS2024 on the 9800x3D or 14900K.

Have anyone found this?

I am in the process of upgrading my hardware, but with this new msfs 2024 I am unsure what to do.

Hi there,

I noticed the following in MSFS2020 and MSFS2024: three different CPU computers, all have RTX 4090, 64 GB MEM @ 6000 CLS 30, not OC, and FPS locked at 85 to prevent spikes in temperature when MSFS is loading and clocking over 300 FPS.

Over NYC at 1,500FT in PMDG, IFLY, and the Airbus in MSFS2020:

 

98300X3D at 4K due to V Cache in MSFS2020 feels more fluid @ 85 FPS.

 

14900K and KS, you might notice one or two stutters; I could not determine why, and they always occur at different locations.

 

On MSFS2024, due to optimization, with default aircraft, all are performing outstandingly on DLSS.

Now, all will suffer on the ground at JFK, and fps will drop occasionally to 74. Personally, I'm not chasing FPS, only fluidity. The biggest violator of fluidity is AI traffic in our cases.

If you share your PC specifications, many people will be able to give you better advice. If I were you, I would wait for PMDG, IFLY, and Airbus to see how the performance looks.

[Rivanov]

My 9800x3D works great in 2020 and 2024.

[KL Oo]

I did some more testing this afternoon with 2024 at different locations, switching the 7950X3D between V-Cache Cores, non V-Cache cores and all cores together.

Noticed near ZERO difference between each. At this stage, I'm leaning towards 2024 not being effected by the V-Cache.....thinking my next upgrade might be to to a 9950X as opposed to an X3D model. Maybe it will take advantage of the higher clock speed.

 

 

[Sethos]

The X3Ds will still be the best CPUs for MSFS, same engine underneath, same high demand for quick access to cache. Nothing has changed. 

[KL Oo]

2 hours ago, Sethos said:

The X3Ds will still be the best CPUs for MSFS, same engine underneath, same high demand for quick access to cache. Nothing has changed. 

Maybe, but if that was the case I'm not sure why its not showing that in my testing. When I run the same test (swaping from X3D cores to non X3D cores) in 2020 the difference in main thread time is instant and massive. In 2024, its zero. 

[Sethos]

5 minutes ago, KL Oo said:

Maybe, but if that was the case I'm not sure why its not showing that in my testing. When I run the same test (swaping from X3D cores to non X3D cores) in 2020 the difference in main thread time is instant and massive. In 2024, its zero. 

Because you simply cannot do any sort of valid test by sitting in the air, with CPU usage all over the place, no actual benchmark process or software and just switch cores on the fly. That simply does not work or give you a correct result.  Is it, with all due respect, a complete junk benchmark. You want proper benchmarking procedures, a clean environment, established baselines and some actual measurable output to A / B test with. 

Edited November 24, 2024 by Sethos

[KL Oo]

4 minutes ago, Sethos said:

Because you simply cannot do any sort of valid test by sitting in the air, with CPU usage all over the place, no actual benchmark process or software and just switch cores on the fly. That simply does not work or give you a correct result.  Is it, with all due respect, a complete junk benchmark. You want proper benchmarking procedures, a clean environment, established baselines and some actual measurable output to A / B test with. 

Thanks for the constructive  feedback.  I don't recall every saying there was any benchmarking going on. Nor professing to give FPS related results or improvement measurements. My observations were simple - That in 2020 using V-Cache cores significantly improved main thread timing. In 2024, for me, its not making any difference. Not in the half dozen locations I've tried it. 

[Allard]

I gleaned from your video that MSFS had about 10 fps more when on cores 0-7 than on 8-15. I also wonder why you were switching Streamdeck all the time. I don't know whether that was part of your test but it may have influenced the test.

[AnkH]

On 11/24/2024 at 8:04 AM, KL Oo said:

I did some more testing this afternoon with 2024 at different locations, switching the 7950X3D between V-Cache Cores, non V-Cache cores and all cores together.

Noticed near ZERO difference between each. 

No wonder if you test in a location where your GPU runs always at around 98%. You are GPU-limited over the whole 2mins of the video. If you want to see if the V-Caches Cores make a difference, you must use a scenario where the GPU load is below 95%, either because you chose your scenario in accordance (e.g. flying above Manhattan) or by lowering the setings that much, that GPU limit can be basically excluded (e.g. 720p). 

[KL Oo]

That wasn't a particularly polite reply. 

In any event, my 4090 (like lots of peoples) is running at 95% in even the most basic of scenarios. Seems to be a common theme with the new sim that the GPU as there are dozens of threads on the official forums about it.

That said, because I'm happy to spend some time testing ad opposed to criticising, I'll load up in NYC tonight and do a similar test.

[KL Oo]

On 11/27/2024 at 11:50 PM, AnkH said:

No wonder if you test in a location where your GPU runs always at around 98%. You are GPU-limited over the whole 2mins of the video. If you want to see if the V-Caches Cores make a difference, you must use a scenario where the GPU load is below 95%, either because you chose your scenario in accordance (e.g. flying above Manhattan) or by lowering the setings that much, that GPU limit can be basically excluded (e.g. 720p). 

Hello mate. After your very constructive feedback I did as you suggested and loaded up above Manhattan my 7950X3D and 4090 shows as main thread limited.

Again, switching between V-Cache and non V-Cache cores showed absolutely zero difference in main thread time or overall FPS. Running the exact same test in 202 gives widely different results where the V-Cahce cores reduce main thread time significantly. 

 

[AnkH]

Now THIS is interesting, thanks for replicating your test over NYC.

[fluffyflops]

On 11/24/2024 at 5:35 AM, Rivanov said:

My 9800x3D works great in 2020 and 2024.

can i ask what overclock you have used please.  

thanks