Ryzen 7 5800X3D - Brilliant MSFS performance

[Noel]

Single-thread perf on PassMark shows I9-12900K roughly 20-25% faster than 5900X, FWIW.  Despite DX-12 multicore improvements I wonder how much single-threaded performance matters for MSFS going forward.

[MySound]

3 hours ago, Alpine Scenery said:

 

Look people can argue and say whatever they want, but anyone using common sense knows that a CPU using the exact same architecture except more added cache cannot jump another CPU by 50% in MSFS. 

 

 

Does your „common sense“ also accept that some applications heavily benefit from more cache?

Like eg MSFS?!

 

You are telling people here that you don’t want to argue on technical levels that are below yours (lol) but then you also don’t understand how benchmarks and especially 3D Mark works. 
And you are ignoring all the big tech channels who reviewed the 3D and just claim that the 3D can’t be that fast because of „common sense“? Ok 😄

You should maybe take a step back and chill a bit 🙂

Edited July 23, 2022 by MySound

[MySound]

33 minutes ago, Noel said:

Single-thread perf on PassMark shows I9-12900K roughly 20-25% faster than 5900X, FWIW.  Despite DX-12 multicore improvements I wonder how much single-threaded performance matters for MSFS going forward.

As you cannot parallelize a lot of things in programming and you will always have a main thread - I am pretty sure Single Core Performance will still be king for the next years. In some games more, less in others. 
But for MSFS I am quite sure it will be the same for a long time (especially in CPU heavy situations like big airports or big cities). 
 

 

[TravelRunner404]

For what it’s worth I swapped a 5900x for a 5800x3d and it was absolutely 10-15 FPS better. All I did was take the CPUs and swap them in my system. I had a 3090 in the case and did my tests at Ultra 200/200.  Naked sim, except for a few airport add ons and WT G1000.  All marketplace installs.  I posted my results in a couple places.  I was in it for one thing and one thing only…MSFS performance.  Worked for that but my understanding is that for general productivity it might not be the best.  

I am tempted to see if it holds up in DX12 with the better multi core optimization, but swapping CPUs is a bit of a chore for me since I don’t have a test bench or anything.

Edited July 23, 2022 by TravelRunner404

[Noel]

3 hours ago, MySound said:

As you cannot parallelize a lot of things in programming and you will always have a main thread - I am pretty sure Single Core Performance will still be king for the next years. In some games more, less in others. 
But for MSFS I am quite sure it will be the same for a long time (especially in CPU heavy situations like big airports or big cities). 

Is it then safe to conclude I9-12900K is more capable of not main-thread limiting, than say 5800x3d or 5900X?  For single core performance I9-12900K is around 28% faster than my 9900K at the same clock speed.  That's a lot, but I don't think upgrading right now would make my experience meaningfully different for the $1200 or so involved to do it.  I already have excellent performance and though it would let me increase T-LOD considerably in a double-blind test I don't think I'd do alot better than 50% correct in discerning the difference between say 150 and 200.  Maybe it's the 3440x1440 screen or what have you but it's a bit underwhelming for $1200.

[fogboundturtle]

1 hour ago, Noel said:

Is it then safe to conclude I9-12900K is more capable of not main-thread limiting, than say 5800x3d or 5900X?  For single core performance I9-12900K is around 28% faster than my 9900K at the same clock speed.  That's a lot, but I don't think upgrading right now would make my experience meaningfully different for the $1200 or so involved to do it.  I already have excellent performance and though it would let me increase T-LOD considerably in a double-blind test I don't think I'd do alot better than 50% correct in discerning the difference between say 150 and 200.  Maybe it's the 3440x1440 screen or what have you but it's a bit underwhelming for $1200.

wrong. in MSFS, 5800X3D is king. not even the 12900Ks can approach it. 

[Noel]

2 minutes ago, fogboundturtle said:

wrong. in MSFS, 5800X3D is king. not even the 12900Ks can approach it. 

In my system Terrain LOD is the CPU stressor beyond other variables.  With a very strong GPU, at what Terrain LOD setting will a 5800x3D start to display main-thread limiting, say at EDDF or FT's KLAS in a complex airplane?  This is the only way I could discern a difference as I stay limited to 30fps by vsync.   

[fogboundturtle]

5 minutes ago, Noel said:

In my system Terrain LOD is the CPU stressor beyond other variables.  With a very strong GPU, at what Terrain LOD setting will a 5800x3D start to display main-thread limiting, say at EDDF or FT's KLAS in a complex airplane?  This is the only way I could discern a difference as I stay limited to 30fps by vsync.   

It varies so much depending on which aircraft you are flying and where you are flying. On the ground is not the same as while in the air. It's impossible question to answer. I can see that I am always above 30fps with LOD 200. Even in the Fenix. But if I fly GA, I can probably do LOD 400

[Noel]

52 minutes ago, fogboundturtle said:

It varies so much depending on which aircraft you are flying and where you are flying. On the ground is not the same as while in the air. It's impossible question to answer. I can see that I am always above 30fps with LOD 200. Even in the Fenix. But if I fly GA, I can probably do LOD 400

I can easily do T-LOD 400 for GA in modest airports, and I have a 9900K, already 28% or so slower than 12900K.  I go from 110 at the most complex ARRIVAL airports in the B78X HD or AS CRJ (same airport as DEPARTURE I would be at around 150 at the two scenarios I mentioned), on up to 400 in much less demanding arrival airports.  My average from where I fly in those two planes is around 150. Unless I could sustain at least T-LOD 300 everywhere in any plane it would hardly be worth switching out my mainboard/cpu/ram. 

[Sky_Pilot071]

AMD is king for MSFS.   Keep in mind that the Series X is an AMD architechture.

sp

[Noel]

4 minutes ago, Sky_Pilot071 said:

AMD is king for MSFS.   Keep in mind that the Series X is an AMD architechture.

sp

AMD is king in Xbox--I wonder if there is a connect there?

[Alpine Scenery]

If someone got a 10-15 FPS improvement, maybe, but unless they were previously getting 20fps - 30fps, that is not a 50%+ increase. There are ways to reduce the error when measuring, but first you have to normalize the data and also note the error from one session to the next, and use graphs not self averaging. 

No-one understands how every benchmark works, even the people that wrote the code in the benchmark don't fully understand it, it's a lot of fiddling. I don't even fully understand the code I wrote last week. I understand some code in how benchmarks are made, not all code, because I don't know everyone's code. All I am saying is, I've never seen anything like this before where it defies the main benchmarks backwards and then leaps ahead that far the reverse direction, and everything tells me the data is wrong, so unless I have proof, I'm not going to believe it.

This is worse than trying to randomly ask different people questions that ate a certain food and now claim there is a 20% chance if you eat carrots + eggrolls you might develop dementia. 

Edited July 23, 2022 by Alpine Scenery

[zecanelao]

2 hours ago, Noel said:

In my system Terrain LOD is the CPU stressor beyond other variables.  With a very strong GPU, at what Terrain LOD setting will a 5800x3D start to display main-thread limiting, say at EDDF or FT's KLAS in a complex airplane?  This is the only way I could discern a difference as I stay limited to 30fps by vsync.   

Quick test at LAX with Fenix A320, 4k, no traffic; 3090.

LOD 200: silk smooth, no main thread limit. 

400: main thread limit all the time . The stutter may be annoying for some. 

300: good balance between the two above. 

[FakeWayne]

I upgraded from a 3700x this weekend. I primarily use the sim in VR.

Its been a pretty nice upgrade so far.

I would say more than anything, it’s just overall much more smooth. Frame times seem MUCH better. Overall avg framerate is definitely higher too in a lot of areas. I still see drops into the mid 30’s at low altitude, over some (not all) of the heavy photogrammetry areas I’ve tested. But again, even with the framerate in the 30’s, it is for sure much more smooth due to what I assume are improved frame times.  This is on 100/100 LOD, flying the Milviz 310 (which seems like a heavier fps hitter than most other GA aircraft for some reason).  If I go 200/200 LOD's I see drops into the high 20's.

I have not tested in 2d yet.  I'm wondering if going 200/200 LOD in 2d will have the same type of hit as it does in VR.    

 

[Alpine Scenery]

Yah, the 3700 to 5800x3D is a fairly sizable upgrade, from a 2667 passmark to a 3126 passmark, + the added cache.

That's a different story than the 5800x to 5800x3D. I may upgrade to the 3D later when it gets cheaper, just to see, but I'm not expecting more than 5% to 10%, or 15% absolute max. Would be nice if I'm wrong.

With the 5800x to 5800x3D you are going down in core performance by 10% and then just adding the cache, so it's still an upgrade, but more borderline.

Edited July 25, 2022 by Alpine Scenery