Ryzen 5 5600x for DCS VR?

[HoneyBadger]

What do you guys think? Would you say the Ryzen 5 5600x is a good balance in terms of cost to performance for DCS in VR? Or would you recommend something different? My budget isn't incredibly tight but I do want to get the best cost-to-performance ratio I can for DCS VR.

Edited December 28, 2020 by HoneyBadger

[Qiou87]

Depends how much you end up paying for the 5600X. It is only a six core CPU, it should cost around 300€/320$. Due to stock problems it sells for more than that right now.

 

Still, it has one of the best single core performance on the market, and DCS loves that. DCS does not care if you have 8, 10 or 12 cores. It is not to say that it won't use more than one core, but the CPU frametime is limited by single core performance nonetheless because it does not balance the workload on multiple cores efficiently.

 

The 5600X provides similar CPU frametimes in DCS to the best CPUs out there, Intel Core i7/i9 10700/10850/10900K, but does it while using 75W max compared to over 150 or even 250W for the Intel chips. If you find it at a decent price I think it is an incredible gaming CPU and it will do very well in VR. I saw some comparisons with the 3700X in VR in sim racing and the 5600X was much faster, providing more stable fps.

 

Now of course, VR is also very hungry on the GPU. There is no point going for the fastest CPU out there if you are stuck with a slow GPU.

 

My own findings when measuring performance in DCS in VR (Oculus Rift S): my RTX 2070 Super is limiting performance, not the CPU, when I play single player missions. However, when I play online with 20+ players, the CPU becomes the bottleneck about 50% of the time, and sometimes I will get some "spike" in frametime, resulting in a slow-down or "lag", due to my CPU (Ryzen 2600X). The GPU workload in multiplayer is the same, I get the same GPU frametimes, only CPU frametime doubles roughly. Because of that I decided to upgrade to the 5600X, before I can also upgrade my GPU in 2021 (stocks are crazy right now, prices as well, so I am hanging on to my card for now).

 

Above the 5600X, it seems you only get a few MHz more for DCS, since basically the extra cores of the 5800X, 5900X and 5950X won't be utilized by DCS. To me 100 or 200MHz was not worth 150€ or more, so in this regard, the 5600X was better value. Also considering I could play around with PBO2 and get those MHz back with a more agressive boost curve, if I win the silicon lottery...

Edited December 29, 2020 by Qiou87

[Ala12Rv-Tundra]

5600 is the way to go, that´s why they vanished from shelves.

[Thinder]

£299.99 that's how much I paid for mine, I'm skinned for a month but I think it was worth it and I just got an email from my provider saying it's been dispatched, I should get it tomorrow or Thurdsay, then I'll run a test bench...

 

cclonline.com AMD Ryzen 5 5600X 3.7GHz 6 Core (Socket AM4) CPU

Edited December 29, 2020 by Thinder

[VirusAM]

I am running a 5800x (since today)...and performance increase over the old 9900k in VR is remarkable...
I am able to mantain 40fps in situations where before I couldn’t(of course i am talking about scenarios where the cpu is the limiting factor, like mp or missions with many AI).
However I am keeping an eye on 5900x or 5600x stocks....I will return the 5800x if i can find one of those before the end of January

[Raven (Elysian Angel)]

Well, we don't know how the new engine ED is working on is going to scale with more cores. But to make sure I'll be able to run DCS properly while also streaming, I got myself a 5900X.

[VirusAM]

Well, we don't know how the new engine ED is working on is going to scale with more cores. But to make sure I'll be able to run DCS properly while also streaming, I got myself a 5900X.

I would have done the same if it was possible to find it at its stock price.....

[Qiou87]

3 hours ago, Raven (Elysian Angel) said:

Well, we don't know how the new engine ED is working on is going to scale with more cores. But to make sure I'll be able to run DCS properly while also streaming, I got myself a 5900X.

That new engine won't come for years, probably. Based on different interviews, it seems like a daunting task, and one that is far from finished. I wouldn't buy hardware now based on what I think DCS will work best on in 4-5 years, because much newer/faster hardware will exist then.

 

That's not to say that I wouldn't buy the 5900X if I had the cash. But if you are watching your expenses, going for 5600X and putting the difference in the GPU will yield far better results in VR. The difference between 5600X and 5900X in DCS should be minimal at best.

Edited December 29, 2020 by Qiou87

[VirusAM]

In 2d gaming sure....but in VR every ms that allows a lower frametime counts...and the 5800x and 5900x are probably the best in that regard. 5600x is rocking fast, and for its price (if you can find it) is great....but probably putting cash (again only if you find it at his price) on a 5900x gives you lower frametimes...in VR a fraction of a millisecond could mean the difference between 45 and 90 fps...or worse a difference between smooth 45 and lower then it (no ASW/Motion Vector/Motion Smoothing magic)

Edited December 30, 2020 by VirusAM

[VirusAM]

Today I received the last piece of the puzzle (2x 32gb crucial ballistic kit).....

These are the result of a benchmark

The single threading score is particularly impressive (the baseline for 5800x is 3508, for 5900x is 3511 and for 5600x is 3386)

This is done with no overclocking (stock Ryzen Master auto-overclocking) and high performance power profile (balanced keeps some lower number but not by much).

 

Duing the test I saw all cores rising up to > 4,7 and 2 of the cores > 4.9 and a third almost at 4.9 impressive given that official specs says that boosting is 4.7 on a single core.

 

Maybe someone with a 5600x and a 5900x can upload his results in similar conditions to see the real winner in single threading performances (which counts very much for DCS and sims in general).

 

Edited December 30, 2020 by VirusAM

[Thinder]

There is a lot more than the CPU for VR performances in DCS.

 

O.S optimization: Memory usage. Page file. Latency. Storage.

 

System Optimization: Cooling/thermal limits. RAM; with Ryzen 3, 4 X sticks can improve performance by up to 10% compared to 2 X sticks, RAM speed and latency.

 

Headset. I notice a clear improvement of performances overal since I swaped my CV1 for a G2.

 

Since my Ryzen 5 5600X will be delivered to me in a couple of hours, I ran tests playing DCS and 3D Mark, it doesn't look like my CPU is the issue here but the bootleneck is clearly the GPU and it is no slouch, O.Ced in Afterburner at 1606MHz Core Clock and 6500MHz Memory Clock speed, the temperature never reach more than 76° C.

 

According to CPU Mark, compared to VirusAM results, I shouldn't even be able to run DCS in VR with a 3600X, yet I played several training missions and scenarios such as 1vs1, 2vs2 in the Ukraine and Nevada maps and my FPS never went below 40FPS in average, it made for a quite comfortable gaming all along, for the story, I run antialiasing settings a X 8 in NVIDIA Control Panel and  140% (3752 X 3660) settings in Steam VR.

 

I don't think a further upgrade of CPU will do much to improve my FPS, this is a case of uprading for a stronger GPU, so the Ryzen 5 5600X I'm waiting for is just part of a complete Zen 3 upgrade, all AMD with an RX 6800XT in mind and a kit of 32GB (4 x 8GB) G.SKILL Trident Z Neo DDR4 SDRAM DDR4 3600.

 

You can check on the bottleneck all you want, with this combination, the 5600X will cope real well while playing DCS in VR.

 

Since I couldn't find a bootleneck test for the AMD card, I tried with an NVIDIA 3080:

 

Quote

AMD Ryzen 5 5600X (Clock speed at 100%) with NVIDIA GeForce RTX 3080 (Clock speed at 100%) x1 will produce only 0.36% of GPU bottleneck on 2160p/4K resolution.

As you can see the percentage of bootleneck is way below 5% at 4K, which is the resolution I'm after for playing VR, today I'm at 3.35% of GPU bottleneck on 2160p/4K resolution.

 

Now, those are the results of this morning games:

 

Quote

fpsVR Report:
App: DCS HMD: HP Reverb Virtual Reality Headset G20 (90.001 Hz, IPD 68.0)
GPU: NVIDIA GeForce GTX 1080 Ti (27.21.14.6079) CPU: AMD Ryzen 5 3600X 6-Core Processor             
Delivered fps: 43.3  Duration: 66.8min.
GPU Frametimes:
Median: 22 ms
99th percentile: >30 ms
99.9th percentile: >30 ms
frametime <11.1ms(vsync): 0.4%
CPU frametime:
Median: 9.4 ms
99th percentile: 19.8 ms
99.9th percentile: 26.6 ms
frametime <11.1ms(vsync): 71.6%
Reprojection Ratio: 0.0% (for Index/Vive/VivePro headsets only)
Dropped frames: 0 or 0.0% (for Index/Vive/VivePro headsets only)
Max. SteamVR SS: 140%
Render resolution per eye: 3752x3664(by SteamVR settings, Max.) (HMD driver recommended: 3171x3095)

 

So, my conclusion are that focusing on the CPU only is a mistake, if one wants to mainstream while gaiming, then fine, I can uderstand the need they have for a 5900X but other than that, a 5600X is more than enough and more to the point, pitted with a  RX 6800XT, it will most probably offer a better bound than my 1080 Ti 11GB does with the 3600X I am using now.

 

After that (considering that I've been over the whole optimization topics already), the RAM upgrade should help too, in short, you need balanced performances and good bound between the whole system devices, especialy in DCS, even if GPU is the weakest link here, bootlenecks are still what causes the loss of single device performances overall.

Edited December 30, 2020 by Thinder

[Raven (Elysian Angel)]

13 minutes ago, Thinder said:

you need balanced performances and good bound between the whole system devices

I couldn't agree more. I spend quite a lot of time optimising Windows, BIOS, etc.

 

As for the core counts: several people have pointed out the new consoles have 8C/16T, and given the ever rising costs of game development, it makes more and more sense to have as much unification as possible between PC and console architecture to minimise the costs of developing for both platforms. In that sense, going for a 8C+/16T+ CPU in a new PC makes a lot of sense. 
And yes, I have seen the benchmarks for the 5600X and they are very impressive, but who knows what a CP2077 equivalent in 2 or 3 years from now will demand (at maximum settings).
I bought my system (see signature) with an upgrade cycle of at least 4-5 years in mind.

Sure, DCS is my main hobby at the moment but I also occasionally play (preferably open-world) RPGs and I plan on returning to the MMORPG genre one day, and there are several interesting titles on the horizon. I also made my purchase with that in mind...

[Qiou87]

Core count doesn’t actually matter. Previous gen consoles had eight cores as well, but slow and low clocked ones. Actual performance counts for the future of your cpu, meaning it’s all-core performance. People have been misleadingly thinking this way for a while although it has been debunked time and again. Or else someone with a six core Phenom II would still have a competitive gaming CPU today.

[Thinder]

13 minutes ago, Raven (Elysian Angel) said:

I couldn't agree more. I spend quite a lot of time optimising Windows, BIOS, etc.

 

As for the core counts: several people have pointed out the new consoles have 8C/16T, and given the ever rising costs of game development, it makes more and more sense to have as much unification as possible between PC and console architecture to minimise the costs of developing for both platforms. In that sense, going for a 8C+/16T+ CPU in a new PC makes a lot of sense. 
And yes, I have seen the benchmarks for the 5600X and they are very impressive, but who knows what a CP2077 equivalent in 2 or 3 years from now will demand (at maximum settings).
I bought my system (see signature) with an upgrade cycle of at least 4-5 years in mind.

Sure, DCS is my main hobby at the moment but I also occasionally play (preferably open-world) RPGs and I plan on returning to the MMORPG genre one day, and there are several interesting titles on the horizon. I also made my purchase with that in mind...

 

I understand that people have different needs because they engage into different activities and/or style of gaming.

 

In my case, I was looking for a mid-range to high-end upgrade to play DCS in VR, so it will be what I'll get at the end, still I had to compromise because of budget and recent upgrades as is the case for my B450 Gaming Plus, but it matter little  if we consider that AMD is likely to change socket with the next generation of CPUs, plus this board is really good for the price.

 

 

[Qiou87]

3 hours ago, VirusAM said:

In 2d gaming sure....but in VR every ms that allows a lower frametime counts...and the 5800x and 5900x are probably the best in that regard. 5600x is rocking fast, and for its price (if you can find it) is great....but probably putting cash (again only if you find it at his price) on a 5900x gives you lower frametimes...in VR a fraction of a millisecond could mean the difference between 45 and 90 fps...or worse a difference between smooth 45 and lower then it (no ASW/Motion Vector/Motion Smoothing magic)

 

I’m sorry but this is wrong. As long as your GPU is the bottleneck for frame time it doesn’t matter how fast the cpu is. VR with very high def screens requires mostly fast GPUs...

[Thinder]

Just now, Qiou87 said:

Core count doesn’t actually matter. Previous gen consoles had eight cores as well, but slow and low clocked ones. Actual performance counts for the future of your cpu, meaning it’s all-core performance. People have been misleadingly thinking this way for a while although it has been debunked time and again. Or else someone with a six core Phenom II would still have a competitive gaming CPU today.

 

Spot on.

 

That's why the Ryzen 5 5600 are selling like cookies, noteveryone needs to spend more to get little more for gaming.

[VirusAM]

I’m sorry but this is wrong. As long as your GPU is the bottleneck for frame time it doesn’t matter how fast the cpu is. VR with very high def screens requires mostly fast GPUs...

This is absolutely not wrong....
The cpu in a flight simulation like dcs is much more important then the GPU.
Of course if you fly alone over Dubai which has a lot of 3d objects to display, the cpu won’t be stressed, and the GPU will.
In a realistic scenario (complex mission or Multiplayer server) the CPU is much more important. The AI, the scripting, the physics is all handled by the poor processor...which for VR have also to prepare every single frame to the GPU for rendering it.
Of course the GPU matters...no one is denying that (I have an rtx 2080ti not the slowest card in the world).
But go have a read to the 3090/6900xt thread, and you will see a lot of disappointed users that changed GPU to virtually see no differences in DCS.
And thrust me, cpu frame time is equally important to the GPU ones...in an ideal world both need to be under 11.2 ms
Under 22 for reprojection.

 
Spot on.
 
That's why the Ryzen 5 5600 are selling like cookies, noteveryone needs to spend more to get little more for gaming.

Yeah this is ok...but I was not able to find it.

[Qiou87]

1 hour ago, VirusAM said:

 

This is absolutely not wrong....
The cpu in a flight simulation like dcs is much more important then the GPU.
Of course if you fly alone over Dubai which has a lot of 3d objects to display, the cpu won’t be stressed, and the GPU will.
In a realistic scenario (complex mission or Multiplayer server) the CPU is much more important. The AI, the scripting, the physics is all handled by the poor processor...which for VR have also to prepare every single frame to the GPU for rendering it.
Of course the GPU matters...no one is denying that (I have an rtx 2080ti not the slowest card in the world).
But go have a read to the 3090/6900xt thread, and you will see a lot of disappointed users that changed GPU to virtually see no differences in DCS.
And thrust me, cpu frame time is equally important to the GPU ones...in an ideal world both need to be under 11.2 ms
Under 22 for reprojection.

 

I actually measured that for myself, not just taking someone else’s suppositions. I recommend everyone to do it before investing further into unnecessary components, as it is not very difficult to do.

 

I measured cpu and GPU frametimes in VR, in single and multiplayer, both alone and in complex missions or on 32 player servers. I have found that even if I have a pretty average 2600X, it is only bottlenecking 50% of the time in the heavy MP scenarios, and actually almost never in other situations.   The GPU was doing most of the work because of the high resolution required for the headset.
 

We are here talking about a good value CPU for VR, not how much you can possibly spend. So talking about 1500€/$ GPUs is not really the focus, those propose absolutely horrible value. Of course if you have the fastest GPU on earth you can find yourself in situations where the cpu becomes the focus, but I resent the fact that every discussion about DCS and VR should only focus on brand new 4000€/$ computers. There are people who play and enjoy VR whilst spending their more limited budget efficiently and they do just fine. I speak from experience.

 

Once you get your cpu below your target frametime (11 or 22ms), it does not matter how much faster the cpu gets. And if you only have a certain amount to invest into your PC, including for DCS in VR, spending more into the GPU will yield higher gains than into the CPU.

Edited December 30, 2020 by Qiou87

[VirusAM]

I actually measured that for myself, not just taking someone else’s suppositions. I recommend everyone to do it before investing further into unnecessary components, as it is not very difficult to do.
 
I measured cpu and GPU frametimes in VR, in single and multiplayer, both alone and in complex missions or on 32 player servers. I have found that even if I have a pretty average 2600X, it is only bottlenecking 50% of the time in the heavy MP scenarios, and actually almost never in other situations.   The GPU was doing most of the work because of the high resolution required for the headset.
 
We are here talking about a good value CPU for VR, not how much you can possibly spend. So talking about 1500€/$ GPUs is not really the focus, those propose absolutely horrible value. Of course if you have the fastest GPU on earth you can find yourself in situations where the cpu becomes the focus, but I resent the fact that every discussion about DCS and VR should only focus on brand new 4000€/$ computers. There are people who play and enjoy VR whilst spending their more limited budget efficiently and they do just fine. I speak from experience.
 
Once you get your cpu below your target frametime (11 or 22ms), it does not matter how much faster the cpu gets. And if you only have a certain amount to invest into your PC, including for DCS in VR, spending more into the GPU will yield higher gains than into the CPU.

Yeah I get that...and I mostly agree.
But...I repeat myself....i test DCS with very extreme situations (lot of scripting and AI, dynamic multiplayer missions) and those scenarios are just unplayable in VR, and (read my signature) I have a quite powerful rig...
Now going from an i9 9900k to an r7 5800x I experienced 40fps in situations where before I simply couldn’t...and that was without changing GPU, and 80fps where before I had 40 because of motion smoothing. So the CPU matters a lot.
In the past I measured this with fpsVR, I still didn’t do that with the new CPU.
So probably on your average SP mission campaigns you won’t have cpu bottlenecking issues...
But as mentioned I love to test extreme situations (look at my forum tag ), and in this situation the bottleneck will be the cpu (unless dcs becomes a fully multithreaded application).
Then I am not saying that you should spend all your money on the best possible CPU and you can happily have a gtx970 (just as an example) and experience good frames in VR.
The components in a gaming Pc needs to be balanced.
So I am only saying to don’t chase these new cards (3090/6900xt) at the prices those are selling now.
Anyway a cheap 5600x will be better then any Intel CPU in DCS and you can live with that.

[Thinder]

So OK. I got my Ryzen 5 5600X a couple of hours ago, I had some trouble getting the O.S to start because of changes of settings in BIOS it kept returning to it but after flashing CMOS it ran OK.

 

I ran diverse test benches and figured one thing: With equal settings in NVIDIA Control panel Ryzen Master and Afterburner, my GPU seems to run a little slower with 3D Mark.

 

Score with the Ryzen 5 3600X was 6.767 IN Fire Strike Ultra.

 

6.767 IN Fire Strike Ultra (V1.1) Graphics score 7.375, 41.20 FPS and 26.25 FPS for Graphic tests 1 and 2.

But physics score was 7.700 and Physics test is 24.45 FPS with a Combined score of 3.761 and Combined test at 17.50 FPS.

 

With the Ryzen 5 5600X, same settings, I got 7.341 Graphics score (-0.034), 41.55 FPS and 25.91 FPS for Graphic tests 1 and for Graphic test 2 which is slower by 0.34.

 

Physics score is 9.734, 2.034 higher, Physics test is 30.90 FPS, 6.45 FPS higher, Combined score is 3.672, 0.089 lower and Combined test 17.08 FPS, 0.42 FPS lower.

 

Visibly the bound between the CPU and GPU creates a larger percentage of GPU bottleneck than was the case with the 3600X.

Settings for the test bench in 3D Mark was 3840 X 2160, MSAA X2.

Settings for both CPUs in Ryzen Master are identical in term of boost, NO manual O.C involved.

 

Something interesting though, I ran a  test in single thread with CPU-Z ID and ended up with this...

 

Edited December 30, 2020 by Thinder

[Qiou87]

1 hour ago, VirusAM said:

The components in a gaming Pc needs to be balanced.
So I am only saying to don’t chase these new cards (3090/6900xt) at the prices those are selling now.
Anyway a cheap 5600x will be better then any Intel CPU in DCS and you can live with that.

 

It seems we agree then: balance is key, the single-threaded performance matters a lot in DCS, and also the highest tier cards are not really worth the extra money especially at the prices we are seeing now. 

 

I am just curious if someone will manage to get us a comparison of 5600X/5800X/5950X in DCS. My assumption would be that the only performance difference will be that of the maximum core clock (making sure all else, especially the memory, is equal), so very minimal, because as we all agree the game engine is not balancing load on multiple cores efficiently. And that difference could even be somewhat aliviated with some overclock of the 5600X, maybe. At least, that's my take on it, and why I think the 5600X is a "smart" buy if you are watching your budget and want to save some money for a higher-end GPU.

[VirusAM]

It seems we agree then: balance is key, the single-threaded performance matters a lot in DCS, and also the highest tier cards are not really worth the extra money especially at the prices we are seeing now. 
 
I am just curious if someone will manage to get us a comparison of 5600X/5800X/5950X in DCS. My assumption would be that the only performance difference will be that of the maximum core clock (making sure all else, especially the memory, is equal), so very minimal, because as we all agree the game engine is not balancing load on multiple cores efficiently. And that difference could even be somewhat aliviated with some overclock of the 5600X, maybe. At least, that's my take on it, and why I think the 5600X is a "smart" buy if you are watching your budget and want to save some money for a higher-end GPU.

Of course the 5600x is a smart buy...best bucks for money considering it is better (for gaming) then the i9 10900k
About the comparison...that is why I posted my passmark score...
Hoping that someone with a 5600x and a 5900x will do the same.
I am still in the Amazon return window...so if I can find a 5600x or 5900x at their legit price I could think about switching

There is a lot more than the CPU for VR performances in DCS.
 
O.S optimization: Memory usage. Page file. Latency. Storage.
 
System Optimization: Cooling/thermal limits. RAM; with Ryzen 3, 4 X sticks can improve performance by up to 10% compared to 2 X sticks, RAM speed and latency.
 
Headset. I notice a clear improvement of performances overal since I swaped my CV1 for a G2.
 
Since my Ryzen 5 5600X will be delivered to me in a couple of hours, I ran tests playing DCS and 3D Mark, it doesn't look like my CPU is the issue here but the bootleneck is clearly the GPU and it is no slouch, O.Ced in Afterburner at 1606MHz Core Clock and 6500MHz Memory Clock speed, the temperature never reach more than 76° C.
 
According to CPU Mark, compared to VirusAM results, I shouldn't even be able to run DCS in VR with a 3600X, yet I played several training missions and scenarios such as 1vs1, 2vs2 in the Ukraine and Nevada maps and my FPS never went below 40FPS in average, it made for a quite comfortable gaming all along, for the story, I run antialiasing settings a X 8 in NVIDIA Control Panel and  140% (3752 X 3660) settings in Steam VR.
 
I don't think a further upgrade of CPU will do much to improve my FPS, this is a case of uprading for a stronger GPU, so the Ryzen 5 5600X I'm waiting for is just part of a complete Zen 3 upgrade, all AMD with an RX 6800XT in mind and a kit of 32GB (4 x 8GB) G.SKILL Trident Z Neo DDR4 SDRAM DDR4 3600.
 
You can check on the bottleneck all you want, with this combination, the 5600X will cope real well while playing DCS in VR.
 
Since I couldn't find a bootleneck test for the AMD card, I tried with an NVIDIA 3080:
 

AMD Ryzen 5 5600X (Clock speed at 100%) with NVIDIA GeForce RTX 3080 (Clock speed at 100%) x1 will produce only 0.36% of GPU bottleneck on 2160p/4K resolution.

As you can see the percentage of bootleneck is way below 5% at 4K, which is the resolution I'm after for playing VR, today I'm at 3.35% of GPU bottleneck on 2160p/4K resolution.
 
Now, those are the results of this morning games:
 

fpsVR Report:
App: DCS HMD: HP Reverb Virtual Reality Headset G20 (90.001 Hz, IPD 68.0)
GPU: NVIDIA GeForce GTX 1080 Ti (27.21.14.6079) CPU: AMD Ryzen 5 3600X 6-Core Processor             
Delivered fps: 43.3  Duration: 66.8min.
GPU Frametimes:
Median: 22 ms
99th percentile: >30 ms
99.9th percentile: >30 ms
frametime CPU frametime:
Median: 9.4 ms
99th percentile: 19.8 ms
99.9th percentile: 26.6 ms
frametime Reprojection Ratio: 0.0% (for Index/Vive/VivePro headsets only)
Dropped frames: 0 or 0.0% (for Index/Vive/VivePro headsets only)
Max. SteamVR SS: 140%
Render resolution per eye: 3752x3664(by SteamVR settings, Max.) (HMD driver recommended: 3171x3095)

 
So, my conclusion are that focusing on the CPU only is a mistake, if one wants to mainstream while gaiming, then fine, I can uderstand the need they have for a 5900X but other than that, a 5600X is more than enough and more to the point, pitted with a  RX 6800XT, it will most probably offer a better bound than my 1080 Ti 11GB does with the 3600X I am using now.
 
After that (considering that I've been over the whole optimization topics already), the RAM upgrade should help too, in short, you need balanced performances and good bound between the whole system devices, especialy in DCS, even if GPU is the weakest link here, bootlenecks are still what causes the loss of single device performances overall.


Well simple 2v2 or 4v4 are not the scenarios where a CPU can be stressed in dcs....
Do you have the Hornet?
Try the stock 1989 cas mission included with the module...go over waypoint 3 and tell me what happens...

Another example? Try the p47 fw190a8 furball instant action in the channel map....
That are scenarios where the CPU can be quite stressed (in VR).

And of course I repeat myself, i am not saying that the GPU doesn’t count...
It counts a lot...but for my system changing cpu made more sense, and the results for now are encouraging.

[Thinder]

22 minutes ago, Qiou87 said:

It seems we agree then: balance is key, the single-threaded performance matters a lot in DCS, and also the highest tier cards are not really worth the extra money especially at the prices we are seeing now. 

 

I am just curious if someone will manage to get us a comparison of 5600X/5800X/5950X in DCS. My assumption would be that the only performance difference will be that of the maximum core clock (making sure all else, especially the memory, is equal), so very minimal, because as we all agree the game engine is not balancing load on multiple cores efficiently. And that difference could even be somewhat aliviated with some overclock of the 5600X, maybe. At least, that's my take on it, and why I think the 5600X is a "smart" buy if you are watching your budget and want to save some money for a higher-end GPU.

 

Well, if you're planning to pit it with a stronger GPU than my 1080 Ti 11GB it's a hell of a processor, looking at the results I got running 3D Mark it shows nearly 21% gain in Physics Test and 6.45 FPS higher, I can't argue with that.

 

The thing is, Ryzen 3 architecture can be used to improve at all levels, a good GPU and RAM can further improve the performances of the whole system, but if one component is weaker, the bottleneck shows, not by much in my case because visibly the 5600X is way faster in single core with equivalent auto O.C settings than the 3600X, as the physics test demonstrate.

 

Note that I use exactly the same cooler than with the 3600X, an Artic Freezer 7 X.

 

17 minutes ago, VirusAM said:

Well simple 2v2 or 4v4 are not the scenarios where a CPU can be stressed in dcs....
Do you have the Hornet?

 

My tests were done playing the F-18C.

 

Now, I invite you to have a look at the CPU-Z ID score of my 5600X in single thread, I don't think it will be the issue in more complicated scenarios and in any case it never was the CPU which was the issue, even with the 3600X, yet no one can claim that a GEFORCE GTX 1080 Ti 11GB is a weak card even by today standards...

Edited December 30, 2020 by Thinder

[Thinder]

Now, let's mitigate those performances a litte and replace them into their context:

 

When I started to upgrade, I needed a new motherboard, RAM and GPU in a hurry because of failure, I was running an Athlon 4000 and my GPU was falling appart, I had to screw a fan on its sink to keep it cool because the stock fans were dead, second hand and completely worn out after years of abuse I must say.

 

I upgraded the GPU twice already this year just to keep up with the demands for VR but I chosed to keep my motherboard because it is not only very good for the price, but also it is likely that AMD will change socket pretty soon, and I'm on a budget, so no hurry to buy a fully new system here, I'm aiming for a mid/high range of system and try to keep cost as low as possible.

 

The MSI B450 GAMING PLUS MAX doesn't support the faster Gen 3 or 4 PCI so its performances are lower when it comes to GPU buses bandwidth, it's a compromise I accepted to make in full knowledge of what I was getting into, and I got my 1080 Ti second hand from CEX with a two years guaranty, just as new as if it came straight out of the box.

 

Quote

• PCI-Express 2.0 sends 500 MB/s per direction
• PCI-Express 3.0 brings that number towards a whopping 1 GB/s.
• https://www.guru3d.com/articles-pages/pci-express-scaling-game-performance-analysis-review,2.html

 

But this doesn't imply the CPU itself nor the RAM, it's a Dual channel, can take 4133 MHz RAM with O.C mode and I'm not planning to get this kind of speed from my RAM, what I am looking for is (again) a ballanced system with the lowest percentage of bottleneck between the 5600X and my future GPU and RAM.

 

So in this regard, the question as if the 5600X is adequate for playing DCS (even in complicated scenarios when pited with a good GPU) is yes, it was designed for gaming at these levels and I'm quite sure that an RX 6800 XT and a kit of 4 X 8GB of DDR4 3600 will allow me to increase graphic settings and gain in FPS, this processor is good enough to achieve that.

 

From the tests I have seen, a 10% improvement in FPS can be obtained by using 4 sticks of RAM, the DDR4 seems to be the sweet spot and there is little gain in comparison by using much faster RAM or more than 32 GB, that's what the Zen 3 architecture allow for, and I'm sure there can be more performances to come with new BIOS and graphic drivers in the future.

 

 

 

 

Edited December 30, 2020 by Thinder

[VirusAM]

 
My tests were done playing the F-18C.
 
Now, I invite you to have a look at the CPU-Z ID score of my 5600X in single thread, I don't think it will be the issue in more complicated scenarios and in any case it never was the CPU which was the issue, even with the 3600X, yet no one can claim that a GEFORCE GTX 1080 Ti 11GB is a weak card even by today standards...

Yeah...
Not arguing on any of that...
Just try the two missions I mentioned in my previous post...especially the Hornet one can bring the most powerful pc to its knees