Jump to content

AMD 5800X3D, the new King for flight simulators?


maxsin72

Recommended Posts

21 hours ago, TED said:

I’m more and more impressed with this cpu the more I test. Today for the first time ever I managed 90fps in vr in the a10c from and around batumi without sacrificing much on graphics settings. Running 100% resolution in openXR and some fsr in the toolkit. Stays above 70 with cpu frametimes in single figures most of the time. I’m now gpu bound, which is no bad place to be and no complaint given I’m using a 6900xt. 
The most notable improvements are those times when I’d expect fps to tank, it just doesn’t and remains totally smooth. For me this was often doing low level gun runs in the a10 and flying close to the ai vehicles or firing off mavericks. Now, not a jitter. It’s totally stable. For me it’s a no brainer now and very clear this works extremely well for my setup. 

What settings are you using? I installed my 5800X3D yesterday, and it has improved my CPU Frametime significantly, from around 33ms to 23, but it is still too high to play at 90 fps. The GPU sits at around 11-12 ms (HP Reverb G1 with a 3080). Everything on High.

Link to comment
Share on other sites

1 hour ago, valrond said:

What settings are you using? I installed my 5800X3D yesterday, and it has improved my CPU Frametime significantly, from around 33ms to 23, but it is still too high to play at 90 fps. The GPU sits at around 11-12 ms (HP Reverb G1 with a 3080). Everything on High.

I wouldn't attempt to play 90fps myself. It was more for testing. I lock my fps normally to 45 with radeon chill. I wanted to just see how much overhead I had and get gpu and cpu times.

Settings I use are 100% openxr resolution with fsr on in toolkit.

In game all fairly high with shadows flat. Visibility ultra, textures high.

I would say my more normal fps would be 60-70 if I didn't lock it. 90 was around 3000' over batumi in a10c. It won't maintain that for long. 


Edited by TED
Link to comment
Share on other sites

On 5/1/2022 at 9:42 AM, valrond said:

What settings are you using? I installed my 5800X3D yesterday, and it has improved my CPU Frametime significantly, from around 33ms to 23, but it is still too high to play at 90 fps. The GPU sits at around 11-12 ms (HP Reverb G1 with a 3080). Everything on High.

What speed ram do you have?  Have you locked in your infinity fabric to be at a 1:1 with the ram speed?

AMD CPU's benefit a lot from higher speed ram up to about 3800mhz.  I think hardware conucks covered the benefits of using higher speed ram with the 3d.


Edited by Strong05

5800X3d, 32GB DDR4@3400, 6800 xt, Reverb G2, Gunfighter/TMWH

Link to comment
Share on other sites

10 hours ago, DeltaMike said:

Wonder if this thing might help on the server side.  

Just got the chance to test in mp. Flying the apache in the Low Level Hell server and I have to say it was a huge difference. Previously I always had to dial down the settings and even then would get little stutters. 

Did a run now on the same settings as I use in sp and the fps never dropped. As smooth as it was in sp. Very impressed.


Edited by TED
  • Like 2
Link to comment
Share on other sites

passmark says otherwise:

https://www.cpubenchmark.net/compare/Intel-i9-12900KS-vs-AMD-Ryzen-7-5800X3D/4813vs4823

The AMD cpu gets trounced all around. 

Derek "BoxxMann" Speare

derekspearedesigns.com 25,000+ Gaming Enthusiasts Trust DSD Components to Perform!

i7-11700k 4.9g | RTX3080ti (finally!)| 64gb Ram | 2TB NVME PCIE4| Reverb G1 | CH Pro Throt/Fighterstick Pro | 4 DSD Boxes

Falcon XT/AT/3.0/4.0 | LB2 | DCS | LOMAC

Been Flight Simming Since 1988!

Useful VR settings and tips for DCS HERE

Link to comment
Share on other sites

3 hours ago, DerekSpeare said:

passmark says otherwise:

https://www.cpubenchmark.net/compare/Intel-i9-12900KS-vs-AMD-Ryzen-7-5800X3D/4813vs4823

The AMD cpu gets trounced all around. 

Yeah it’s worse CPU for the PassMark Benchmark or just overall productivity when compared to 12900k

It’s highly likely that it’s current best CPU for DCS (we do not have comparisons with 12900k there) 

And it’s hands down the best for MSFS by a large margin
I believe Star Citizen it's almost 80% faster than the next competitor with the latest patch 

Basically its a complete “trouncing” by 30%+ in Sim type games even when compared to 12900K there

Depends on your use case - if you only cared about sims DCS/MSFS etc... it's easily the best 


Edited by nikoel
  • Like 3
Link to comment
Share on other sites

10 hours ago, DerekSpeare said:

passmark says otherwise:

https://www.cpubenchmark.net/compare/Intel-i9-12900KS-vs-AMD-Ryzen-7-5800X3D/4813vs4823

The AMD cpu gets trounced all around. 

Passmark isn't a game, it's a synthetic benchmark. Synthetic benchmarks can be (if they're not trash) very useful, but they aren't the gold standard for how CPU's will function in games. For that, you have to actually run them in games.

And honestly I don't know why you'd refer to passmark over something like Cinebench.

I'm hoping to see AMD continue with the 3d cache chips going forward, so you get to choose between say a 7800X that has excellent multithreaded performance for rendering workloads etc, and a 7800X3D that loses some multithreaded performance but has significantly better gaming performance.

The rumor is that AM5 may launch without the option for a 3d cache chip, which would be a shame. I'd hold off until they release one if that ends up being the case.

Either way, here's hoping it gets intel to cram more cache in as well.


Edited by MoleUK
  • Like 1
Link to comment
Share on other sites

Yall are killing me.  I'm looking at that little tower cooler in there... should be enough probably maybe.  Meanwhile my PSU is saying, "It can nae take any more Captain...she's gonna BLOW!"  Killing me.  

  • Like 1

Ryzen 5600X (stock), GBX570, 32Gb RAM, AMD 6900XT (reference), G2, WInwing Orion HOTAS, T-flight rudder

Link to comment
Share on other sites

11 hours ago, DerekSpeare said:

passmark says otherwise:

https://www.cpubenchmark.net/compare/Intel-i9-12900KS-vs-AMD-Ryzen-7-5800X3D/4813vs4823

The AMD cpu gets trounced all around. 

Very good for passmark. My experience directly comparing directly back to back my overclocked 3800xt to the 5800x3d in dcs is vastly different and is a very significant improvement especially in mp.

I guess it all depends on how and what you test for. One thing I know for sure is the x3d works extremely well in dcs. 

Link to comment
Share on other sites

11 hours ago, DerekSpeare said:

passmark says otherwise:

https://www.cpubenchmark.net/compare/Intel-i9-12900KS-vs-AMD-Ryzen-7-5800X3D/4813vs4823

The AMD cpu gets trounced all around. 

Passmark does not count anything in gaming benchmark. You must look at specific benchmarks just like this where you can understand that also 12900ks gets trounced all around:

f31284d1f7cb46bb0f169d32f0af646a43b909c8

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

On 5/2/2022 at 8:43 PM, Strong05 said:

AMD CPU's benefit a lot from higher speed ram up to about 3800mhz.

 

Lol, no. At least not in DCS. The difference of running my 3600MHz, CL14 Ram without XMP (that is 2133 MHz), was about 1% less FPS.

In general, High-MHz RAM is vastly overrated. There may be some very special applications, but even there you get diminishing returns above 3.600-3800 Mhz.
You just introduce a source for instabilities.


Edited by Hiob
  • Like 1

"Muß ich denn jedes Mal, wenn ich sauge oder saugblase den Schlauchstecker in die Schlauchnut schieben?"

Link to comment
Share on other sites

On the other hand, such benchmarks in 1080p tend show the best case scenario with high FPS no GPU limit. When playing at higher resolutions the gains are usually not that high, as many times GPU is the limiting factor.  

Here are several test snapshots from MSFS forum when tested at 4k resolution https://forums.flightsimulator.com/t/amd-5800x3d-performance/510937/98

85b45e00f3a7e890bba0fa915ecee73cf9b7fa2f

 

And a good analysis in VR.

 

 

  • Like 1

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

37 minutes ago, some1 said:

On the other hand, such benchmarks in 1080p tend show the best case scenario with high FPS no GPU limit. When playing at higher resolutions the gains are usually not that high, as many times GPU is the limiting factor.  

Here are several test snapshots from MSFS forum when tested at 4k resolution https://forums.flightsimulator.com/t/amd-5800x3d-performance/510937/98

85b45e00f3a7e890bba0fa915ecee73cf9b7fa2f

 

And a good analysis in VR.

 

 

Yeah if you're benchmarking at 4k you're benchmarking the GPU, even a significantly older/slower CPU can keep up with a top end GPU at 4k.

Of course games like DCS can be an exception, where they require high end GPU's for 4k/ultra but also need faster CPU's due to poor optimisation.

And VR is a different thing altogether with the additional CPU overhead.

41 minutes ago, Hiob said:

Lol, no. At least not in DCS. The difference of running my 3600MHz, CL14 Ram without XMP (that is 2133 MHz), was about 1% less FPS.

In general, High-MHz RAM is vastly overrated. There may be some very special applications, but even there you get diminishing returns above 3.600-3800 Mhz.
You just introduce a source for instabilities.

 

I believe CL14 3200mhz DDR4 still competes with very fast DDR5. It's an odd situation, might be a year or two till DDR5 really pulls ahead.


Edited by MoleUK
  • Like 1
Link to comment
Share on other sites

Hello everyone, this is my first post on this forum but i've been following it ever since 2017 when I started flying dcs exclusive in VR.

I would like to know how this cpu does in dcs liberation?

I currently have a 5600X paired with a 6900XT and wondered if an upgrade to the 5800X3D would be worth it?

Here in Belgium the chip is available but costs 500euro's...

Link to comment
Share on other sites

8 hours ago, some1 said:

On the other hand, such benchmarks in 1080p tend show the best case scenario with high FPS no GPU limit. When playing at higher resolutions the gains are usually not that high, as many times GPU is the limiting factor.  

 

85b45e00f3a7e890bba0fa915ecee73cf9b7fa2f

 

 

Your graphic...  assuming I'm understanding it correctly...  clearly shows the opposite of what you say it does yes?

In all the cases where the CPU was the 3900X the limitation is the main thread (CPU) aside from the bottom one and in all cases where the CPU is the 5800x3D the limitation is the GPU...

Also in all cases there the gains were 7% in the case of the one that was ALREADY GPU BOUND...  and the rest are from 13% to 46%...

 

So...  looks like a huge win for the 5800x3D yeah???

 

  • Like 1

Nvidia RTX3080 (HP Reverb), AMD 3800x

Asus Prime X570P, 64GB G-Skill RipJaw 3600

Saitek X-65F and Fanatec Club-Sport Pedals (Using VJoy and Gremlin to remap Throttle and Clutch into a Rudder axis)

Link to comment
Share on other sites

It simply shows a more realistic usage scenario than playing at 1080p. There are places in the simulator where the new CPU helps considerably, there are others where there's not much difference. Even with a 3080 the game hits GPU limit before the CPU can reach its full potential. 

And that's compared to an older CPU generation. Compared to 5800 non-X3D, the difference will be smaller. When considering an upgrade, best to check if your GPU has some headroom, or is it already loaded near the maximum at the settings you normally use. 

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

It appears to my eye that the graphic you provided shows a few things...

If you have enough CPU ALREADY and are ALREADY GPU bound...  you can maybe gain 7% by upgrading your CPU to the 5800x3D.

If you are already CPU bound and your GPU is the one waiting...  You can gain a lot more.  How much depends what GPU you have of course.

  • Thanks 1

Nvidia RTX3080 (HP Reverb), AMD 3800x

Asus Prime X570P, 64GB G-Skill RipJaw 3600

Saitek X-65F and Fanatec Club-Sport Pedals (Using VJoy and Gremlin to remap Throttle and Clutch into a Rudder axis)

Link to comment
Share on other sites

8 minutes ago, some1 said:

It simply shows a more realistic usage scenario than playing at 1080p. There are places in the simulator where the new CPU helps considerably, there are others where there's not much difference. Even with a 3080 the game hits GPU limit before the CPU can reach its full potential. 

And that's compared to an older CPU generation. Compared to 5800 non-X3D, the difference will be smaller. When considering an upgrade, best to check if your GPU has some headroom, or is it already loaded near the maximum at the settings you normally use. 

Well...  OK.  I'll grant you that it's likely that not too many of us are still running 1080p.  At least not the ones with top shelf CPU/GPU combos...

As far as the fact that there are places where it helps considerably and places where it doesn't...  that's pretty much just always true all the time really.

You graphic specifically shows a case where the system was ALREADY GPU limited...  and it that case it gained 7%.  That's phenomenal.

 

Of course the difference compared to a non 3d chip that's faster than whatever it was compared to here would show smaller gains.  Again...  that's kinda just how things work...

But...  Smaller gains than what?  That smallest 13% gain you see here?  Or smaller than the 46% gain?

I don't care.

Say we get 1/2 of the smallest gain there...  so about 7%.  

That's worth it sir.  That's 3FPS at 45FPS.   That's nothing to sneeze at in the VR world.

 

I get that in the GPU world gains of 30-40% per generation are normal...

 

It don't work that way in the single thread CPU world friend.  7% gains (please keep in mind I've gone through great lengths here to be ULTRA conservative with the estimation) is...

 

HUGE.

 

Nvidia RTX3080 (HP Reverb), AMD 3800x

Asus Prime X570P, 64GB G-Skill RipJaw 3600

Saitek X-65F and Fanatec Club-Sport Pedals (Using VJoy and Gremlin to remap Throttle and Clutch into a Rudder axis)

Link to comment
Share on other sites

So realistically, my 3070 that is pretty much saturated already compared to my 3600 is likely to see less benefit.

I do get CPU limited occasionally in SP but it's only for very brief periods. I think I'd see more cost:benefit in moving to a 5600 (180 quid) than a X3D (420 quid).

  • Like 1

Ryzen7 7800X3D / RTX3080ti / 64GB DDR5 4800 / Varjo Aero / Leap Motion / Kinect Headtracking
TM 28" Warthog Deltasim Hotas / DIY Pendular Rudders / DIY Cyclic Maglock Trimmer / DIY Abris / TM TX 599 evo wheel / TM T3PA pro / DIY 7+1+Sequential Shifter / DIY Handbrake / Cobra Clubman Seat
Shoehorned into a 43" x 43" cupboard.

Link to comment
Share on other sites

As I mentioned in an earlier post, the x3d works "very well in my setup". With my 6900xt, I was previously almost always hitting the cpu limit first. So in my case it was a no brainer in order to release more from the gpu. Therefore the benefits are significant. I would suggest unless you are running at least a 3080/3090 or 6900xt then the gains will not so easily be found. 


Edited by TED
  • Like 2
Link to comment
Share on other sites

In VR, 90% of GPU tuning is figuring out how many (virtual) pixels you can drive.  That doesn't change when you run a complex mission.  (That doesn't mean you can run the exact same settings.  But it's close.)  

I don't know that I would get too excited about this chip for SP/DLC.  3600x is enough and 5600x is more than enough 99% of the time

If you're playing Liberation, you can try dusting off an old computer, use that as a server.  That definitely works.  I'll bet the 3D does also.  Interested to see a head to head.  It's clearly doing something.  

As for MP... there's only one benchmark that matters around here. Sounds to me like we have a winnah

CPU times and GPU times are not unrelated, both are gonna go up in MP.  But I don't think any of us are confused about where the bottlenecks are.  


Edited by DeltaMike

Ryzen 5600X (stock), GBX570, 32Gb RAM, AMD 6900XT (reference), G2, WInwing Orion HOTAS, T-flight rudder

Link to comment
Share on other sites

I think you can run Liberation on a dedicated server on another core if you have the mem for it.

Nvidia RTX3080 (HP Reverb), AMD 3800x

Asus Prime X570P, 64GB G-Skill RipJaw 3600

Saitek X-65F and Fanatec Club-Sport Pedals (Using VJoy and Gremlin to remap Throttle and Clutch into a Rudder axis)

Link to comment
Share on other sites

In VR, 90% of GPU tuning is figuring out how many (virtual) pixels you can drive.  That doesn't change when you run a complex mission.  (That doesn't mean you can run the exact same settings.  But it's close.)  
I don't know that I would get too excited about this chip for SP/DLC.  3600x is enough and 5600x is more than enough 99% of the time
If you're playing Liberation, you can try dusting off an old computer, use that as a server.  That definitely works.  I'll bet the 3D does also.  Interested to see a head to head.  It's clearly doing something.  
As for MP... there's only one benchmark that matters around here. Sounds to me like we have a winnah
CPU times and GPU times are not unrelated, both are gonna go up in MP.  But I don't think any of us are confused about where the bottlenecks are.  
That is one side of the equation..the other is the complexity of the calculation needed. Many AI units and many scripts equals to more calculations for that tiny core (waiting for multi-core here)

🖥️ R7-5800X3D 64GB RTX-4090 LG-38GN950  🥽 Meta Quest 3  🕹️ VPForce Rhino FFB, Virpil F-14 (VFX) Grip, F-15EX Throttle, MFG Crosswinds v3, Razer Tartarus V2 💺SpeedMaster Flight Seat, JetSeat

CVW-17_Profile_Background_VF-103.png

Link to comment
Share on other sites

So finally got it installed, and used the fpsVR logging to capture the CPU frame times of the F-14B cold start with the 5800X and 5800X3D.

With the 5800X and DDR4-3600 CAS 16 memory the median CPU frametime was about 9.2ms. The 5800X3D with DDR4-2666 it was 7.8ms, and with the 5800X3D with DDR4-3600 CAS 16, it was 7.2ms.

Frame rates were still bottlenecked by the GPU, but it looks to be providing a lot of additional CPU performance in VR. I also saw similar performance improvements in the CPU limited Il-2 VR benchmark. 

For standard 2D gaming, it does seem to be providing around a 7-12% lift over the 5800X.

I think what we're seeing is the impact of the dual rendering of VR. It seems like having two almost identical render scenes benefit greatly from very large cache sizes, to a degree that other types of games do not. 

Overall, I think us VR users can more than double the GPU power on one of these without becoming CPU limited.

  • Like 4
Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...