SYN_Vander BENCHMARK v6 to measure IL-2 performance in monitor & VR

[DBCOOPER011]

Here's the best I can get with the newest IL2.

 

Motherboard: EVGA Z690 Dark
CPU:         12900K
CPU Freq:    Cores 2x55,4x54,6x53,7x52,8x51
Cores:       8
Threads:     8
RAM size:    32Gb (2x16GB)
RAM Freq:    6800 MHz
RAM timings: 32-39-39-80 1T
GPU:         3090

 

CPU Test: E-cores/Hyperthreading off, ring 51, 1 core at 5.6Mhz

2022-09-30 09:23:09 - Il-2
Frames: 9029 - Time: 60000ms - Avg: 150.483 - Min: 132 - Max: 211

2022-09-30 09:29:39 - Il-2
Frames: 8992 - Time: 60000ms - Avg: 149.867 - Min: 132 - Max: 205

2022-09-30 09:35:47 - Il-2
Frames: 9024 - Time: 60000ms - Avg: 150.400 - Min: 134 - Max: 205

Daily game setting. E-cores and hyperthreading off, ring 50, 2 cores at 5.5Mhz

 

2022-09-30 09:53:05 - Il-2
Frames: 8750 - Time: 60000ms - Avg: 145.833 - Min: 130 - Max: 202

VR2 Test: 130Mhz core, Mem at 1000Mhz, OpenXR W/O Fsr/Nis, HP G2 at 3162:

 

1 CPU core at 5.6Mhz

2022-09-30 14:04:28 - Il-2
Frames: 5314 - Time: 60000ms - Avg: 88.567 - Min: 77 - Max: 91

 

2 CPU cores at 5.5Mhz

2022-09-30 14:12:27 - Il-2
Frames: 5294 - Time: 60000ms - Avg: 88.233 - Min: 74 - Max: 91

Edited September 30, 2022 by DBCOOPER011

[=SFG=BoostedStig]

System Specs

Motherboard:  ASUS TUF Gaming X570 Plus (Wi-Fi)
CPU:                AMD Ryzen 9 5950X
CPU Freq:        5000mhz (This moves around a lot. I've seen it boost as high as 5150mhz during gaming)
Cores:              16
Threads:          32
RAM size:        32Gb (4x8GB)
RAM Freq:       3800 MHz
RAM timings:  16-16-16-35 1T
GPU:                3080Ti

 

Windows 10

IL-2 v5.001

 

CPU Test:

Frames: 7516 - Time: 60000ms - Avg: 125.267 - Min: 106 - Max: 188

Frames: 7444 - Time: 60000ms - Avg: 124.067 - Min: 105 - Max: 176

Frames: 7333 - Time: 60000ms - Avg: 122.217 - Min: 105 - Max: 184

 

GPU Test 4K:

Frames: 8695 - Time: 60000ms - Avg: 144.917 - Min: 102 - Max: 216

Frames: 8669 - Time: 60000ms - Avg: 144.483 - Min: 101 - Max: 209
Frames: 8664 - Time: 60000ms - Avg: 144.400 - Min: 103 - Max: 206

Edited October 2, 2022 by =SFG=BoostedStig

[chiliwili69]

On 9/30/2022 at 11:47 PM, DBCOOPER011 said:

1 CPU core at 5.6Mhz

2022-09-30 14:04:28 - Il-2
Frames: 5314 - Time: 60000ms - Avg: 88.567 - Min: 77 - Max: 91

 

Many thanks for re-runing the test with the new IL-2 release. We know now where it is our top performer CPU.

 

Being almost at 90fps in your VRTEst2, it shows that the settings defined in these VR tests would be too low for the new CPUs arriving this year. So perhaps it is time to raise the bar of the settings of the VR tests if we want to have measurable differences.

9 hours ago, =SFG=BoostedStig said:

Motherboard:  ASUS TUF Gaming X570 Plus (Wi-Fi)
CPU:                AMD Ryzen 9 5950X
CPU Freq:        5000mhz (This moves around a lot. I've seen it boost as high as 5150mhz during gaming)

 

Thank you for your test.

For the 4K test you got exactly the same numbers than @Firdimigdi

[=SFG=BoostedStig]

UPDATE: Changed from 5950X to 5800X3D

 

System Specs

Motherboard:  ASUS TUF Gaming X570 Plus (Wi-Fi)
CPU:                AMD Ryzen 7 5800X3D
CPU Freq:        4500mhz
Cores:              8
Threads:          16
RAM size:        32Gb (4x8GB)
RAM Freq:       3800 MHz
RAM timings:  16-16-16-35 1T
GPU:                3080Ti

 

Windows 10

IL-2 v5.001

 

CPU Test:

Frames: 7759 - Time: 60000ms - Avg: 129.317 - Min: 110 - Max: 181

Frames: 7812 - Time: 60000ms - Avg: 130.200 - Min: 114 - Max: 181

Frames: 7845 - Time: 60000ms - Avg: 130.750 - Min: 117 - Max: 182

Edited October 3, 2022 by =SFG=BoostedStig

[chiliwili69]

5 hours ago, =SFG=BoostedStig said:

Changed from 5950X to 5800X3D

Thank you for running again the bench after the upgrade.

Here you went from a 64MB L3 cache (which is already a nice cache) to a 96MB cache, so the peformance increase was lower than other people going from 32MB to 98MB.

[=SFG=BoostedStig]

1 hour ago, chiliwili69 said:

Thank you for running again the bench after the upgrade.

Here you went from a 64MB L3 cache (which is already a nice cache) to a 96MB cache, so the peformance increase was lower than other people going from 32MB to 98MB.

Yes, unfortunately I didn't test IL-2 on the 5950X in VR which seems to have been a mistake for data collection. In games that I play more regularly, like iRacing, I saw similar improvements on monitor, in the 5% range. But in VR, on iRacing, I saw 20-25% improvements, depending on settings. I really wish I would have done the VR bench ahead of time, but here they are as I've just done them.

 

System Specs

Motherboard:  ASUS TUF Gaming X570 Plus (Wi-Fi)
CPU:                AMD Ryzen 7 5800X3D
CPU Freq:        4500mhz
Cores:              8
Threads:          16
RAM size:        32Gb (4x8GB)
RAM Freq:       3800 MHz
RAM timings:  16-16-16-35 1T
GPU:                3080Ti

 

Windows 10

IL-2 v5.001

 

HP Reverb G2

VR Test1

Frames: 5343 - Time: 60000ms - Avg: 89.050 - Min: 76 - Max: 91

Frames: 5315 - Time: 60000ms - Avg: 88.583 - Min: 73 - Max: 91

 

VR Test2

Frames: 5250 - Time: 60000ms - Avg: 87.500 - Min: 71 - Max: 91

Frames: 5243 - Time: 60000ms - Avg: 87.383 - Min: 72 - Max: 91

[chiliwili69]

11 hours ago, =SFG=BoostedStig said:

Frames: 5343 - Time: 60000ms - Avg: 89.050 - Min: 76 - Max: 91

 

This is very good mark for VRtest1. You should run very smooth with the G2.

[-3RW-ghyslain]

I'm curious how will the Ryzen 7000's will fare against the Intel 13th gen.

[Angry_Bird]

Same. Or at least results 7600X vs 5800X3D

[chiliwili69]

9 hours ago, -3RW-ghyslain said:

I'm curious how will the Ryzen 7000's will fare against the Intel 13th gen.

 

Yes, many people will upgrade PC along the end of this year and they will need to decide.

According to the single-thread the 13th gen will be above Ryzens 7000´s , but until this is tested here we don´t really know.

In any case, any of the new CPUs will be plenty for the monitor or VR needs right now.

[-3RW-ghyslain]

14 hours ago, chiliwili69 said:

 

Yes, many people will upgrade PC along the end of this year and they will need to decide.

According to the single-thread the 13th gen will be above Ryzens 7000´s , but until this is tested here we don´t really know.

In any case, any of the new CPUs will be plenty for the monitor or VR needs right now.

At last indeed, new cpu's will finally be able to keep the fps steady. I'm personally running a Pimax 8KX at 75hz currently, off of a R9 5900x // RTX 3080 TI rig. At 0.5 render scale resolution, I can keep the fps up, with some stutters in MP servers when over busy areas (most likely cpu bottlenecked). Looking forward to upgrade to a 13900k with a 4090 and hopefully be able to get butter smooth FPS with 0.75 render scale. I still won't quite be able to crack 1.0 render scale, due to a gpu bottleneck on even the 4090 (anticipated). Crazy the amount of pixels the Pimax 8KX sports!

[Capt_Hook]

Thanks to all contributing to this thread - the cost/benefit calculation in IL2 VR seems pretty tricky.  I'm running a 3090 with a 3900X CPU, and was going to replace it all late this year with a 7900X / 4090 system (or equivalent Intel if a better deal), but now I'm wondering if a much simpler 5800X3D drop in to the existing system along with the 4090 (when I can get my hands on one) would yield similar results.  This is driving a Reverb G2.  

[Alonzo]

39 minutes ago, Capt_Hook said:

Thanks to all contributing to this thread - the cost/benefit calculation in IL2 VR seems pretty tricky.  I'm running a 3090 with a 3900X CPU, and was going to replace it all late this year with a 7900X / 4090 system (or equivalent Intel if a better deal), but now I'm wondering if a much simpler 5800X3D drop in to the existing system along with the 4090 (when I can get my hands on one) would yield similar results.  This is driving a Reverb G2.  

 

I think you'd see a decent boost for the $400 it costs to go 3900X -> 5800X3D, depending on your RAM speed also. I have a 5800X and am even considering the X3D, because the platform costs of switching to DDR5 are pretty hefty. I have a 3080 driving also a Reverb G2. I think my bottlenecks are still CPU related.

[chiliwili69]

On 10/17/2022 at 2:18 AM, Alonzo said:

I think you'd see a decent boost for the $400 it costs to go 3900X -> 5800X3D, depending on your RAM speed also.

 

Yes, I think is worth. There is a big step in VR performance from Zen2 to Zen3. And even also from 5800X to 5800X3D. Voyager was getting +10fps in VR.

[FTC_ChilliBalls]

Hey guys, given the recent price drops in GPU prices, I've been thinking of upgrading to a RX 6800 non-XT.

However, I was told that the RX6000 series is seriously underperforming IL2 generally and in VR particularly.

 

Is that the case or was it at some point solved with driver updates?

 

Best regards 

[chiliwili69]

10 hours ago, So_ein_Feuerball said:

Is that the case or was it at some point solved with driver updates?

 

Latest test in this benchmark the issue was still there. Also, in the VR section there is a thread and thread2 of a guy with issues with a 6900XT. I think all he issues are 6900XT related.

But anyone with a 6800, 6900, 6800Xt, 6900XT can run the bench to confirm.

[1Sascha]

On 10/15/2022 at 11:36 AM, chiliwili69 said:

According to the single-thread the 13th gen will be above Ryzens 7000´s , but until this is tested here we don´t really know.

Sample size is extremely low still, but there are a couple of 13900 bench results up on Passmark.

13900KF, single thread rating: 4844

13900K, single thread rating:  4789

 

For comparison:

AMD Ryzen 9 7950X: 4336

Intel i9 12900KS: 4407

... and because it's the one I use:

i5-12600K: 3981 (average) - 4045 (mine).

 

 

I guess I should probably re-run at least the VR tests now that I'm running the Win11-update, the most recent GF-drivers and IL-2 5.x. I seem to be having more trouble with FPS than before, but only with certain settings and in certain circumstances (well populated career missions with lots of bombers around). So I'm not sure my results will be all that valid or "enlightening".

 

But I'm still not sure what is causing my recent FPS drops and am trying to figure out where the bottleneck is or what setting might be responsible. When everything runs normal in a career mission (steady 80 FPS), my GPU is sitting between 80 and 85% load with my current settings and my CPU isn't really stressed, either.

 

Weird.

 

S.

[FTC_ChilliBalls]

6 hours ago, chiliwili69 said:

 

Latest test in this benchmark the issue was still there. Also, in the VR section there is a thread and thread2 of a guy with issues with a 6900XT. I think all he issues are 6900XT related.

But anyone with a 6800, 6900, 6800Xt, 6900XT can run the bench to confirm.

Too bad, I found an RX 6800 for 580€ and was seriously considering it.

[Voyager]

3 hours ago, So_ein_Feuerball said:

Too bad, I found an RX 6800 for 580€ and was seriously considering it.

Well, at that price, you really need to be comparing it to how the 3070/3070 Ti perform too, not the 3090 numbers. 

 

That said, if you're ok with a used GPU, I'm seeing 3070 Ti's for $450, and 2080 Ti's for under $300. Depending on what you're on now, that could be a pretty solid boost. 

[FTC_ChilliBalls]

19 minutes ago, Voyager said:

Well, at that price, you really need to be comparing it to how the 3070/3070 Ti perform too, not the 3090 numbers. 

 

That said, if you're ok with a used GPU, I'm seeing 3070 Ti's for $450, and 2080 Ti's for under $300. Depending on what you're on now, that could be a pretty solid boost. 

Found an RTX 3070 for 500€ and I would compare it to the 6750XT in that case as that's the same price I found for that one.

[1Sascha]

Just re-ran 1080p, 4k (in 1440p) and VR 1 - because Windows 11 22H2/GRDriver 522.25/IL-2 5.002.

 

My Specs:
 Motherboard: MSI MAG Z690 Tomahawk Wifi DDR4
 CPU: Intel i5-12600K, slightly OCed via BIOS turbo-offset (+1) on the P- and E-cores, cooled by a Lian Li Galahad 240
 CPU Freq: P-cores max 1080 run: 4.98 GHz, P-cores max in VR test1: 4.78 GHz
 L3 cache: 20 MB
 Cores: 6P + 4E = 10
 Threads: 16
 RAM type: Corsair Vengeance Pro RGB, 3600 DDR4 (XMP is on)
 RAM size:  2x8 GB, dual channel
 NB Freq: 3600 MHz
 RAM Freq: 3600 MHz
 RAM Latency:  CL18
 GPU: Gigabyte RTX 3070 Eagle OC LHR. Undervolted, GDDR OCed by +800MHz. Effective clock speed: 1925 MHz @0.925V, GDDR: 7800 MHz.

 

 

Both 2D results were virtually identical to my last runs, so I won't put them up again.

 

VR result wasn't all that different, either, but average FPS went up a bit.

Not sure what I had done last time, but this time I made sure Windows was set to highest performance power plan - I usually have it on balanced since it doesn't make a difference in non-VR titles. Plus I closed a bit of background crap like discord, disabled GFE Overlay, etc. before running the benchmark.

 

Old result from August:

Frames: 3191 - Time: 60000ms - Avg: 53.183 - Min: 38 - Max: 81

 

 

 

Current result from today:

2022-10-19 23:20:25 - Il-2
Frames: 3638 - Time: 60000ms - Avg: 60.633 - Min: 39 - Max: 81

 

Slight tangent, but I thought I'd mention this as well:

I accidentally had Fraps Benchmark running still when I had re-configured the game for actual flying (with reduced SS to 120% in SteamVR). I think what ruined my framerates last time must've been me going a bit overboard with the SS (or forgetting to switch it back to reasonable levels after benchmarking, lol!) .... because:

 

2022-10-19 23:35:56 - Il-2
Frames: 4798 - Time: 60000ms - Avg: 79.967 - Min: 78 - Max: 81

2022-10-19 23:43:35 - Il-2
Frames: 4798 - Time: 60000ms - Avg: 79.967 - Min: 78 - Max: 81

 

 

Still... the only config that pushed my GPU close to 100% usage was the 4K-test (run in 1440p due to lack of a 4k monitor). Both the VR1-test and my settings for actual VR-gameplay resulted in GPU usage only slightly over 80%. Even "Battletech" (Unity-based strategy game) on 1440p and Ultra settings is more demanding on my GPU than that... lol.

And yet... despite what the gaming Fraps runs seem to suggest, the missions were naturally much longer than 60 seconds and I did encounter *slight* FPS drops (low 70s) in certain situations. Generally speaking I was at 79/80 FPS for 90, 95% of the time, but there were some stutters in large furballs - Stalingrad summer, 20+ planes on the map, ground activity set to "scattered". Re-checked CPU usage in HWMonitor and it too didn't show any huge spikes or high values in CPU-usage or power-draw.

 

EDIT: Removed the 60 second time limit on FRAPS and benchmarked the same "attacker intercept" career mission on the Stalingrad summer map from start to finish with my gaming settings and 1.22 PD:

 

2022-10-20 19:47:20 - Il-2
Frames: 93151 - Time: 1165828ms - Avg: 79.901 - Min: 59 - Max: 81

 

... and I still can't tell whether I'm CPU or GPU limited with these settings, since I couldn't see any obvious spikes in either CPU or GPU load/power draw/temps. I guess I'd need a different method of monitoring/logging sensors than looking at HWMonitor's max values after the mission.

 

 

 

 

 

S.

 

Edited October 21, 2022 by 1Sascha

[Alonzo]

I currently have an AMD 5800X, with G.Skill 3600-CL16 RAM (I believe it's Samsung, model is F4-3600C16-16GTZN). My impression is that it's CPU speed that's holding me back in VR, rather than GPU. I have a 3080 and a Reverb G2 but I don't crank the graphics.

 

For CAD$500 I could get a 5800X3D, and get a bit more life out of this PC. Should I pull the trigger on an upgrade, or wait and maybe build an entirely new rig, try to sell the old rig? Or something else? I did try to tune my memory timings but couldn't get anything stable, I was using one of those pre-canned tools though, maybe that's not the way forward.

[Voyager]

16 hours ago, Alonzo said:

I currently have an AMD 5800X, with G.Skill 3600-CL16 RAM (I believe it's Samsung, model is F4-3600C16-16GTZN). My impression is that it's CPU speed that's holding me back in VR, rather than GPU. I have a 3080 and a Reverb G2 but I don't crank the graphics.

 

For CAD$500 I could get a 5800X3D, and get a bit more life out of this PC. Should I pull the trigger on an upgrade, or wait and maybe build an entirely new rig, try to sell the old rig? Or something else? I did try to tune my memory timings but couldn't get anything stable, I was using one of those pre-canned tools though, maybe that's not the way forward.

 

I did that earlier this year, and in VR it was about a 15% improvement in average frame rate and a 19% improvement in minimum frame rates. At this point I'm now GPU bound in practical VR. 

 

I can post some FPS VR frame time graphs tonight so you can see what I mean. 

 

I suspect the 5800X3D will match the RTX 4090/RX 7950 XT GPU performance in VR. 

 

A future 7800X3D will probably beat it, and I expect the Raptor Lake and Zen 4 parts will trade blows with it, but the Zen 4 Vcache isn't out, and there isn't a GPU that's going to over match any of the CPUs that are out, so if you're already on AM4, I think the 5800X3D is your best bet, and will keep you in top VR performance for at least until the next generation of graphics cards comes out (ie Blackwell and RDNA4).

[MilitantPotato]

Using a reverb G2 for both tests

Motherboard: X570 AORUS MASTER
 CPU:                 Ryzen 5 5800X
 CPU Freq:        4.9 Ghz
 Cores:               8 
 Threads:           16 
 RAM size:        32Gb (2x16GB)
 RAM Freq:       3600 MHz 
 NB Freq:          1800 MHz 
 RAM timings:  14-15-15-35
 GPU:                 3080 10 gig

 

VR test 2: 
2022-10-17 12:51:25 - Il-2
Frames: 4418 - Time: 60000ms - Avg: 73.633 - Min: 55 - Max: 91

2022-10-17 12:53:53 - Il-2
Frames: 4184 - Time: 60000ms - Avg: 69.733 - Min: 55 - Max: 91

--------------------------------------------------------------------------------
 

Motherboard: X570 AORUS MASTER
 CPU:                 Ryzen 5 5800X
 CPU Freq:        4.9 Ghz
 Cores:               8 
 Threads:           16 
 RAM size:        32Gb (2x16GB)
 RAM Freq:       3600 MHz 
 NB Freq:          1800 MHz 
 RAM timings:  14-15-15-35
 GPU:                 4090

 

VR test 2: 
 

2022-10-20 11:50:09 - Il-2
Frames: 4487 - Time: 60000ms - Avg: 74.783 - Min: 58 - Max: 91

2022-10-20 11:56:19 - Il-2
Frames: 4380 - Time: 60000ms - Avg: 73.000 - Min: 56 - Max: 91


 

My 4090 scores are exactly the same as a 3080. I think we need a 35mp+ benchmark, as even a 4090 is bottlenecked with the current benchmark setup @SYN_Vander

Edited October 20, 2022 by MilitantPotato

[=SFG=BoostedStig]

On 10/16/2022 at 6:36 PM, Capt_Hook said:

Thanks to all contributing to this thread - the cost/benefit calculation in IL2 VR seems pretty tricky.  I'm running a 3090 with a 3900X CPU, and was going to replace it all late this year with a 7900X / 4090 system (or equivalent Intel if a better deal), but now I'm wondering if a much simpler 5800X3D drop in to the existing system along with the 4090 (when I can get my hands on one) would yield similar results.  This is driving a Reverb G2.  

 

I'll hopefully be testing this combination mid next week as I was able to score a 4090 from a drop earlier today.

[MilitantPotato]

1 hour ago, =SFG=BoostedStig said:

 

I'll hopefully be testing this combination mid next week as I was able to score a 4090 from a drop earlier today.

I saw zero change from a 3080 to a 4090 on a 5800x for VR test 2. Completely CPU limited ?  Looking at your results I may snag a 5800X3D myself.  Hopefully 7000x3d cpus will be 30% better than a 5800x3d, cause it looks like it's needed for steady 90fps.

[Alonzo]

8 hours ago, Voyager said:

I did that earlier this year, and in VR it was about a 15% improvement in average frame rate and a 19% improvement in minimum frame rates. At this point I'm now GPU bound in practical VR. 

 

I can post some FPS VR frame time graphs tonight so you can see what I mean. 

 

I suspect the 5800X3D will match the RTX 4090/RX 7950 XT GPU performance in VR. 

 

When you say "practical VR" would you include multiplayer in that? That's my use-case and I've found things can get bogged down a lot in multiplayer, but it's difficult to benchmark or really know how far off I am from hitting the framerate.

[Voyager]

4 hours ago, Alonzo said:

 

When you say "practical VR" would you include multiplayer in that? That's my use-case and I've found things can get bogged down a lot in multiplayer, but it's difficult to benchmark or really know how far off I am from hitting the framerate.

 

Unfortunately I don't really do multiplayer, but what you can do is get FpsVR, and turn on its logging. That will give you a distribution graph of your frame times so you can see where your CPU and GPU are performing.

[MilitantPotato]

14 minutes ago, Voyager said:

 

Unfortunately I don't really do multiplayer, but what you can do is get FpsVR, and turn on its logging. That will give you a distribution graph of your frame times so you can see where your CPU and GPU are performing.

Il-2 still slows down game speed when the CPU is overwhelmed in single player yea?
I think replaying a replay with lots of planes and ground clutter is a better way to see if your CPU is managing OK, maybe.

 

Edited October 21, 2022 by MilitantPotato

[chiliwili69]

On 10/19/2022 at 11:29 PM, 1Sascha said:

(with reduced SS to 120% in SteamVR).

 

Thanks for your tests.

The VRtest1 is just putting an extra oversampling over the Rift-S since it only runs at 80Hz. So to compare with other devices the required SS is 184%, so a bit more that the 9.5 million pixels of the VRTest1.

At your 3070 is the limiting factor for the VRtest1 at 184%, but at 120% is quite OK your obtain the 80fps.

 

Regarding your drops in fps in Furballs, there your CPU could also constrains the system. the best tool to check that is fpsVR, it is in Steam.

The CPU or core % load indicators of tools like HWmonitor are useless for IL-2.

13 hours ago, MilitantPotato said:

My 4090 scores are exactly the same as a 3080. I think we need a 35mp+ benchmark, as even a 4090 is bottlenecked with the current benchmark setup

 

Many thanks for providing the first tests of a 4090 with this benchmark.

 

The reason for not having almost any gain in your comparison is because your main bottleneck for the VRtest2 is your CPU and the WMR system for G2.

If you look at the table, people with the 5800X and the G2 were in the 71-73 fps region. So yous system behaves exactly as expected.

If you upgrade your CPU from 5800X to 5800X3D then you will get around 88-89 in the VRtest2.

 

We could create a third VRtest3 to increase the %SS and render a resolution required from top devices like Aero and Pimax8KX (or any device using a higher SS, for example a G2 at 140%SS).

The VRtest1 is rendering 9.5 million pixels, the VRtest2 is rendering 19.5 million pixels, so a new VrTest3 could render something like 29.5 million pixels which is the pixels demanded by the Aero at 35ppd and 100%SS. Sound that reasonable?

9 hours ago, MilitantPotato said:

I saw zero change from a 3080 to a 4090 on a 5800x for VR test 2. Completely CPU limited ?  Looking at your results I may snag a 5800X3D myself. 

I see this now. I read posta in FIFO mode. So you know this.

[firdimigdi]

Got around to doing some VR tests

 

HP Reverb G2 with OpenComposite/OpenXR Runtime 112.2209.30002

 

vrtest1 (50% - 2232x2180)
Frames: 5344 - Time: 60000ms - Avg: 89.067 - Min: 78 - Max: 91
Frames: 5332 - Time: 60000ms - Avg: 89.017 - Min: 79 - Max: 91

 

vrtest2 (100% - 3156x3084)
Frames: 5292 - Time: 60000ms - Avg: 88.200 - Min: 77 - Max: 91
Frames: 5305 - Time: 60000ms - Avg: 88.900 - Min: 78 - Max: 91

 

EDIT:

In hindsight I remembered that I had forgotten to set the power management mode to "Prefer maximum performance" in the nvidia control panel as requested by the test conditions but I guess this just goes to verify that it really makes no difference in such situations as the GPU is always in boosted mode.

Edited October 21, 2022 by firdimigdi

[chiliwili69]

31 minutes ago, firdimigdi said:

Avg: 88.900 - Min: 78 - Max: 91

This is quite nice performance. Did you run the tests before with SteamVR for WMR? (just for comparison)

[chiliwili69]

Hey guys, I have made three changes to the instructions of the SYN_VANDER test for the VR area. They are:

 

1.- The VR test1 will use exactly the same settings than the CPU test, but in VR mode with 9.5millionpixels. I think that now more and more CPUs are better and reaching numbers near 90fps we don´t measure anything. So, the settings are raised to load more the CPU and analyze what  CPUs are better.

 

2.- The VRtest2 follow the same aproach than VRtest1 but raising to 19.5 million pixels.

 

3.- A new VRtest3, same but now with 29.5 million pixels. So we can compare new cards and top resolution devices like Aero or Pimax8KX or future.

 

So, it means that all previous VR tests can not be compare with the new ones, since they will use different settings.

 

I think all would agree with those changes.

Edited October 23, 2022 by chiliwili69

[firdimigdi]

46 minutes ago, chiliwili69 said:

This is quite nice performance. Did you run the tests before with SteamVR for WMR? (just for comparison)

 

No unfortunately, I did this quickly before jumping in to play but I'll make a note next time to give it a whirl just for the sake of comparison as you say.

 

About performance: yes it performs fantastically well which just adds to the frustration when the game engine is tripping over its own feet (as I've described all too many times by now).

 

Edited October 21, 2022 by firdimigdi

[RAAF492SQNOz_Steve]

On 10/20/2022 at 10:57 AM, Alonzo said:

I currently have an AMD 5800X, with G.Skill 3600-CL16 RAM (I believe it's Samsung, model is F4-3600C16-16GTZN). My impression is that it's CPU speed that's holding me back in VR, rather than GPU. I have a 3080 and a Reverb G2 but I don't crank the graphics.

 

For CAD$500 I could get a 5800X3D, and get a bit more life out of this PC. Should I pull the trigger on an upgrade, or wait and maybe build an entirely new rig, try to sell the old rig? Or something else? I did try to tune my memory timings but couldn't get anything stable, I was using one of those pre-canned tools though, maybe that's not the way forward.

I upgraded from a 5800X to the 5800X3D some time ago and noticed very little difference, for IL 2, between the FPS results.
May have picked up 1 FPS average but this could also be due testing being done on different versions of IL 2 when I was confirming performance.

My opinion, for a least IL2, is that it is not worth changing to a 5800X3D for high res VR users i.e. HP G2, Varjo Aero etc.

[MilitantPotato]

23 hours ago, chiliwili69 said:

 

Thanks for your tests.

-snip-

I did indeed use WMR runtime for these tests. I can do an openXR runtime benchmark if that'd be useful, as well as a 140% ss test, if that's where you think it should be.  I will say that at 5944x5795 (350% SS) quick missions and multiplayer the 4090 holds 90fps near 80% usage (when not CPU bound) with fairly high in-game settings, so 29.5mp might not be enough.

 

 

 

@chiliwili69 here's a 350% SS benchmark with the 4090 to express my concern.

 

 

 

2022-10-22 01:03:02 - Il-2
Frames: 4452 - Time: 60000ms - Avg: 74.200 - Min: 58 - Max: 91

 

 

 

 

Here's 500%
2022-10-22 01:13:36 - Il-2
Frames: 4418 - Time: 60000ms - Avg: 73.633 - Min: 59 - Max: 91

Edited October 22, 2022 by MilitantPotato

[MilitantPotato]

And finally, here's 500% SS (7104x6928) at what i'd consider "very nice" settings for VR

2022-10-22 01:28:50 - Il-2
Frames: 3529 - Time: 60000ms - Avg: 58.817 - Min: 51 - Max: 91
 

 

I think VR test 3 should be run with *much* higher settings, and potentially maxed super sampling, as a 4090 is capable of maxing out all in-game settings (aside from MSAA (which can use too much VRAM) at insane super sampling levels. I'd expect the 4090 TI to hold 65fps average with the settings I posted.

 

These tests are done without open_fsr, OpenXR, no GPU overclocking, and the exact Nvidia control panel settings required for the benchmark.  

 

 

I'd be happy to run some benchmarks with different settings and SS levels, feel free to PM me on discord if you'd like MilitantPotato#1978

 

Edited October 22, 2022 by MilitantPotato
extra image

[chiliwili69]

I have run the test with the new benchmark VR settings with IL-2 current version 5.002b:

 

CPUtest (1080p)
Frames: 6705 - Time: 60000ms - Avg: 111.750 - Min: 93 - Max: 165
Frames: 6869 - Time: 60000ms - Avg: 114.483 - Min: 95 - Max: 159
Frames: 6851 - Time: 60000ms - Avg: 114.183 - Min: 93 - Max: 162

 

GPUtest (4K)
Frames: 7396 - Time: 60000ms - Avg: 123.267 - Min: 97 - Max: 148
Frames: 7413 - Time: 60000ms - Avg: 123.550 - Min: 96 - Max: 149
Frames: 7383 - Time: 60000ms - Avg: 123.050 - Min: 98 - Max: 148

 

VRtest1 (9.5Mpixels) Index 106%SS
Frames: 3700 - Time: 60000ms - Avg: 61.667 - Min: 48 - Max: 80
Frames: 3699 - Time: 60000ms - Avg: 61.650 - Min: 48 - Max: 81
Frames: 3765 - Time: 60000ms - Avg: 62.750 - Min: 50 - Max: 80

 

VRtest2 (19.5Mpixels) Index 216%SS
Frames: 3207 - Time: 60000ms - Avg: 53.450 - Min: 43 - Max: 79
Frames: 3123 - Time: 60000ms - Avg: 52.050 - Min: 43 - Max: 76
Frames: 3136 - Time: 60000ms - Avg: 52.267 - Min: 44 - Max: 73

 

VRtest3 (29.5Mpixels) Index 326%SS
Frames: 3154 - Time: 60000ms - Avg: 52.567 - Min: 43 - Max: 77
Frames: 3207 - Time: 60000ms - Avg: 53.450 - Min: 42 - Max: 80
Frames: 3090 - Time: 60000ms - Avg: 51.500 - Min: 42 - Max: 75

 

Now I can see that with these new VR settings which load more the CPU (like in CPUtest 1080p monitor test), my 5600X is not able to mantain the 90fps.

So, there is room to improve with higher CPUs and we can measure it.

 

So, I believe that in the VRtest1 I am CPU constrained, but in the VRTest2 I am both CPU & GPU constrained (depeding on the scene)

 

But the strange thing is that the VRtest3 should produce lower fps (more GPU constrained) but it doesn´t. It gives exactly the same results than VRtest2.

 

I can not find an answer for that.

 

Perhaps I hit another bottleneck (BUS data transfer?, any limitation on resolution on SteamVR or IL?,....).

 

 

4 hours ago, MilitantPotato said:

I did indeed use WMR runtime for these tests. I can do an openXR runtime benchmark if that'd be useful

 

If you can, you could run run the VRTest1 with the new settings using SteamVRforWMR and another run  OpenXR, so you will be able to compare if OpenXR will help the CPU to achieve a better framerate.

4 hours ago, MilitantPotato said:

5944x5795 (350% SS) quick missions and multiplayer the 4090 holds 90fps near 80% usage

 

All this results using a ver high number of pixels seems to give the same performance regardless of the %SS. That´s  is very strange.

I think there is a limit somewhere limiting the total resolution. It makes no sense that increase %SS the peformance remains the same. I also experiment the same in my tests.

[chiliwili69]

I am not sure, but in the past SteamVR was limiting the maximum reder resolution based on the GPU model.

 

https://www.reddit.com/r/oculus/comments/jyc70z/steamvr_limits_your_render_resolution_depending/

 

For example, for my 3080 and the Index the suggested SS% is 150%, it is was I normally use.

And I don´t know if above 200% or 250% or whatever, the SS% has no further effect in changing the resolution. I need to investigate this more.

 

[MilitantPotato]

Here's my initial tests with the new VR3 settings

SteamVR+WMR 3888x3790 (150% in steamVR)
Frames: 3432 - Time: 60000ms - Avg: 57.200 - Min: 48 - Max: 80

OpenXR (stock settings) results at 3888x3790 (150% I believe)

Frames: 3537 - Time: 60000ms - Avg: 58.950 - Min: 49 - Max: 88

 

I tried 5944x5796 and got similar results, so your theory about a SS cap in steamVR may be true, or one of the settings being used stalls the render pipeline.

 

Also worth noting, my GPU usage was only 40% and power draw was 180 watts, instead of 350-400 watts.

I'll try OpenXR next.    OpenXR Results added @chiliwili69

 

 

Here's a frametime/usage graph showing the GPU is barely being used

Edited October 22, 2022 by MilitantPotato
Added OpenXR VR3 results