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

[StefanGebhardt]

The memory is running at 3600MHz CL16-16-16-32. I tried to raise it to 3800MHz but it wasn't working right away so I reverted back to 3600MHz. I think running four sticks will add a layer of complexity ?

 

I am not too worried about the (temporary) issues of AMD CPU's and Windows 11. They will sort it out, I am confident. 

 

Is there a good overview of the video settings in IL2 for a setup using a 3080TI or 3090?

 

 

[FTC_Mephisto]

This is a great thread. What is the latest view for best CPU to run VR with Il-2? I was able to get my hands on 3090 but currently I am bottlenecked by i7 CPU and considering changing this… Other recommendations for Mobo and RAM would be also much appreciated. Thank you!

[StefanGebhardt]

3 hours ago, FTC_Mephisto said:

This is a great thread. What is the latest view for best CPU to run VR with Il-2? I was able to get my hands on 3090 but currently I am bottlenecked by i7 CPU and considering changing this… Other recommendations for Mobo and RAM would be also much appreciated. Thank you!

At this stage (release of Intel's 12xxx series coming up next) AMDs 5900 is probably the best gaming CPU. Personally I decided for more cores even if this slightly impacts the gaming performance (less headroom for higher frequencies) and picked the 5950X. 

 

My last two builds sticked to Asus mainboard and AMDs X570-E is normally on the recommendation list on many sites for an AMD based machine. Guess it comes down to the budget. An slighly cheaper alternatives is the Asus X570-F board with less fancy extra components (e.g. no 2.5GHz network port). Based on a variaty of reviews there shouldn't be any significant performance differences between boards of the same chipset. 

 

Depends on the availabilty Corsair and GSkill are good choices for memory moduls. 3600MHz is probbably the go-to frequency but most of AMDs 5000 CPUs should be capable of running at 3800MHz. As I said in my initial post it seems to be harder to reach this speed if you are dealing with four rather than the typical two moduls. I also do photoediting (beside gaming) and selected two kits adding to up 64GB at 3600MHz and CL16-16-16-32 timings. 3600MHz CL14 are way more expensive while 3600MHz CL18 moduls are way cheaper again. In addition the CL14 moduls require higher voltages for stability reasons.

 

As always it comes down to the amount of money you are willing to invest. Thanks to the GPU shortage a big portion of the budget needs to be allocated to the video card... Speaking for myself I foced mainly on CPU and videocard and tried to balance out all other components. My previous machine from 2015 (i7 5960X, Titan Xp (higher end version of the GTX 1080 TI) both watercooled) was/is still able to run games like Ghost Recon Breakpoint at 45 fps at this highest settings in 4K. I haven't tried IL2 with VR alot but this should change with my new computer setup ?

Edited October 18, 2021 by StefanGebhardt

[StefanGebhardt]

Here are some none-VR benchmarks. Hard to say if my current setup isn't optional or if the slower than expected CPU performance comes from the AMD/Windows 11 issues.

 

 

 

 

 

CPU 1080p:

Frames: 6161- Yime: 60000ms - AVG: 102.683 - Min: 90 - Max: 139

 

GPU 4K:

Frames: 8554- Time: 60000ms - AVG: 142.567 - Min: 128 - Max: 187

 

VR Test 1:

Frames: 3630- Time: 60000ms - AVG: 60.5 - Min: 49 - Max: 85

 

Summary:

I can't say for sure if my setup is not correctly configured yet or if Windows 11 has quite some impact on it. I am tempted to setup the machine using Windows 10...

 

Edited October 18, 2021 by StefanGebhardt

[chiliwili69]

20 hours ago, FTC_Mephisto said:

Mobo and RAM

 

If I would go today for a new PC, I would pick AMD Zen3 processor with a known brand Mobo (Asus, MSI). I would review the best RAM (high speed, low latency) model which is QVL certified for that particular Mobo.

For example:

https://www.gskill.com/qvl/165/326/1605159977/F4-3800C14D-32GTZN-Qvl

15 hours ago, StefanGebhardt said:

CPU 1080p:

Frames: 6161- Yime: 60000ms - AVG: 102.683 - Min: 90 - Max: 139

 

Definetely something weird is happenning here. I will not upload this result until we find the cause.

You can also check this old post, just in case:

 

 

 

If the Win11+AMD is really the issue, then it is a good warning to all AMD users here.

 

If anyone with a 5950X can run or run again the CPU test just to confirm nothing changed in IL-2 code.

[FTC_Mephisto]

Thank you. Looking at the prices is i7-11700K not a more viable option vs. 5950X? 

[StefanGebhardt]

Hello,

 

I decided to have a fresh start using Windows 10 for the best possible performance at this stage. There seems to be some issues with AMD processors and Windows 11, preventing the hardware to work as fast as it should be at this stage. I was aware that some of the mandatory features like secure boot etc will come with a minor performance penalty (based on testing performed by computerbase.de) but the overall gaming experience wasn't up to what I was hoping for. I am confident that I will revisit Windows 11 at a later stage. Below are the results of my benchmarking today.

 

Notes:

• HP Reverb G1
• Nvidia 496.13

 

 

 

CPU 1080p:

Frames: 6882- Yime: 60000ms - AVG: 114.700 - Min: 101 - Max: 160

 

GPU 4K:

Frames: 9464- Time: 60000ms - AVG: 157.733 - Min: 138 - Max: 201

 

VRTest 1:

Frames: 4811 - Time: 60000ms - AVG: 80.183 - Min: 60 - Max: 91

 

VRTest 2:

Frames: 4662 - Time: 60000ms - AVG: 77.700 - Min: 58 - Max: 91

 

Summary:

• Personally I will wait for updates from Microsoft and or AMD to address the issues linked to the use of AMD processors and Windows 11.
• The test results above seems to be within the expected range. One way to improve them would be to remove two memory Moduls and run them at 3600MHz CL-14 or 3800MHz. The X-570 chipset doesn't seems to be to kind to the use of four moduls when it comes to really fast timing and frequencies. I prefer to stick to my current configuration and timing due to the lack of 2x32GB moduls with 3600MHz CL-14 or the chipset/mainboard abaility to run at 3800MHz with decent timings and four moduls installed. 
• A comparision of the results above with the ones collected in the table linked on the first page seems to indicate a drop of performance between IL2 4.5xx and 4.6xx. All users reaching >85 fps in VRTest1 or VRTest2 benchmarked their systems using IL2 4.5xx. All publised results using the version 4.6xx are limited to <82 fps in these tests. It would be nice if chiliwilli69 or DBCOOPER01 could rerun the tests using 4.6xx. I could be wrong and their machines are just significant faster but it would be good get confirmation on this topic.

Edited October 20, 2021 by StefanGebhardt

[chiliwili69]

17 hours ago, StefanGebhardt said:

CPU 1080p:

Frames: 6882- Yime: 60000ms - AVG: 114.700 - Min: 101 - Max: 160

 

Good to see that your switch to Windows 10 improved your CPU Passmark and also your results in IL-2.

Your numbers are now better aligned with your peers and I put them into the table

 

I have re-run the full benchmark in my system just to check if the lastest versions affects performance.

 

v604b
CPU test
Frames: 6903 - Time: 60000ms - Avg: 115.050 - Min: 100 - Max: 155
Frames: 6906 - Time: 60000ms - Avg: 115.100 - Min: 100 - Max: 155

 

4K GPU test
Frames: 8436 - Time: 60000ms - Avg: 140.600 - Min: 111 - Max: 167
Frames: 8482 - Time: 60000ms - Avg: 141.367 - Min: 112 - Max: 171

 

VRtest1 (Index at 106% SS)
Frames: 5238 - Time: 60000ms - Avg: 87.300 - Min: 77 - Max: 92
Frames: 5250 - Time: 60000ms - Avg: 87.500 - Min: 76 - Max: 92

 

VRtest2 (Index at 216% SS)
Frames: 5181 - Time: 60000ms - Avg: 86.350 - Min: 73 - Max: 92
Frames: 5198 - Time: 60000ms - Avg: 86.633 - Min: 75 - Max: 91

 

 

The CPU and GPU tests are about 4-5fps below previous test with 4.601, and VR tests about 1 fps less.

 

This would explain a bit also the lower results of @BH_Adabadoo_VR and @Voyager who used the v604 version.

18 hours ago, StefanGebhardt said:

One way to improve them would be to remove two memory Moduls and run them at 3600MHz CL-14 or 3800MHz.

Yes, this test would be interesting

[StefanGebhardt]

23 minutes ago, chiliwili69 said:

 

Good to see that your switch to Windows 10 improved your CPU Passmark and also your results in IL-2.

Your numbers are now better aligned with your peers and I put them into the table

 

I have re-run the full benchmark in my system just to check if the lastest versions affects performance.

 

v604b
CPU test
Frames: 6903 - Time: 60000ms - Avg: 115.050 - Min: 100 - Max: 155
Frames: 6906 - Time: 60000ms - Avg: 115.100 - Min: 100 - Max: 155

 

4K GPU test
Frames: 8436 - Time: 60000ms - Avg: 140.600 - Min: 111 - Max: 167
Frames: 8482 - Time: 60000ms - Avg: 141.367 - Min: 112 - Max: 171

 

VRtest1 (Index at 106% SS)
Frames: 5238 - Time: 60000ms - Avg: 87.300 - Min: 77 - Max: 92
Frames: 5250 - Time: 60000ms - Avg: 87.500 - Min: 76 - Max: 92

 

VRtest2 (Index at 216% SS)
Frames: 5181 - Time: 60000ms - Avg: 86.350 - Min: 73 - Max: 92
Frames: 5198 - Time: 60000ms - Avg: 86.633 - Min: 75 - Max: 91

 

 

The CPU and GPU tests are about 4-5fps below previous test with 4.601, and VR tests about 1 fps less.

 

This would explain a bit also the lower results of @BH_Adabadoo_VR and @Voyager who used the v604 version.

Yes, this test would be interesting

 

Do we know if we can compare the results from the HP Reverb G1 with the one from the Index 1-to-1? I am surprised that my setup performs better in the 4K GPU test while falling behind in the VR tests. Not worried, just interested. Do we have a quide for all the settings in Windows Mixed Reality, Steam VR etc?

[chiliwili69]

4 minutes ago, StefanGebhardt said:

Do we know if we can compare the results from the HP Reverb G1 with the one from the Index 1-to-1? I am surprised that my setup performs better in the 4K GPU test while falling behind in the VR tests. Not worried, just interested. Do we have a quide for all the settings in Windows Mixed Reality, Steam VR etc?

 

All WMR add an extra layer of software "WMR for SteamVR" app. That´s why you can not compare 1-to-1 the results. You need to take into account that this benchmark measure the hardware, but also the software required to run it.

 

Most of the VR tests with top cards (3080 and above) are not really bottlenecked by the card, but the CPU which need to compute the geometry all those elements twice.

The only change from VRtest1 to VRtest2 is resolution, so if GPU is not bottlenecked most of the 3080 (and above) achieve almost the same.

 

That´s why the CPU test is important for VR. In fact, the CPU test is highly correlated to the VRtests if the GPU is not bottlenecked.

 

I was using G1 and G2 in the past, and I remember that I was having less performance than with the Index (look this post).

 

Right now I am not familiar with the lastest tricks regarding WMR, but there is section in the VR subforum.

[StefanGebhardt]

I can give it a shot tomorrow or the day after and check if I can run just two Moduls at 3600MHz CL-14 or 3800MHz CL-16 (?). Will report back when it’s done. 
 

The Moduls should be capable of doing so…

https://www.tomshardware.com/reviews/gskill-trident-z-neo-ddr4-3600-c16-2x16gb-review

 

Update:

I did a quick run of the VRTest 1 with 2 x 16GB at the following timings and results:

• 3600MHz CL14-14-14-35-52-2T --> +1fps
• 3600MHz CL16-16-16-36-52-1T --> Baseline
• 3800MHz CL18-18-18-36-89-1T --> +2fps

 

I did another round of testing today, this time with all typical applications installed (meaning more background processes running) and my results are slightly lower than before.

 

VRTest1:

Frames: 4673 - Time: 60000ms - Avg: 77.883 - Min: 58 - Max: 92

compared to

Frames: 4811 - Time: 60000ms - AVG: 80.183 - Min: 60 - Max: 91

before

 

To backup my claim that the HP Reverb using WMR will not be able to keep up with the Index I would like to compare the results of chiliwili69 latest tests with mine:

 

CPU 1080p

Stefan

Frames: 6882- Yime: 60000ms - AVG: 114.700 - Min: 101 - Max: 160

chiliwili69

Frames: 6906 - Time: 60000ms - Avg: 115.100 - Min: 100 - Max: 155

 

--> Both systems performing nearly identical 

 

GPU 4K 

Stefan

Frames: 9464- Time: 60000ms - AVG: 157.733 - Min: 138 - Max: 201

chiliwili69

Frames: 8482 - Time: 60000ms - Avg: 141.367 - Min: 112 - Max: 171

 

--> Stefan's system is roughly 10% faster 

 

VRtest1

 

Stefan - HP Reverb G1

Frames: 4673 - Time: 60000ms - Avg: 77.883 - Min: 58 - Max: 92

chiliwili69 Index

Frames: 5250 - Time: 60000ms - Avg: 87.500 - Min: 76 - Max: 92

 

--> - Index leads the HP Reverb G1 by 10fps

 

My final summary:

Adjusting the memory to faster timings or higher frequencies is not a gamer changer for me. I will stick to the rather common 3600MHz CL16 setup. The choice of VR hardware will have an impact on the performance due to the inclusion of WMR (for the HP Reverb G1/G2) and this makes it less straight forward to compare the results with each other. As shown above similar results in the CPU1080p and GPU4K test does not necessarily result in nearly identical performing VR tests. 

 

Edited October 21, 2021 by StefanGebhardt

[chiliwili69]

On 10/20/2021 at 1:35 PM, StefanGebhardt said:

I did a quick run of the VRTest 1 with 2 x 16GB at the following timings and results:

• 3600MHz CL14-14-14-35-52-2T --> +1fps
• 3600MHz CL16-16-16-36-52-1T --> Baseline
• 3800MHz CL18-18-18-36-89-1T --> +2fps

 

Thanks for your additional tests. But I see that when you use the test you use the VRTest1. This is fine, but for RAM or Settings changes I normally use just the CPU test, because it is not capped by the 90fps.

Remember that if you avg fps for the VRtest1 is 80fps (from 90fps has the max), it means that most likely about 70% or more of the time you are at 90, so the extra fps by the RAM changes are not impacting the average.

 

Note also that the VRtest1 is just a syntethic test using High presets with everything else to low. You actual game setting would be quite different.

 

 

On 10/20/2021 at 1:35 PM, StefanGebhardt said:

Stefan

Frames: 6882- Yime: 60000ms - AVG: 114.700 - Min: 101 - Max: 160

chiliwili69

Frames: 6906 - Time: 60000ms - Avg: 115.100 - Min: 100 - Max: 155

 

--> Both systems performing nearly identical

 

This is also a bit strange. The 5950X should get more than the 5600X since it overclock higher. Other test of @Jaws2002 or other with the same version and same memory were about +15fps than me in the CPUTest1.

 

Perhaps, if someone with a 5900X or 5950X could make just the CPUtest again we could have more insight.

On 10/20/2021 at 1:35 PM, StefanGebhardt said:

The choice of VR hardware will have an impact on the performance due to the inclusion of WMR (for the HP Reverb G1/G2) and this makes it less straight forward to compare the results with each other

 

Yes, this is right. The benchmark take that impact into account since all headset are running for the same 90Hz mode and same pixel count.

 

That impact will affect the game play in the same way, so better to know it.

 

But many headsets has other extra frequencies (60Hz, 72Hz, 80Hz) so you can play with it to improve the experience. But in the benchmark we defined 90Hz since it is the most common frequency of all headset. (except RiftS which only has 80Hz).

 

I believe that some G1/G2 owners use the 60Hz mode, but not sure.

 

[Jaws2002]

On 10/21/2021 at 2:45 PM, chiliwili69 said:

Perhaps, if someone with a 5900X or 5950X could make just the CPUtest again we could have more insight.

I just did a quick test. I didn't have the fans running at 100%, and no tweaks. CPU is stock + XMP at 3600MHz.

 

2021-10-24 13:27:50 - Il-2
Frames: 7663 - Time: 60000ms - Avg: 127.717 - Min: 112 - Max: 178

2021-10-24 13:29:36 - Il-2
Frames: 7565 - Time: 60000ms - Avg: 126.083 - Min: 111 - Max: 171

 

I think, in my case, the new overclock 3080TI doesn't help much in CPU score, because it burns some 450W of power and the pass through cooler design blows a lot of hot air straight across the CPU and into the CPU radiator. 

 Everything runs quite a bit hotter since i upgraded from 1080ti to 3080ti. Obviously, while I think the CPU performance got a hit, the GPU is significantly more powerful.

 

@StefanGebhardt  Your CPU can definitely do better.

Edited October 24, 2021 by Jaws2002

[StefanGebhardt]

21 hours ago, Jaws2002 said:

I just did a quick test. I didn't have the fans running at 100%, and no tweaks. CPU is stock + XMP at 3600MHz.

 

2021-10-24 13:27:50 - Il-2
Frames: 7663 - Time: 60000ms - Avg: 127.717 - Min: 112 - Max: 178

2021-10-24 13:29:36 - Il-2
Frames: 7565 - Time: 60000ms - Avg: 126.083 - Min: 111 - Max: 171

 

I think, in my case, the new overclock 3080TI doesn't help much in CPU score, because it burns some 450W of power and the pass through cooler design blows a lot of hot air straight across the CPU and into the CPU radiator. 

 Everything runs quite a bit hotter since i upgraded from 1080ti to 3080ti. Obviously, while I think the CPU performance got a hit, the GPU is significantly more powerful.

 

@StefanGebhardt  Your CPU can definitely do better.

That's what I thought too but all benchmarks I was running so far outside the IL2 world are returning back good to very good results. Not quite sure why IL2 is not aligned. Do you have a chance to run the CPU-Z benchmark? 

 

Just a quick question when it comes to in-game performance. I used the IL2 VR settings guide from gamersbynight.com on my install and did a few quick missions (Rheinland, Kuban) with 2 + 8 planes and my setup was running at 88fps in average according to fpsVR or 90 based on the in-game framecounter (HP Reverb G1). My settings are pretty much identical to the example screenshot below with the following few exceptions:

• Clounds - High
• Terrain roughness - Medium
• HDR - Off

 

 

 

 

Edited October 25, 2021 by StefanGebhardt

[Jaws2002]

I used the settings on the first page of this thread. I don't have VR, so no VR relate tweaks.  

My CPU is running stock. I got really good results, in game, with the last windows install, and i don't mess with it anymore. Looks like yours can do better in 2d Il-2. Maybe some of those VR settings affect 2d monitor performance.

 

Edited October 25, 2021 by Jaws2002

[StefanGebhardt]

13 hours ago, Jaws2002 said:

I used the settings on the first page of this thread. I don't have VR, so no VR relate tweaks.  

My CPU is running stock. I got really good results, in game, with the last windows install, and i don't mess with it anymore. Looks like yours can do better in 2d Il-2. Maybe some of those VR settings affect 2d monitor performance.

 

 

Thanks for posting your results. The issue we are facing is that IL2 CPU benchmarks are behaving (not sure if it is limited to me) strange and not aligned to other benchmarks on the same platform. The key for me is to make sure that the system is working within the expected range when it comes to the fps. The benchmarks really helped me to establish that something was wrong for me in Windows 11 but right now I tend to think I am fine when it comes to IL2 in VR with the settings I posed above. 

 

Can someone tells me his video settings in IL2 for VR when it comes to Shadows etc. and the resulting fps in the game?

Edited October 26, 2021 by StefanGebhardt

[Jaws2002]

AMD released new Chipset drivers for Windows 11. They are saying the performance bug is fixed. See if this helps.

 

 

https://www.igorslab.de/en/new-driver-for-all-ryzen-under-windows-11-amd-clears-the-cppc2-bug-for-solved/

Edited October 26, 2021 by Jaws2002

[chiliwili69]

On 10/24/2021 at 7:44 PM, Jaws2002 said:

2021-10-24 13:27:50 - Il-2
Frames: 7663 - Time: 60000ms - Avg: 127.717 - Min: 112 - Max: 178

 

Thanks, this confirms also that the lastest versions 4.604 has a small impact over CPU test. The GPU is not the guilty.

14 hours ago, StefanGebhardt said:

not sure if it is limited to me

 

I have been thinking about why you perform quite well in all other becnhmarks but not in the CPU test.

I think the problem is not VR related, and if we solve the issue in 2D, then you will get better VR results as well.

 

Firstly, I assume that you run the CPU exactly with the settings indicated in the first post, and also you uninstall any IL-2 mod or shader or any other thing beyond IL-2.

 

Secondly, I have seen that your monitor is 4K with 31". Some people using 4K monitors (including me) tend to use a 150% scaled mode (Display Settings/Scale) of windows. So the 1920x1080 IL-2 window (in not full screen) use more than a 1/4 of the 4K screen. Technically a 1920x1080 is 1/4 of a 4K.

To run the CPU test, you have to put the scale mode of Windows at just 100%, so the IL-2 screen uses just exactly 1/4 of the 4K screen.

This has a small impact and probably not justify you lower values, but just worth to check.

[StefanGebhardt]

I had another look at the CPU related 'issue'. My screen is running at 100% scalling already, no need to change anything. I copied over the install directory from my old machine and wasn't 100% sure if I applied any mods or customisations in the past (haven't touched IL2 for a long time). I backed up the folder and downloaded the whole content from scratch - No change in the outcome. 

 

Nevertheless I think I figured out the reason for the lower than expected fps for the CPU test. I kept MSI Afterburner running side by side with the benchmark and noticed that the GPU core clock does not ramp up to 2GHz as normal when running the IL2 cpu benchmark. This means it is running roughly 300MHz lower compared to the 4K or VR tests. Not sure why the Asus card is behaving like this (or is it the same for you guys too?) but it would be an explanation. I am aware it is a CPU focused test but still, the video card will have an impact - right?

Edited October 28, 2021 by StefanGebhardt

[dburne]

9 hours ago, StefanGebhardt said:

I had another look at the CPU related 'issue'. My screen is running at 100% scalling already, no need to change anything. I copied over the install directory from my old machine and wasn't 100% sure if I applied any mods or customisations in the past (haven't touched IL2 for a long time). I backed up the folder and downloaded the whole content from scratch - No change in the outcome. 

 

Nevertheless I think I figured out the reason for the lower than expected fps for the CPU test. I kept MSI Afterburner running side by side with the benchmark and noticed that the GPU core clock does not ramp up to 2GHz as normal when running the IL2 cpu benchmark. This means it is running roughly 300MHz lower compared to the 4K or VR tests. Not sure why the Asus card is behaving like this (or is it the same for you guys too?) but it would be an explanation. I am aware it is a CPU focused test but still, the video card will have an impact - right?

 

Yes it was the same for me also. Benchmark would not boost as much as it does during regular play.

[chiliwili69]

16 hours ago, StefanGebhardt said:

the video card will have an impact - right?

 

Your GPU (3090) is not really loaded during the CPU test, so it doesnt need to boost the frequency and that has not an impact in the CPU test result. If you do manual overclock over the GPU you will see that no gain is get in the CPU test.

 

I really don´t know what else you can try.

 

I assume you updated Chipset and BIOS from Mobo webpage.

 

It shouldnñt affect, but just in case:

 

- Review if there is any process running in background consuming some CPU time. (no other app should be running during CPU test)

- use another 1080p monitor or TV just in case,

- remove two sticks of memory  and just test it with 2x16Gb (put them as indicated by your Mobo: SIMM_A2 and DIMM_B2)

- Disable the hypertreading (SMT) in BIOS

- Disable some of the slower cores (using AMD Ryzen Master or BIOS)

[StefanGebhardt]

9 hours ago, chiliwili69 said:

 

Your GPU (3090) is not really loaded during the CPU test, so it doesnt need to boost the frequency and that has not an impact in the CPU test result. If you do manual overclock over the GPU you will see that no gain is get in the CPU test.

 

I really don´t know what else you can try.

 

I assume you updated Chipset and BIOS from Mobo webpage.

 

It shouldnñt affect, but just in case:

 

- Review if there is any process running in background consuming some CPU time. (no other app should be running during CPU test)

- use another 1080p monitor or TV just in case,

- remove two sticks of memory  and just test it with 2x16Gb (put them as indicated by your Mobo: SIMM_A2 and DIMM_B2)

- Disable the hypertreading (SMT) in BIOS

- Disable some of the slower cores (using AMD Ryzen Master or BIOS)

All drivers etc are fully up to date. I am also followed up on Windows modification (game mode off) based on recommendations in the DCS forum. I need to say that I am not to bothered about the CPU results any longer. The IL2 CPU benchmark is the only one showing underwhelming results while all other IL2 related and generic benchmarks indicating good performance. I was already running a two stick setup but focused on the VR benchmark. Still even if there is an impact on the CPU side of things at the end it comes down on the impact on the real world speed. 
 

Just one interesting observation I made when I looked into the DCS forum and VR related settings. The main focus there seems to be on frame-times rather than fps online. Thoughts on this one?

[chiliwili69]

On 10/29/2021 at 6:07 AM, StefanGebhardt said:

Just one interesting observation I made when I looked into the DCS forum and VR related settings. The main focus there seems to be on frame-times rather than fps online. Thoughts on this one?

 

Average Frametimes are equivalent to Average fps when the fps are always above 45fps and below 90fps in a 90Hz VR device.

Frametimes give you information in all conditions, when you are below 90fps and also when you are over 90fps (ie when frametimes are below 11.11 milisecons, which is 1000/90).

 

So, when the fps counter is at the maximum 90fps you dont know how well you system is doing (how large is the margin). And then the Frametime counter would indicate 11ms or 10 or 6.

 

So, for VR performance, it is better to use Frametimes since it is not capped by the 90 fps limit. For example, fpsVR tells you CPU and GPU frametimes, so you can see in realtime how large is the margin to the 11.11ms.

 

In the monitor test (1080p or 4K) we don´t have that problem since Fraps measure more fps than the native Hz of the monitor.

 

This SYN_VANDER benchmark is essentially a monitor test for IL-2 and use Fraps for that (so we know best CPU and GPU independently). And I was also adding the VR tests as an option and trying to put the settings in VR to produce values between 45-90

[A_radek]

Alder lake, Anyone building yet?

 

Urgently need an upgrade but not looking forward to the detox/withdrawal-period if il2 goes 'total war syndrome' using alder lake.

Edited November 5, 2021 by SvAF/F16_radek

[DBCOOPER011]

14 hours ago, SvAF/F16_radek said:

Alder lake, Anyone building yet?

 

Urgently need an upgrade but not looking forward to the detox/withdrawal-period if il2 goes 'total war syndrome' using alder lake.

 

I'm waiting on a 12900K/Asus Z690-A scheduled to arrive next Tuesday, and just received 32GB DDR5-5200mhz yesterday. Will post the results after I get it together...

[chiliwili69]

9 hours ago, DBCOOPER011 said:

I'm waiting on a 12900K/Asus Z690-A scheduled to arrive next Tuesday, and just received 32GB DDR5-5200mhz yesterday. Will post the results after I get it together...

 

Nice!.

I have added a new column to track the OS version (windows10 vs Windows11) just in case there is any influence.

 

While you wait you could just make a last run with your 5800X with the lastest IL-2 just for the CPU test, so we can truly get the net gain.

[DBCOOPER011]

5 hours ago, chiliwili69 said:

 

Nice!.

I have added a new column to track the OS version (windows10 vs Windows11) just in case there is any influence.

 

While you wait you could just make a last run with your 5800X with the lastest IL-2 just for the CPU test, so we can truly get the net gain.

 

No problem, my 5800X is in my backup computer now with a 3060XC Gaming, this is what I got with my valve index at 90mhz. My 3090 is waiting for the new motherboard/CPU..

 

Motherboard: Gigabyte X570 Aorus Elite
CPU:                Ryzen 5800X    
CPU Freq:       3700Mhz +150 PBO2 and CO
Cores:             8
Threads:         16      
RAM size:       32Gb (4x8GB)
RAM Freq:      3600 MHz
NB Freq:         1800 MHz (or Uncore Frequency or UCLK)
RAM timings: 14-14-14-28
GPU:               EVGA 3060XC Gaming
OS:                  Win11

 

CPU Test:
Frames: 7409 - Time: 60000ms - Avg: 123.483 - Min: 109 - Max: 172
Frames: 7075 - Time: 60000ms - Avg: 117.917 - Min: 102 - Max: 154
Frames: 7131 - Time: 60000ms - Avg: 118.850 - Min: 103 - Max: 151

 

VR Test 1:
Frames: 5152 - Time: 60000ms - Avg: 85.867 - Min: 76 - Max: 91

 

VR Test 2:
Frames: 2701 - Time: 60000ms - Avg: 45.017 - Min: 43 - Max: 47

 

Edited November 6, 2021 by DBCOOPER011

[zhutwo]

It should be interesting to see what wins out between Alder Lake and the upcoming 3D Cache Zen 3 chips. Alder Lake has quite a large gain in single-thread performance, but a massive L3 cache upgrade should help with complex simulations where lots of data needs to be continually referenced, as well as unoptimized code, which a smaller studio may have less resources to dedicate toward.

[chiliwili69]

18 hours ago, DBCOOPER011 said:

No problem, my 5800X is in my backup computer now with a 3060XC Gaming, this is what I got with my valve index at 90mhz.

 

Many thanks for your test. It the first in three things:

1.- The first with Windows 11

2.- The first with version 4.605 of IL-2

3.- The first with a 3060 card

 

On the CPU test you are in 120fps now and in one of your lastest test with the same 5800X  (IL-2 V4.506, Windows 10 and RAM at 1833MHz) you were at 123 fps.

 

We have seen an small impact (about -5fps) in the CPU test from 4.604 version. And we still don´t know the impact that Windows 11 could have in IL-2, probably very small.

 

I also see that your 3060 cards performs quite well in the VRTest1, with 9.5 Million pixels. This info is useful for people with lower resolution headsets like Index, Odysey, Vive Pro, Rift-S etc.

 

Let´s see now how it goes with the new 12900K

[DBCOOPER011]

Just got this thing put together with a new install of win11. Was able to do 3 cpu-test runs in IL2 and some benchmarks. Everything is stock except I put +2 for tvb and used the DDR5 xmp profile at 5200mhz. I may be doing something wrong, but it doesn't appear that it overtook AMD in IL2, at least in the cpu test. Will do the VR tests tomorrow. The temps were real good for me, pretty much the same as my 5800X..

 

2021-11-09 19:05:18 - Il-2
Frames: 7983 - Time: 60000ms - Avg: 133.050 - Min: 116 - Max: 181

2021-11-09 19:10:28 - Il-2
Frames: 8188 - Time: 60000ms - Avg: 136.467 - Min: 116 - Max: 193

2021-11-09 19:16:04 - Il-2
Frames: 8214 - Time: 60000ms - Avg: 136.900 - Min: 122 - Max: 186

 

 

[A_radek]

Many thanks Dbcooper. Had expected il2 to get a boost using ddr5, not so much on ddr4 vs amd with better memory. But, now we know it actually runs on alder lake. Are you on win11?

Edited November 10, 2021 by SvAF/F16_radek

[FoxbatRU]

I'm thinking of replacing my 8600k with 12900k while I tested it on the existing system.

 

CPU test
Frames: 5767 - Time: 60000ms - Avg: 96.117 - Min: 85 - Max: 128

 

GPU test
Frames: 8310 - Time: 60000ms - Avg: 138.500 - Min: 117 - Max: 177

 

 Motherboard:  Gigabyte Z390 AORUS PRO
 CPU:                 8600K
 CPU Freq:        4.9 Ghz (50-AVX 1)
 L3 cache:         9 MB
 Cores:               6
 Threads:           6
 RAM type:        DDR4
 RAM size:         32 GB (4x16)
 NB Freq:           4700 MHz
 RAM Freq:        3900 MHz 
 RAM Latency:  18-18-18-36-350
 GPU:                 3080

 OS:                    Windows 10 (x64, 20H2)

 IL2:                   4.605

temp.zip

Edited November 10, 2021 by FoxbatRU

[chiliwili69]

On 11/10/2021 at 4:47 AM, DBCOOPER011 said:

I may be doing something wrong, but it doesn't appear that it overtook AMD in IL2, at least in the cpu test.

 

Many thanks for this first test of the 12900K. But I think it overtook the AMD. Take into account only the CPU Test after IL-2 version 4.604 (to compare apples with apples).

Jaws2002 was achieving 127.7 fps, and your tests are around 133-137. So currently the 12900K is the top performer.

 

Perhaps I could make a new tab in the spreadsheet from version 4.604. But not sure.

[chiliwili69]

On 11/10/2021 at 4:47 AM, DBCOOPER011 said:

Everything is stock except I put +2 for tvb

 

I don´tt how Turbo Bost 2.0, Turbo Boos Max 3.0 and TVB works in your 12900K, but I think there are two modes for Thermal Velocity Boost (TVB), applied to 1 cores (from the 2 faster) or to all cores.  Which one did you applied?

 

I see in your test that you reach 5.4GHz for cores 0 and 1.

 

Perhaps the difference in the CPU tests (133fps vs 137fps) could be due to where heavy IL-2 threads are executed or how TVB apply the boost based in Temp.

 

I don´t know if disabling some P-cores and/or E-cores the power&Temps will decrease and frequencies can be keept at 5.4 during all full test. (and game play!).

18 hours ago, FoxbatRU said:

I'm thinking of replacing my 8600k with 12900k while I tested it on the existing system.

 

Thanks for your test. Let´s us know how it goes if you upgrade.

[DBCOOPER011]

17 hours ago, chiliwili69 said:

 

I don´tt how Turbo Bost 2.0, Turbo Boos Max 3.0 and TVB works in your 12900K, but I think there are two modes for Thermal Velocity Boost (TVB), applied to 1 cores (from the 2 faster) or to all cores.  Which one did you applied?

 

I see in your test that you reach 5.4GHz for cores 0 and 1.

 

Perhaps the difference in the CPU tests (133fps vs 137fps) could be due to where heavy IL-2 threads are executed or how TVB apply the boost based in Temp.

 

I don´t know if disabling some P-cores and/or E-cores the power&Temps will decrease and frequencies can be keept at 5.4 during all full test. (and game play!).

 

Thanks for your test. Let´s us know how it goes if you upgrade.

 

Was able to tinker around with it today and pretty much had the same cpu scores I posted earlier. I did notice that the cores never went beyond 5.3mhz and mainly stayed at 5.2mhz during the IL2 cpu test, even though I had the some of the cores set for 5.4mhz with +2 TVB (5.6mhz). Most other benchmarks I used so far utilized the full 5.6mhz. So far, I think my sweet spot is a peak of 5.4mhz. Temperatures are relatively low during gaming (low-mid 40C), and it doesn't take a lot of cpu package power (less then 100W)during gaming, but it can get hot with all core benchmarks with a high mhz setting.

[SCG_Wulfe]

Just finished putting together my new setup. Quite Pleased.  

 

Motherboard:    Gigabyte B550 Aorus Pro AC
 CPU:                 5600x
 CPU Freq:        4.65 Ghz
 Cores:               4 
 Threads:           12 
 RAM size:        16Gb (2x8GB)
 RAM Freq:        3800 MHz 
 NB Freq:          1900 MHz
 RAM timings:  16-16-16-30
 GPU:                 RTX 3080

OS:                     Windows 10 x64 (21H2)

IL2:                     4.605b

Headset:            Index

CPU Test:
Frames: 7170 - Time: 60000ms - Avg: 119.500 - Min: 103 - Max: 165

GPU 4K Test:
Frames: 8613 - Time: 60000ms - Avg: 143.550 - Min: 112 - Max: 171

 

VR Test 1:
Frames: 5296 - Time: 60000ms - Avg: 88.267 - Min: 81 - Max: 92

 

VR Test 2:
Frames: 5130 - Time: 60000ms - Avg: 85.500 - Min: 75 - Max: 91

Edited November 19, 2021 by SCG_Wulfe

[chiliwili69]

5 hours ago, SCG_Wulfe said:

CPU Test:
Frames: 7170 - Time: 60000ms - Avg: 119.500 - Min: 103 - Max: 165

Thank for this test. Hey! you have almost an indentical setup than me, including Index!

The only difference is the RAM speed, this perhaps explains this extra +4.4fps in the CPU test and +2.5fps in the GPU test.

[SCG_Wulfe]

5 hours ago, chiliwili69 said:

Thank for this test. Hey! you have almost an indentical setup than me, including Index!

The only difference is the RAM speed, this perhaps explains this extra +4.4fps in the CPU test and +2.5fps in the GPU test.

 

Absolutely, thank you for putting this all together.

 

I do indeed have a very similar setup, and not by happenstance. I largely used the results from this benchmark to pick components and build my PC in the most effective/cost effective way possible, specifically for VR in IL2. 

 

The only thing that I'm combatting now as I spent some time last night working on settings for regular use, is that while I am able to maintain 90FPS in VR, I have almost constant micro stutters. I'm not sure if its a performance issue (and needs settings further dialed back), or an overclocking issue, or what. 

 

I'm thinking of starting by running the 5600x at factory settings and removing my overclock and seeing if that can still hold 90 fps with my settings and maybe lose the micro stutters. Secondarily, I may try FSR and see if that gives me enough of a boost to kick them. 

 

edit* also doing some quick reading, I had forgotten about possibly needing to turn SMT off and make sure HAGS is off as well. Will try those first when I get home. 

Edited November 19, 2021 by SCG_Wulfe

[chiliwili69]

22 hours ago, SCG_Wulfe said:

I have almost constant micro stutters. I'm not sure if its a performance issue (and needs settings further dialed back), or an overclocking issue, or what. 

 

That´s strange.

Do you have those microstutters at all conditions? (low or high altitude, clouds vs no clouds, any map? or plane?)

 

You can use fpsVR app to see when those stutters appear, and also try other games.

[SCG_Wulfe]

1 hour ago, chiliwili69 said:

 

That´s strange.

Do you have those microstutters at all conditions? (low or high altitude, clouds vs no clouds, any map? or plane?)

 

You can use fpsVR app to see when those stutters appear, and also try other games.

Yes seem to be constant and in all conditions. Tried with various things last night and nothing really changed them.

 

I will try logging in fpsVR to see if/when the appear in other games. The one thing I forgot about and will try today is deleting GeForce experience/disabling its overlay.