Jump to content
chiliwili69

Measuring rig performance: Common Baseline (for IL-2 v3.010)

Recommended Posts

Congrats for a successful delid operation!

 

I also has been thinking more about why the test doesn´t show a proportional increase from certain overclock frequency. First reason is that the closer you are to 90 the less proportional is the increase since the asymptotic max value is 90. You can try to run exactly the same test but with SS=1 and see how much you GPU is bottlenecking the test.

 

Another important value from the 8700K and 7700K tests is the min fps value. The 8700K is around 84 and 7700K is around 60-62. This weight the overall average meaning that 7700K handle the worst final test  scene a little bit worse than the 8700K. But for remaining test both are at 90fps.

 

You can try to OC your CPU to 5.1 or 5.2 until you achieve a STMArk of 3031 and then run the VR test. (since it is a short test not very demanding in terms of temperature there should not be a problem). Obiously this is only for testing purposes not to use in normal gaming.

Temperatures are not the bottleneck in my Overclock now, fortunately. The bottleneck is the voltage and continuous stability by now.

 

At 5GHz (now with 1.4V because only that gives me continuous stability in IL-2 and in x264-over-night), the max temperature was on core#0 and 69°C. Hence I still got a bit of headroom. But the voltage multiplier is hindering me, I set the limit to 1.4V while playing IL-2 (I go back to 4.6GHz 1.28V at normal operation). I need to see if 5.1GHz remains stable using the same voltage of 1.4V

 

The better min FPS of the 8700K are very interesting indeed!

 

 

 

P.S. In my opinion we may get more expressive results for the most modern machines by creating a secondary, higher demanding set of settings for running the test track. That way we can see much easier how the different CPUs and clocks behave at a heavier load of Ultra+High Shadows+4x terrain. The asymptotes for CPU or GPU or RAM clock you mentioned are more diagonal that way and not largely leveled out yet, hence way more expressive. I like the way you think, are you also Engineer?

Edited by 2./JG51_Fenris_Wolf

Share this post


Link to post
Share on other sites

Thank you for reporting the test results for the first Odyssey test!!

 

We will need to know also you RAM size and speed. You can know that using the CPU-Z free tool.

Your CPU passmark of 5690 is quite low, it should be higher for your CPU. Please, verify with CPU-Z that your CPU is really running at 4.0 GHz during the Passmark test and also during IL-2 test.

Report also your Single Threaded Passmark (it is displayed when you run the CPU Passmark).

 

As you might have seen in the table of results, there is another person (doog442) with the i5-4690K CPU. We still don´t know why this particular CPU is not giving the expected results in IL-2 VR with the Rift (it should deliver about 66 fps avg but only gives 51).

 

In any case, if you want to have a fluent experience in IL-2 you could do the following:

 

1.- Try to OC your CPU as much as your CPU temperatures allow.(what cooling system do you have?)

2.- Do and report six runs of VR tests: With LOW settings (SS=1.0 and SS=1.7), with BALANCED (SS=1.0 and SS=1.7), with HIGH (SS=1.0 and SS=1.7)

3.- Then, it you can afford it, we will see if you would need a better CPU or GPU to have a better fps.

reporting again:

 

i5-4690k@4.5Ghz

12Gb RAM @1600 (799MHz)

SSD

zotac nvidia 1070 stock

 

CPUmark after OC: 

8571

single threaded 2630

 

Samsung Odyssey

SteamVR says it renders:

2x 1428x1777@ SS1.0

2x 1862x2317@SS1.7

FRAPS:

low SS 1.0
2017-12-04 16:40:24 - Il-2
Frames: 2789 - Time: 60000ms - Avg: 46.483 - Min: 44 - Max: 70
 
low SS 1.7
2017-12-04 16:44:53 - Il-2
Frames: 2872 - Time: 60000ms - Avg: 47.867 - Min: 40 - Max: 80
 
balanced SS 1.0
2017-12-04 17:20:45 - Il-2
Frames: 2733 - Time: 60000ms - Avg: 45.550 - Min: 43 - Max: 61
 
balanced SS 1.7
2017-12-04 16:47:46 - Il-2
Frames: 2706 - Time: 60000ms - Avg: 45.100 - Min: 43 - Max: 54
 
high SS 1.0
2017-12-04 17:03:03 - Il-2
Frames: 2683 - Time: 60000ms - Avg: 44.717 - Min: 43 - Max: 46
 
high SS 1.7
2017-12-04 16:52:31 - Il-2
Frames: 2686 - Time: 60000ms - Avg: 44.767 - Min: 43 - Max: 46
 
 
I ve noticed that Im affected by the fov/ipd bug, so my hmd was showing bit more image than the screenshot in the first post.
 
 

 

full passmark:

93351204872.png

Share this post


Link to post
Share on other sites
Samsung Odyssey SteamVR says it renders: 2x 1428x1777@ SS1.0 2x 1862x2317@SS1.7

 

Thank you for reporting these tests with the Odyssey, but before going to your topic it surprises me the reported Odyssey resolution in SteamVR for SS=1 and SS=1.7.

 

I made some numbers shown in this image:

post-18865-0-66577000-1512427786_thumb.png

The strange thing is that the Odyssey has 2 panels of 1440x1600, but the rendered pixels for SS=1 are 2x1428x1777.

What?? they do subsampling in the horizontal!!

That´s very different from what the Rift do. (2x1080x1200 physical panels and 2x1344x1600 pixels at SS=1)

 

So, the number of pixel for SS=1 is 4.3Million for Rift and 5Million for the Odyssey.

So in terms of pixels rendered and GPU load, running the Rift with SS=2.0 (SteamVR) is equivalent to run the Odyssey with SS=1.7. But obviously the image in the Odyssey will be much better since it is down-sampled less to fit in the better physical display.

Edited by chiliwili69

Share this post


Link to post
Share on other sites

 

 

low SS 1.0 2017-12-04 16:40:24 - Il-2 Frames: 2789 - Time: 60000ms - Avg: 46.483 - Min: 44 - Max: 70   low SS 1.7 2017-12-04 16:44:53 - Il-2 Frames: 2872 - Time: 60000ms - Avg: 47.867 - Min: 40 - Max: 80

 

It is a bit strange that you achieve a Max fps of 80 with SS=1.7 than with SS=1.0 and Max=70.

Be sure you switched ASW off and follow all instructions as the first post.

Nevertheless, I think all these tests tell us two things:

 

1.- The SS is not really affecting significantly the avg numbers, so your GPU is not really stressed. So keep your 1070.

 

2.- Even at LOW graphics settings your avg fps does not increase significantly and your system can not even reach 90 fps at any moment. So despite you have a reasonable ST Passmark (2630) your CPU is not performing well with IL-2 in VR. I don´t know the reason but it is the same case than the doog442 with the same CPU than you.

 

You can spend some more time trying to know why this CPU is not performing as expected. For example you can run the test with monitor as explained in the post.

 

If you can not solve you will need a rig upgrade:

 

1. Upgrade CPU: You can go to 4790K if you want to keep your Mobo, but as the new 8700K is in the market at almost same price, I would go for 8700K with a new Mobo and DDR4 ram

2. Upgrade RAM: Try to go above 2600 MHz 

Share this post


Link to post
Share on other sites

It is a bit strange that you achieve a Max fps of 80 with SS=1.7 than with SS=1.0 and Max=70.

Be sure you switched ASW off and follow all instructions as the first post.

Nevertheless, I think all these tests tell us two things:

 

1.- The SS is not really affecting significantly the avg numbers, so your GPU is not really stressed. So keep your 1070.

 

2.- Even at LOW graphics settings your avg fps does not increase significantly and your system can not even reach 90 fps at any moment. So despite you have a reasonable ST Passmark (2630) your CPU is not performing well with IL-2 in VR. I don´t know the reason but it is the same case than the doog442 with the same CPU than you.

 

 

I would suggest to not put too much stock in that based on that test track.

As I found, running that track was no where near indicative of the actual performance I get flying missions in BoS. If I went by my results then I could say my system can not even reach 90 fps at any moment,   where in fact flying many combat missions , and many hours in the game I am at 90 fps the bulk of the time, dropping into the 80's and 70's occasionally during very heavy combat but still very smooth and fluid. And I am 182 missions into a PWCG campaign with a lot of activity going on.

 

Performance for me in the actual game itself does not even compare to what I got running that track file, it is way better.

And how it performs flying the sim, is what really counts. 

Share this post


Link to post
Share on other sites

I would suggest to not put too much stock in that based on that test track.

As I found, running that track was no where near indicative of the actual performance I get flying missions in BoS. If I went by my results then I could say my system can not even reach 90 fps at any moment,   where in fact flying many combat missions , and many hours in the game I am at 90 fps the bulk of the time, dropping into the 80's and 70's occasionally during very heavy combat but still very smooth and fluid. And I am 182 missions into a PWCG campaign with a lot of activity going on.

 

Performance for me in the actual game itself does not even compare to what I got running that track file, it is way better.

And how it performs flying the sim, is what really counts. 

 

 

Completely agree.  While the benchmark test is useful for a variety of comparisons, you're very likely to get better performance under normal game play.  For example, with my fairly maxed out settings (including using OVRdrop to inject the map into the game in VR), I got the following on the VR test track during one set of tests:

 

Avg: 52.383 - Min: 43 - Max: 90

 

but with the same settings I was getting more like the following over the length of a typical PWCG mission:

 

Avg: 82.587 - Min: 12 - Max: 91  (and I think the 12 may have been when I crashed, IIRC).

 

Not a completely scientific, apples to apples comparison, but a good indication that as dburne says you may well be able to have a very acceptable VR experience actually playing the game even if your benchmark test results are a bit marginal.

Edited by TG-55Panthercules

Share this post


Link to post
Share on other sites

That RAM is a real stinker. no question. Buy some new DDR4 16GB min and you'll be in the game.

Share this post


Link to post
Share on other sites

Hello!

I am new to setting up my PC for playing VR. I have made som changes but I feel unsure about OC and stuff. I have been playing WT for a long while white the Rift and used oculus tray tools to SS. Now I have reed that in IL2 Bos you don't use tray tools.

 

I would really appreciate som help to set up my computer to IL2 and help me get those extra FPS.

Right now when I play IL2 Bos I get many 40-50 FPS. 

I would really be happy if some body could tell me how I can boost the FPS and graphics to max with my rig

 

I have a:

 

i7-7700

3.6 GHz (base), 4,2 GHz (Turbo)

16 Gb DDR4

Nvidia GeForce GTX 1070

Share this post


Link to post
Share on other sites

Hello!

I am new to setting up my PC for playing VR. I have made som changes but I feel unsure about OC and stuff. I have been playing WT for a long while white the Rift and used oculus tray tools to SS. Now I have reed that in IL2 Bos you don't use tray tools.

 

I would really appreciate som help to set up my computer to IL2 and help me get those extra FPS.

Right now when I play IL2 Bos I get many 40-50 FPS. 

I would really be happy if some body could tell me how I can boost the FPS and graphics to max with my rig

 

I have a:

 

i7-7700

3.6 GHz (base), 4,2 GHz (Turbo)

16 Gb DDR4

Nvidia GeForce GTX 1070

 

I use the Oculus Tray Tool with a profile set for BoS, and it works very well.  ASW set to off, switches audio to my Rift when Oculus Home starts.

 

You might list how you have your graphics options set up. Don't use the HUD, it is a performance killer. No SSAO or HDR. 

In Nvidia Control Panel, make sure you have it set to " prefer maximum performance", I think by default it sets to " Optimal Performance".

Share this post


Link to post
Share on other sites

 

 

In my opinion we may get more expressive results for the most modern machines by creating a secondary, higher demanding set of settings for running the test track. That way we can see much easier how the different CPUs and clocks behave at a heavier load of Ultra+High Shadows+4x terrain. The asymptotes for CPU or GPU or RAM clock you mentioned are more diagonal that way and not largely leveled out yet, hence way more expressive. I like the way you think, are you also Engineer?

 

I wanted to be astronaut as many of us. So first stop is to be a fighter pilot. Since shortsighted (lasered 15 years ago) could not make my dream true, so study something else like electrical engineering...but we have IL-2 VR to make our dreams a reality!!!

 

Anyhow, back to topic, yes, I also think than someday next year we have to create an even more demanding track to measure top rigs. We will need to define what settings we should use for this top rigs test. We also might have to create a new track with another plane with mirrors (Spifire for example) and some mountains and clouds around. We can start to think how to design this top rig track.

Share this post


Link to post
Share on other sites

 

 

I would suggest to not put too much stock in that based on that test track. As I found, running that track was no where near indicative of the actual performance I get flying missions in BoS. If I went by my results then I could say my system can not even reach 90 fps at any moment,   where in fact flying many combat missions , and many hours in the game I am at 90 fps the bulk of the time, dropping into the 80's and 70's occasionally during very heavy combat but still very smooth and fluid. And I am 182 missions into a PWCG campaign with a lot of activity going on.   Performance for me in the actual game itself does not even compare to what I got running that track file, it is way better. And how it performs flying the sim, is what really counts. 

 

You are quite right here. The VR test is not really representative of the normal average fps of a full mission. It only represent the heavy combat side of that. The 1 minute long track of the VR test is taking place in Stalingrad where a Yak is hitting three He-111 in just one minute!. A lot of smoke and many planes around. So the purpose of the VR test is just to make measures and compare rigs and performance hits of settings.

 

I think in your particular case your VR test results were showing a worst performance than expected (I really don´t know why), but even thought you had 88 Max fps.

 

The poster above (Mick_00) were having a max of 80 at LOW settings! and the avg was 49. I am sure that you have much better results with the VR test in LOW settings.

 

If you go to the other tabs in the results spreadsheet, you will see that I had 57.9 at ULTRA but 89.4 at LOW. That´s why I was recommending to go for a rig upgrade.

 

In any case, the fps tolerance is very subjective. Some people is fine with 50 fps all time, but other need 80 for a confortable experience. The best thing is to try with missions and normal combat scenarios and if the user is fine with the experience then forget all about this VR test. We just make it to take objective and informed decisions about the optimum settings and hardware investment. And also to compare our rigs with our peers.


 

 

but with the same settings I was getting more like the following over the length of a typical PWCG mission:   Avg: 82.587 - Min: 12 - Max: 91  (and I think the 12 may have been when I crashed, IIRC).

 

This is because 90% of your mission you are not in a combat situation (firing, smoke,  closer planes around, etc) and you are at 90fps most of the time.

Only in heavy moments your fps go below 90 and then your avg start to count below 90. but just for an small fraction of the whole mission.

The purpose of the VR test was to analyse the performance exactly in this heavy scenario.


 

 

I would really be happy if some body could tell me how I can boost the FPS and graphics to max with my rig   I have a:   i7-7700 3.6 GHz (base), 4,2 GHz (Turbo) 16 Gb DDR4 Nvidia GeForce GTX 1070

 

In this post there is some guidance to new VR users:

 

https://forum.il2sturmovik.com/topic/31914-vr-flight/?p=538703

 

It is a pitty that it is a 7700 and not a 7700K which allow Overclock.

 

With the 7700 the only option is to use always at turbo freq (4.2) freq.  You can run the VR test to determine how well your CPU is performing vs the 7700K. 

Share this post


Link to post
Share on other sites

 

This is because 90% of your mission you are not in a combat situation (firing, smoke,  closer planes around, etc) and you are at 90fps most of the time.

Only in heavy moments your fps go below 90 and then your avg start to count below 90. but just for an small fraction of the whole mission.

The purpose of the VR test was to analyse the performance exactly in this heavy scenario.

 

Precisely why that test is not a good indication of the VR performance one is going to get in the game.

Share this post


Link to post
Share on other sites
Precisely why that test is not a good indication of the VR performance one is going to get in the game.

 

Well, people using this game want to have a smooth experience not only when the fly in open sunny sky, no planes around and enjoying the awesome views...

But also when the thing get more interesting and you throw some bullets to your enemy.... in fact in this heavy combats is when you want to max your fps.

 

It should be noted also that although it is a heavy scenario, the VR test is using HIGH settings but with most of the other settings at normal or not maxed out. Running the VR test with ULTRA and maxing out all other settings the fps will decrease considerably.

 

As you know most of the Oculus games are designed to NEVER go below 90fps with a min spec Oculus rig. Even at the most heavy scenario.

Edited by chiliwili69

Share this post


Link to post
Share on other sites

Precisely why that test is not a good indication of the VR performance one is going to get in the game.

 

Being at the lower end of scale on here I know where I'd rather be....I think  :biggrin:  

 

If we use this test track as a tool for those wishing to upgrade I think its a decent indicator. I'm probably due that next step with my i5 4690K having seemingly squeezed every last bit of juice out of it.

Share this post


Link to post
Share on other sites

Well, people using this game want to have a smooth experience not only when the fly in open sunny sky, no planes around and enjoying the awesome views...

But also when the thing get more interesting and you throw some bullets to your enemy.... in fact in this heavy combats is when you want to max your fps.

 

 

 

Yes, and I too in all my campaign flying - quite a lot of hours - have very smooth and enjoyable experience during the heavy combat situations as well. Certainly not the performance that test showed for me. ( vid card was actually downclocking). During my campaign flying my vid card stays at full boost.

 

Certainly a new rig with a 7700k or 8800k and DDR4 ram would be even better for me, but likely not as much as the test results show.

 

Maybe consider creating a test track that is more in line with how we would normally fly our missions.

Edited by dburne

Share this post


Link to post
Share on other sites

I think the test track is fine. Once you make the switch to multiplayer you will find you take another fps hit compared to single player. At least that's my experience.

 

I don't want to scare anyone off getting VR, it was awesome even running a 2500k @4.5 and a 970 video card. But the reason for the test is more to find out what hardware or software changes can help during intense compute requirements.

 

Showing a scene where we all get a gleeful 90fps the whole time won't help that goal.

Share this post


Link to post
Share on other sites

Showing a scene where we all get a gleeful 90fps the whole time won't help that goal.

 

That was not what I was suggesting either... 

Back to some awesome campaign flying for me.

Edited by dburne

Share this post


Link to post
Share on other sites

 

 

You are quite right here. The VR test is not really representative of the normal average fps of a full mission. It only represent the heavy combat side of that.

 

This is what people should take into account when running the test. Having an average of 55 or 60 fps doesn´t mean that your rig is not valid to play IL-2 in VR. 

With this test measurements we might be unconsciously scaring people who are approaching IL-2 VR.  And this is something I really want to avoid in all ways. Since IL-2 VR is game changer in all senses.

 

So, I will add a note to the first post to state that.

 

If someone create a record file of a "normal campaign" (about 10 or 15 minutes should be enough, using same settings than test) I can include it in the first post and people could also run that optionally to have a more realistic idea about the avg they will achieve in a normal campaign.

Share this post


Link to post
Share on other sites

You don't need to soften up feelings mate. A test only has value if it taxes.

 

It's the simple truth that we need a modern system to run VR properly.

 

 

People who buy VR with +3 year old CPUs and gtx970 cards and then get a horrific experience with stutter and nausea, and no >2 SuperSampling values, while complaining about a low-resolution pixelated image , they are what is cutting the VR market.

 

 

I have heard so many times "I will wait for better resolution" as a comment to why people did not buy VR yet - They simply have no idea what a good GPU and proper settings do to quality, and base their ideas on a badly set-up demo in an electronics market, or badly set-up system of a friend, or "because someone else said that". Fortunately there is a small percentage who thinks and experiences on their own. 1070 or even 1080 is the entry. Going with SteamVR SS >2.9 at 90Hz HMD output with min at 60Hz is mandatory for a proper picture - one that doesn't kill IDing and spotting due to ingame AA&Sharpen settings.

Edited by 2./JG51_Fenris_Wolf

Share this post


Link to post
Share on other sites

I dont have VR and dont plan to get it any time soon, at least not until next generation drops, but after getting today my AIO and overclocking this and that, I've decided to run track and see where do I stand. Score is obviously not comparable but as a peculiarity, perhaps someone will take a note of it.

Rig:

GPU- 1080Ti@2025 Mhz core

CPU - Core i7 2600k@5005 Mhz at 1.41 Vcore (will try tomorrow see if I can drop below that, maybe 1.40 or even 1.39 would be possible)

RAM - 24 GB at 1600 Mhz, CL11

Drive-  SSD

Il-2 BOS with reshade on (I couldnt be bothered to turn off filters, if anything it affects performance and gives even lower scores) at max settings and 3440x1440 resolution.

 

Passmark score:

CPU mark - 11449

Single threaded - 2705

 

Frames: 5517

Min: 70

Max: 102

Avg: 92

 

Tomorrow will lower settings to match the first post ones and see what scores I get then. 

Share this post


Link to post
Share on other sites

Hi all

 

 

I just ran the benchmark after a significant amount of testing time to get the game running at a feasible performance (campaign blazing steppe is horrible in VR, so far).

 

My rig

i7 6700K @ 4 GHz (vanilla style, i.e., no OC)

1070 GTX (no OC)

MB: Asus Z 170-E

16 GB ram

 

Frames: 3676 - Time: 60000ms - Avg: 61.267 - Min: 44 - Max: 89

Passmark CPU Test: 11957, Single Thread 2480

 

In short, this is unplayable to me.

 

Do you veterans believe that my system is performing less than it should based on the specs?

Should I investigate the possibility of over clocking the CPU? I did some OC in the late 90's /  early 2000s. My experience is that OC does virtually nothing with respect to FPS in videogames.

 

Additionally I noticed that the expected FPS in the google doc indicates that it is a function of single thread performance only. Why is it that the GPU has virtually no impact on  average FPS?

 

 

   Exceptionally glad if someone could shed some light on this / Fortran (g77)

Edited by Fortran

Share this post


Link to post
Share on other sites

Hi all

 

 

I just ran the benchmark after a significant amount of testing time to get the game running at a feasible performance (campaign blazing steppe is horrible in VR, so far).

 

My rig

i7 6700K @ 4 GHz (vanilla style, i.e., no OC)

1070 GTX (no OC)

MB: Asus Z 170-E

16 GB ram

 

Frames: 3676 - Time: 60000ms - Avg: 61.267 - Min: 44 - Max: 89

Passmark CPU Test: 11957, Single Thread 2480

 

In short, this is unplayable to me.

 

Do you veterans believe that my system is performing less than it should based on the specs?

Should I investigate the possibility of over clocking the CPU? I did some OC in the late 90's /  early 2000s. My experience is that OC does virtually nothing with respect to FPS in videogames.

 

Additionally I noticed that the expected FPS in the google doc indicates that it is a function of single thread performance only. Why is it that the GPU has virtually no impact on  average FPS?

 

 

   Exceptionally glad if someone could shed some light on this / Fortran (g77)

 

I would think you should get better performance, I have flown the Blazing Steppe campaign on my rig ( specs below ) and got very good performance.

 

Insure you have in Nvidia Control Panel, set to " prefer maximum performance for BoS. I believe by default it sets to optimal, which is not good.

Everything else in Nvidia Control Panel, set to " application controlled".

 

You " should" get very reasonable performance at the High preset, you can try Balanced first to see how it does.

Are you using any SS set , either in Oculus Tray Tool ( if you have Rift), or Steam VR?

I would not go any higher than 1.5 in OTT, or 2.25 in Steam VR. ( do not set in both, only one or the other).

 

Try AA at 2X. No SSAO and no HDR. 

Also , keep the HUD off - it is a real performance killer especially in VR.

Share this post


Link to post
Share on other sites

Hi all

 

 

I just ran the benchmark after a significant amount of testing time to get the game running at a feasible performance (campaign blazing steppe is horrible in VR, so far).

 

My rig

i7 6700K @ 4 GHz (vanilla style, i.e., no OC)

1070 GTX (no OC)

MB: Asus Z 170-E

16 GB ram

 

Frames: 3676 - Time: 60000ms - Avg: 61.267 - Min: 44 - Max: 89

Passmark CPU Test: 11957, Single Thread 2480

 

In short, this is unplayable to me.

 

Do you veterans believe that my system is performing less than it should based on the specs?

Should I investigate the possibility of over clocking the CPU? I did some OC in the late 90's /  early 2000s. My experience is that OC does virtually nothing with respect to FPS in videogames.

 

Additionally I noticed that the expected FPS in the google doc indicates that it is a function of single thread performance only. Why is it that the GPU has virtually no impact on average FPS?

 

 

   Exceptionally glad if someone could shed some light on this / Fortran (g77)

 

nope sounds about right. (currently, you have to play 'balanced' and w lower-ish settings to get 80+ fps in campaigns - i did it plenty)

you should absolutely oc with the right motherboard yes, asap - that alone will get you to high settings. bigtime boost (it aint what it used to be - check out my progression in the chart - same as you basically) the k processors are made for this

Share this post


Link to post
Share on other sites

nope sounds about right. (currently, you have to play 'balanced' and w lower-ish settings to get 80+ fps in campaigns - i did it plenty)

you should absolutely oc with the right motherboard yes, asap - that alone will get you to high settings. bigtime boost (it aint what it used to be - check out my progression in the chart - same as you basically) the k processors are made for this

 

 

Hi

 

Thank you for answering. I installed a 1080 ti and it took the average fps from 61 to 72. However still not good enough for playing at high settings, as you indicated. I have to look into the whole OC thing again.

 

Question though, with respect to the spread sheet: you report 2 different single thread numbers before you OC, 2153 and 2354 (user vs. public). How are they derived?

Share this post


Link to post
Share on other sites

Hi

 

Thank you for answering. I installed a 1080 ti and it took the average fps from 61 to 72. However still not good enough for playing at high settings, as you indicated. I have to look into the whole OC thing again.

 

Question though, with respect to the spread sheet: you report 2 different single thread numbers before you OC, 2153 and 2354 (user vs. public). How are they derived?

 

user, by using the passmark benchmark software. public,Im not sure, perhaps a public average expectation

Share this post


Link to post
Share on other sites

 

 

Frames: 5517 Min: 70 Max: 102 Avg: 92

 

Thank you for posting your results with just monitor. I put them in the monitor tab. Did you used HIGH or ULTRA preset?

Reaching 5.0 with your 2600K is a nice OC.

Based on your monitor results in 4K, I think your system will be quite valid for VR. Have you tried VR?

 

For comparative purposes, it is better to run the test using the settings of the first post. And then optionally try other higher or lower settings.

Share this post


Link to post
Share on other sites

 

 

i7 6700K @ 4 GHz (vanilla style, i.e., no OC) 1070 GTX (no OC) MB: Asus Z 170-E 16 GB ram

 

Thank you for posting your test. What RAM speed do you have? (use CPU-Z to know that) 

 

Regarding OC, I think your MoBo is well prepared for OC. So, if you have a decent CPU cooler there is not reason for not doing it. It is free. (unless you burnt the CPU which nowadays is very difficult since CPUs have their own Power and temperature protections systems)

 

Some months ago I didn´t know anything about overclocking, and now it is a routine thing everytime I play IL-2 in VR.

 

You can read more here:

https://forum.il2sturmovik.com/topic/29881-overclocking-4790k-better-bos-performance/

Share this post


Link to post
Share on other sites

 

 

Additionally I noticed that the expected FPS in the google doc indicates that it is a function of single thread performance only. Why is it that the GPU has virtually no impact on  average FPS

 

We created this thread with the purpose of analyzing what are the limiting factors (hardware or settings) for IL-2 in VR.

We discover that CPU was the most limiting factors, so people with the best GPUs (1080Ti) but worse CPU were not achieving a good fps.

We also saw that the important factor for IL-2 VR is the single-thread performance. We used the public Passmark benchmark to compare (but I believe any other public CPU benchmark reporting single-thread performance can be used as well).

 

If the CPU is the bottleneck, then it doesn´t matter how good is the GPU. 

Share this post


Link to post
Share on other sites
Thank you for answering. I installed a 1080 ti and it took the average fps from 61 to 72

 

Hey! you bought the top GPU before trying OC! :o:

Anyhow, with that GPU you will be able to run higher levels of SS (2.5 or 3.0 in SteamVR) and you visual quality with be better:

 

https://forum.il2sturmovik.com/topic/30771-how-much-ss-desirable-image-quality-samples/

 

Could you please report your full test results with the new card and no OC? (min, max, avg, frames)

Edited by chiliwili69

Share this post


Link to post
Share on other sites

 

 

public,Im not sure, perhaps a public average expectation

 

The public passmark numbers (CPU Passmark and Single-Thread CPU passmark) are taken from the Passmark webpage.

They are an average for all the test performed by the users with no use of OC (but CPU turbo is allowed):

 

Just go here a search your CPU:

https://www.cpubenchmark.net/cpu_list.php 

 

your 6700K is here:

https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-6700K+%40+4.00GHz&id=2565

 

With 9337 samples (from users) It has a CPU passmark of 11110 and a CPU STMark of 2351. These numbers might change a little bit due to new users reporting samples to Passmark.

 

Your 6700K has a base clock of 4.0 and 4.2 with Turbo (this turbo is not considered as real overclocking by Passmark).

 

I think that when you (and fortran) were running the passmark test (and also IL-2 test), the real freq of the CPU was 4.1 or 4.2, that´s why you had a higher than average STMArk.

You can try to run multiple test of the CPU Passmark with diferent OC (4.0, 4.1, 4.2 .. until your max OC).

 

You can see that the more OC the more STMArk you achieve:

https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/?p=523673

Share this post


Link to post
Share on other sites

Just a comment on the Passmark test. Although it seems to provide a base for "expectations" regarding the performance in BoX, I would say it is of note that this is only the case between CPUs of a similar class.

 

I just finished building an 7900X system and I see things diverge there. The Passmark is a very, very, very short test. CPU frequency is very dynamic and thermal capping is an issue.

 

After an overnight torture run with Prime95, I ran Passmark on the 7900X that I clocked to 4.5 GHz. Give or take. If you take a look at the cores, the frequencies vary between 4.2 and 4.7 GHz, being mostly around 4.4 GHz. If you go above that with you need a SERIOUS cooler, nothing you could get of the shelf. At around 4.4 GHz on all 10 cores (20T), my be quiet! 280mm AIO cooler barely manages to keep temps below 90°C.

 

But at least, it runs stable. With the boiling water inside, running Passmark I get this:

passmark.jpg

 

Now, turning back to my old system with a 4770K @ 4.4 GHz, cleaning it out, new thermal paste, open cage and run passmark on a COLD air cooler. I get this:

 

passmark_old.jpg

 

Oh, my, almost 10% less for single thread performance on the 7900X set at basically the same frequencies than the old 4770k?

 

Answer is thermal throttling. If I had ran Prime95 runs on the old system with a hot cooler, I doubt I would reach those numbers. In BoX, thermal throttling is not so much an issue, as if your system can be cooled on 10 cores, no matter what you do on 2 cores, it will be nothing to worry about regarding heat.

 

When I start BoX (monitor, 1440p, ULTRA, all maxed out), I get about 20% increase in FPS on the 7900K than in the old system with the identical GPU. BoX cannot really put heat on the system like Prime95, and if it is basically just 2 cores that get a hammering, they clock nicely.

 

The 8700K really seems to be a good option if BoX is your only concern. If it isn't, The multithread readings of passmark tells you what you can expect from the 7900X. But with that one, you really pay a premium for having a workstation chip that is somewhat able to run BoX. Besides it still being a rushed product. Ah.. competition...

 

Share this post


Link to post
Share on other sites

Hey! you bought the top GPU before trying OC! :o:

Anyhow, with that GPU you will be able to run higher levels of SS (2.5 or 3.0 in SteamVR) and you visual quality with be better:

 

 

 

Heck I went from a 1080 to a 1080 Ti, and was thrilled with the performance increase I got in playing BoS in VR ( along with other games as well)

Share this post


Link to post
Share on other sites

 

 

The Passmark is a very, very, very short test.

 

Passmark is just a performance test, not an stress test. To measure the performance you don´t need a long test.

 

It is like a Lamborghini car. To test the top speed you only need to speed up to 350Km/h for few seconds, and that´s all.

Now, another topic would be to maintain those 350Km/h during several hours.

 

So, a performance test doesn´t need to be long.

A stress test (like Prime95 or others) need to be long to be sure the system is stable and don´t overheat.

Share this post


Link to post
Share on other sites

 

 

After an overnight torture run with Prime95

 

What torture test do you use? Bench? How many threads? 4?

Why are you running the torture test for 1 full night? is it not enough with 2-3 hours? 

 

I was also using Prime95 some months ago with my previous aircooling, reported here:

 

https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/?p=514455

 

If your goal is to run IL-2, you will see that IL-2 (in VR) is much less demanding than the Prime95 test in terms of CPU heating generation.

So, if you goal is just IL-2, you could achieve a higher stable overclocking not capped by thermal throttling or CPU power limits.

Share this post


Link to post
Share on other sites

 

 

With the boiling water inside, running Passmark

 

Can you try to run the CPU Passmark with the 7900X with no boiling water inside? (just to be sure there is not thermal throttling) 

 

BTW, what software or technique do you use to verify that there is no thermal throttling during a performance test?

Share this post


Link to post
Share on other sites

 

 

Although it seems to provide a base for "expectations" regarding the performance in BoX, I would say it is of note that this is only the case between CPUs of a similar class.

 

In this post, I showed that different CPUs delivers different STMark values for the same CPU speed:

 

https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/?p=523673

 

You could do a series of Passmarks tests with the 7900X at different fixed speeds (4.0, 4.1, etc) being sure there is not thermal throttling. 


 

 

I get about 20% increase in FPS on the 7900K than in the old system with the identical GPU

 

We didn´t get any test results for the 7900X. It would be interesting to know how it performs with the IL-2 VR test, but using only the monitor. The instructions are in the first post. 

Share this post


Link to post
Share on other sites

 

 

Heck I went from a 1080 to a 1080 Ti, and was thrilled with the performance increase I got in playing BoS in VR

 

If you got an fps increase it was because your old 1080 was fully loaded. Probably because you used 1.5 SS (or more) in OTT (which is  2.25 in SteamVR) and this could be a bit too much for a 1080 card.

As you know, the SS impacts heavily the fps you achieve and everyone have to find their own limit to keep GPU below 100%.

 

It is a compromise of fps versus quality:

 

https://forum.il2sturmovik.com/topic/31307-how-ss-decreases-your-fps-testing-results/

Share this post


Link to post
Share on other sites

If you got an fps increase it was because your old 1080 was fully loaded. Probably because you used 1.5 SS (or more) in OTT (which is  2.25 in SteamVR) and this could be a bit too much for a 1080 card.

 

 

No not really. 1080 Ti is just a nice improvement over the stock 1080.

Share this post


Link to post
Share on other sites

Thing is, I can play quick missions (16 planes, lots of clouds etc.) with what I believe is a good set of settings, i.e., not pushing it too hard with the i7+1080ti, e.g., High general, SS 1.4, not too much on the other parameters. However, it's the Blazing Steppe campaign that really hurts the gameplay. As a simple test, I set the game to low settings and ran the benchmark. I ended up with the following:

 

Frames: 5359 - Time: 60000ms - Avg: 89.317 - Min: 87 - Max: 90

 

However, with the same settings, mission 3 in the campaign stutters. It's not a lot really, but it's enough to break the immersion. It might not be such a surprise since in that mission you have a full squadron of Stukas bombing stuff, Stalingrad is on fire and there is fighting going on on ground level. Additionally, you have an abundance of other planes as well.

 

In short, I don't believe that even a successful overclock would solve my issue :( 

Share this post


Link to post
Share on other sites

Just for grins, I ran another benchmark test today with the usual VR benchmark settings except that I turned on the dynamic resolution setting at .8. Wanted to see how it compared to the last test I ran about a month ago, but this time with more recent nVidia drivers (388.31).

 

The results were basically identical (slightly better min FPS today):

 

Previous test:  Frames: 5236 - Time: 60000ms - Avg: 87.267 - Min: 63 - Max: 91

 

Today's test:   Frames: 5236 - Time: 60000ms - Avg: 87.267 - Min: 66 - Max: 91

 

Gonna be running some further tests of the new Oculus Home/Dash beta, but wanted to get an updated baseline for the newer drivers and also to see if the dynamic resolution feature would help any.  At least at these benchmark settings, it doesn't look like the dynamic resolution thing is helping any.  But I'm gonna try testing with my mostly-maxed-out settings to see if maybe dynamic resolution will matter more under those conditions.

 

[EDIT] Intriguing - when I ran the benchmark test track with my mostly-maxed-out settings but with dynamic resolution turned on at .8, I saw quite a difference from the tests I ran ran last month:

 

Last month (without dynamic resolution on, i.e., at 1.0):  Frames: 3922 - Time: 60000ms - Avg: 65.367 - Min: 44 - Max: 91

 

Today (with dynamic resolution on at .8):  Frames: 4796 - Time: 60000ms - Avg: 79.933 - Min: 48 - Max: 91

 

Now I've got to fly a few missions and see if I can tell any visual quality difference when the dynamic resolution thing kicks in.  Also gonna have to try some tests with OVRdrop on to see if the improvement in FPS shows up under those conditions as well.

 

[EDIT 2]  OK, now I'm really confused.  I ran the test again a couple of times using my mostly-maxed out settings and with OVRdrop on and map injected (my normal flying settings), but with dynamic resolution turned on at .8, but unlike above I saw absolutely no FPS improvement over my tests from last month (in fact, both test runs slightly worse, but not by much).  Gonna have to try a few more things to see if I can figure that out - would like to have that extra 15-20% FPS that the test above indicated, especially as I didn't really notice much if any visual degradation when I ran one of the tests with the VR headset on just to see how it looked.

Edited by TG-55Panthercules

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...