Jump to content
chiliwili69

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

Recommended Posts

4 hours ago, lightbulbjim said:

Just ran this benchmark.

 

System Specs

  • CPU: i5 9600K (listed as 3.7 GHZ but CPU-Z shows it running at 4.3 GHz, even at idle)
  • RAM: 16GB 2600 Mhz, 19-19-19-43 (I know, timings could be better)
  • GPU: GTX 1080
  • Storage: Samsung 970 EVO SSD 500 GB
  • HMD: Oculus Rift CV1
  • IL-2 Version: 3.010c

 

Benchmark Results

  • Frames: 7020
  • Avg: 58.500
  • Min: 42
  • Max: 88
  • CPU Mark score: 13549

 

Notes

  • When I set the SteamVR resolution to 150% the OTT overlay was reporting a pixel density of 1.23. To get the required 1.32 pixel density I had to use a SteamVR resolution of 172%. 172% was used for the benchmark.
  • Usually I fly with OpenComposite, pixel density 1.0, in-game AA 4x, ASW auto, and I'm happy with the performance.
  • I should really experiment with overclocking my CPU a bit. Everything is running at stock speeds for now. I have a Hyper T4 cooler, so there's probably some thermal headroom there.
  • I should have bought better RAM.

 

Then first overclock your Ram and see if this does have any effect. 

 

Share this post


Link to post
Share on other sites
On 3/2/2019 at 12:55 PM, blitze said:

The newer versions of my Laptop might also do better in CPU Mark as they are either a i9-8950HK or i7-8750H CPU's

 

Yes, those CPUs are really good for a laptop at single-thread performance, specially the i9-8950HK that reach 4.8GHz on Turbo.

 

I have taken note of the resolution at 26% SS. I was also asking for the 100% in the Video tab. Could you confirm you achieve these values with 100%, 120%, 150%:

8k.png.ec4cd22023d2950cf1cb6288ed2b63c1.png

 

For run a fair test with the Pimax8K, I think you will need to set the SS to the minimun (20%) which would give you about 6.9 Million pixels, which will be the closest res to 6.5 million (which is our baseline for pixels)

On 3/3/2019 at 11:29 AM, EAF51_Jimmi said:

Checked the spreadsheet  all Odyssey are max capped at 46... 

i feel ther's something wrong in the test metodology with this headset...

 

I have not an Odyssey to test it myself, but in the previous test ICDP was reporting a max of 52 fps with Odyssey.

I think there should be a way to be sure you deactivate the WMR Motion Reprojection, and also the "SteamVR Motion Smoothing" which are equivalents of Oculus ASW.

When you have the ingame counter in a solo flight, verify you have values in between 45 to 90, not just 45 or 90.

 

I believe you test are not fully valid.

On 3/3/2019 at 1:24 PM, Dutch2 said:

“the last update in 2013” FRAPS

 

For sure there are many apps more sophisticated than Fraps, for example fpsVR  which I use it (but it is not free).

 

Fraps just do the job,  just giving the min, max and avg. And also frametimes and fps in csv files if you want.

So far we have used it along these years and it is simple enough to do a simple test.

 

In any case I am open to use any other tool which comply with this criteria:

- Free

- Simple

- Able to run with NVidia and AMD cards

- Able to run for a number of second after pulsing a customizable key.

- Don´t consume extra resources from CPU/GPU.

 

Old software which just do the job is as valid as any new modern tool. Another example is Prime95. It is old and free. And good to stress your CPU.

Share this post


Link to post
Share on other sites
19 hours ago, lightbulbjim said:

When I set the SteamVR resolution to 150% the OTT overlay was reporting a pixel density of 1.23.

 

First at all thank you for posting your test results.

When you set SteamVR_SS to 150%, it is fine with OTT reporting PD of 1.23, since this is the equivalent of 150% in OTT. So you don´t need to run the test at 172%, do it at 150%.

 

Also, you need to report the Single-Thread Mark number (this the important number for this test) as described in the instructions.

 

You have an excellent CPU for overclocking, try to see where is your limit with that cooler. 

Share this post


Link to post
Share on other sites
7 minutes ago, chiliwili69 said:

 

First at all thank you for posting your test results.

When you set SteamVR_SS to 150%, it is fine with OTT reporting PD of 1.23, since this is the equivalent of 150% in OTT. So you don´t need to run the test at 172%, do it at 150%.

 

Also, you need to report the Single-Thread Mark number (this the important number for this test) as described in the instructions.

 

You have an excellent CPU for overclocking, try to see where is your limit with that cooler. 

Oops, my bad. Thanks for clarifying.

 

I just ran CPU Mark again and I get 13952 for the main score and 2762 for the single threaded score.

 

I don't have time to run another VR benchmark right now but I'll do it later today at 150% resolution.

Share this post


Link to post
Share on other sites
7 hours ago, chiliwili69 said:

 

For sure there are many apps more sophisticated than Fraps, for example fpsVR  which I use it (but it is not free).

 

Fraps just do the job,  just giving the min, max and avg. And also frametimes and fps in csv files if you want.

So far we have used it along these years and it is simple enough to do a simple test.

 

In any case I am open to use any other tool which comply with this criteria:

- Free

- Simple

- Able to run with NVidia and AMD cards

- Able to run for a number of second after pulsing a customizable key.

- Don´t consume extra resources from CPU/GPU.

 

Old software which just do the job is as valid as any new modern tool. Another example is Prime95. It is old and free. And good to stress your CPU.

 

You are pushing to much to the age of FRAPS here is my old post, Think the whole idea around this is not quite optimal. For measurements I still think we need FCAT VR and not “the last update in 2013” FRAPS.   

If seeing the video’s about VR measurements in FCAT VR, I think it is clear FRAPS is not telling the whole VR performance story. 

Share this post


Link to post
Share on other sites
7 hours ago, lightbulbjim said:

2762 for the single threaded score.

 

Thank you for the new data.

 

There something strange in your results. You said you overclock to 4.3GHz, since this is the freq you see in CPU-Z. But CPU-Z is not a good tool to monitor CPU freq. It is better to use MSI Afterburner (it is free) and use the trendlines to monitor CPU freq during the test (and many other variables you may want).

If you don´t fix your CPU freq (in BIOS or with OC software) it is possible that the CPU varies the freq depending on the CPU load. So 4.3 could be for idle, and 4.6 the Freq with load which is 4.6 (in fact the 4.6 is the max freq of your turbo).

 

According to your STMark achievement (2762) you should be running around 4.6 GHz, not 4.3. Look to the "STMark vs OC" page on the test online sheet:

stmark.jpg.a9c4d567b743eeec4d858f7ca5f6c004.jpg

Share this post


Link to post
Share on other sites
59 minutes ago, Dutch2 said:

If seeing the video’s about VR measurements in FCAT VR, I think it is clear FRAPS is not telling the whole VR performance story.

 

Here we are talking just about a tool for the IL-2 test, where ASW is set OFF.

So, what exactly FCAT-VR will tell you that log files of FRAPS is not telling you?

 

For the IL-2 test, Ideally I would want to use a tool like fpsVR but free. Which is telling you the frametimes for CPU and GPU.

https://steamcommunity.com/app/908520

 

The current problem of the IL-2 test with FRAPS is that you don´t know if the constraint is due to CPU or GPU.

 

Please, show us with images and test how to use FCAT and what additional information would be useful to have from the IL-2 test.

Share this post


Link to post
Share on other sites
9 hours ago, chiliwili69 said:

 

Yes, those CPUs are really good for a laptop at single-thread performance, specially the i9-8950HK that reach 4.8GHz on Turbo.

 

I have taken note of the resolution at 26% SS. I was also asking for the 100% in the Video tab. Could you confirm you achieve these values with 100%, 120%, 150%:

8k.png.ec4cd22023d2950cf1cb6288ed2b63c1.png

 

For run a fair test with the Pimax8K, I think you will need to set the SS to the minimun (20%) which would give you about 6.9 Million pixels, which will be the closest res to 6.5 million (which is our baseline for pixels)

 

Using PiTool 1, SteamVR Video 100% and App 26% gives 2260 x 1991 target res.  4,499,660 pixels.

 

My normal setting of PiTool 1.75 with SteamVR Video at 20% and App at 50% gives me 2453 x 2160 or 5,298,480 pixels.  It seems to run well and is at native res on the HDM's displays vertical ( horizontal culled a little for the Normal FOV 150 degrees)

 

Anyway - I don't get this SS system with SteamVR - it is not logical and not conducive to getting the best image clarity out of ones headset.  I personally would like settings that can be set that describe the actual render res target post lens correction aka what is sent from the GPU to the Headsets Screens and having SS altering values as a slider with maybe one decimal place.  So I can then see SS as 1 x the native screen resolution of my headset or 1.4 or 2 or 1.8

 

I am still of the opinion that the render data sent from the GPU to the headset needs to be in relation the the native resolution of the headsets screens for us to get the best image.  No different to a monitor.  The Lense Correction cals are pre processes that take place before the GPU crunching not post.

 

At the moment it is a dogs dinner. 🤮

Share this post


Link to post
Share on other sites
Posted (edited)

I’m not going to discus wrong or false and if you like the way you are testing is for me OK. I’m only saying I think there is maybe a better way to determine the VR performance. I admid not sure because I did not test it myself. BTW you do not need FRAPS as Afterburner can also log frametimes, CPU/GPU etc, in a very simple way. 

 

fpsVR is in a frequent Steamsale $2,xx not a mega cost. I do have this tool, very helpful, it brings all the needed info, but it can not log, so rather useless if wanting to compare/tweak your own BoX adjustments. Because I still think that is all about; the best eyecandy at 90FPS straight and to detect you bottleneck. See lot of members publish the adjustments, thats good, because some hidden adjustments will now come to the surface. But to see the impact of all those small steps you need to log everything. 

 

Right now I do not have that much time in testing FCAT VR or even play BoX, keep in mind the whole VR generating techniques are rather complex, so also is FCAT VR and it does need Python software, in this case Anaconda. But I made a printout from all the info at the below sources for my research and did install the software on my BoX boot. If my family and my work is going to an normal level then I can go further. 

 

If wanting to go for more info about FCAT VR then visit the links I have made a print from and watch the video’s, think it is very informative for VR tweaking:

https://www.geforce.com/whats-new/guides/fcat-vr-download-and-how-to-guide

https://www.geforce.com/hardware/technology/fcat/downloads

https://www.tomshardware.com/reviews/vr-benchmark-fcat,4943.html

Benchmarking VR

 

 

 

 

 

Edited by Dutch2

Share this post


Link to post
Share on other sites
1 hour ago, chiliwili69 said:

 

Tha like nk you for the new data.

 

There something strange in your results. You said you overclock to 4.3GHz, since this is the freq you see in CPU-Z. But CPU-Z is not a good tool to monitor CPU freq. It is better to use MSI Afterburner (it is free) and use the trendlines to monitor CPU freq during the test (and many other variables you may want).

If you don´t fix your CPU freq (in BIOS or with OC software) it is possible that the CPU varies the freq depending on the CPU load. So 4.3 could be for idle, and 4.6 the Freq with load which is 4.6 (in fact the 4.6 is the max freq of your turbo).

 

According to your STMark achievement (2762) you should be running around 4.6 GHz, not 4.3. Look to the "STMark vs OC" page on the test online sheet:

 

I haven't done any overclocking yet. I did a bit of experimenting today with Intel XTU  . My CPU is advertised as 3.7GHz with up to 4.6GHz under turbo boost.

 

Under no load I could indeed get it to run at 4.6GHz. When I loaded up all six cores to 100% it hit the TDP of 95W and throttled to 3.7GHz. Loading only some of the cores (or all of them only partially) would let it ramp up the frequency until it either hit 4.6GHz or consumed 95W.

 

Intel XTU allows for adjusting the max allowed TDP. If I raised it above the stock 95W it would happily run at higher frequencies, but to be honest I wasn't comfortable with the cooling capacity of my cooler to run it at anything other than the stock TDP. Even at 100W it got too hot when fully loading all cores, which I want to be able to comfortably do for safety.

 

Now of course, ideally I would leave the TDP limit set at a sensible number for safety and increase the maximum frequency via the multiplier. That way workloads that don't fully saturate all six cores would allow the CPU to run nice and fast until it hit the TDP limit. Unfortunately I have a B360 motherboard so my overclocking options appear to be limited. Intel XTU won't let me adjust the frequency (via the multiplier) above 4.6GHz. My BIOS looks like it has all the right options, but if you input any multipliers which would give a frequency above 4.6GHz it ignores them. So I guess... no overclocking for me.

Share this post


Link to post
Share on other sites

With new update to 3.011 flight record stopped working, error counter #1000. Please werify

Share this post


Link to post
Share on other sites

Yeah, previous flight records doesn`t work any more. 

It is the third time this happen due to changes in engine etc (which BTW are welcome)

 

But I think we already know enough about what influences the IL-2 VR performance.

Therefore i will not work anymore in this post or online sheet

 

As said before, it would be nice if in future the dev team and Jason agreed to include some kind of pattern base flight (availalble always) to measure performance and allow new users to check their hardware/software/settings is according to expectations.

 

It has been a nice two years of testing but I give up. :cray:

  • Thanks 2

Share this post


Link to post
Share on other sites
2 hours ago, chiliwili69 said:

Yeah, previous flight records doesn`t work any more. 

It is the third time this happen due to changes in engine etc (which BTW are welcome)

 

But I think we already know enough about what influences the IL-2 VR performance.

Therefore i will not work anymore in this post or online sheet

 

As said before, it would be nice if in future the dev team and Jason agreed to include some kind of pattern base flight (availalble always) to measure performance and allow new users to check their hardware/software/settings is according to expectations.

 

It has been a nice two years of testing but I give up. :cray:

 

TY very much for your work!

 

1) i think right now interleaved reprojection is always on! You can't disable it anymore in steam vr! If FPS are low it switch the freqeuncy to 45hz and that's why there are some results capped at 45fps!

 

2) IMHO Fligh Sims are super good to stress pc so should be always present in pc reviews! Having a DEMO of the game featuring a proper benchmark inside would be a way to show flight sims to a broader public! Devs think of it!

Share this post


Link to post
Share on other sites
On 3/7/2019 at 8:44 PM, chiliwili69 said:

It has been a nice two years of testing but I give up. :cray:

 

Thanks for your efforts chili ! Your contribution the IL2-VR community is outstanding. I just wish I had the will to carry on your work, a common performance test will always be useful.

Share this post


Link to post
Share on other sites
On 3/8/2019 at 2:44 AM, chiliwili69 said:

But I think we already know enough about what influences the IL-2 VR performance

I hope devs know it too.

 

This topic is 27 pages long (the longest in VR section) but there were no comments from devs about this research.

Share this post


Link to post
Share on other sites
3 hours ago, TUS_Samuel said:

I hope devs know it too.

 

This topic is 27 pages long (the longest in VR section) but there were no comments from devs about this research.

Unfortunately the VR crowd is extremely small, make that the ww2 combat flight sim VR crowd and we are speaking to like two dozen people. They've definitely worked on some things for VR, the plane view limits were extremely limiting and tight before, now they are all pretty good, we've also had performance enhancements not necessarily tied to VR but helped none the less.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...