Jump to content

RHINELAND BENCHMARK MISSION


Recommended Posts

VERSION 1.0

Good for a 4 or 5 minute test right now.

I recommend the external chase camera after autopilot has been activated.

 

RHINELAND BENCHMARK

 

Un-Pause, then hit A (autopilot)

Time Dilation test subtitles will begin shortly after the mission starts. 

3...2...1...Begin. Use a stopwatch and you can check your TD.

 

This isn't likely the final version of the mission, so any note you make about TD will not be relevant after the mission is tweaked pending feedback etc.

 

I deleted a flight of B-25's and some Flak, a few of the complex tanks and thus reduced the TD  to half of what it was initially, as I indicated I will tweak it further but figured I might as well post the first build.

 

 

Eventually I will eliminate the player aircraft and use an animated camera instead - I've just never done this and need to figure out how.

My own TD on the last run (Ultra, 4K) was 1 min mission time, to 1 min 18 seconds real time...at the 2 minute mark it was 2 minutes 38 seconds...pretty harsh still, so I'll tweak it a bit when I get a chance.

 

 

Edited by Gambit21
  • Thanks 1
  • Upvote 1
Link to post
Share on other sites

Nice test. I think this is a very useful project.

 

4 hours ago, Gambit21 said:

My own TD on the last run (Ultra, 4K) was 1 min mission time, to 1 min 18 seconds real time...at the 2 minute mark it was 2 minutes 38 seconds...pretty harsh still, so I'll tweak it a bit when I get a chance.

 

If the goal is to get easily measurable TD across a wide range of hardware, I'd advise against lightening the load. On a rather mediocre system with a i7-6700K, the TD was only about three additional seconds per minute. Such small differences are somewhat difficult to measure accurately with a stopwatch. I suspect they would get even smaller with modern hardware.

Link to post
Share on other sites
On 11/10/2020 at 8:08 AM, Mitthrawnuruodo said:

If the goal is to get easily measurable TD across a wide range of hardware, I'd advise against lightening the load. On a rather mediocre system with a i7-6700K, the TD was only about three additional seconds per minute. Such small differences are somewhat difficult to measure accurately with a stopwatch. I suspect they would get even smaller with modern hardware.

 

Agree. If the goal of this becnhmark is to measure TD then we should not lighten it

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

What means 'time dilation' in this context, please?

 

Slows the game speed down.

1 second in game might be 1.2 seconds in real time for example.

 

If you in a cockpit with a clock with second hand,  you can get pretty close to judging when you are in time dilation by observing the second hand movement.

  • Like 1
  • Upvote 1
Link to post
Share on other sites
1 minute ago, dburne said:

 

Slows the game speed down.

1 second in game might be 1.2 seconds in real time for example.

 

If you in a cockpit with a clock with second hand,  you can get pretty close to judging when you are in time dilation by observing the second hand movement.

 

Yep

In this mission I've used subtitle messages to mark time so that you can compare with your stopwatch.

1 min, 2 min and I think 5 min.

 

In actual heavy mission builds (not trying to cause problems as I've done here) the TD is not nearly this severe, and might not even be noticed by many, but I still try to bring it back to 1:1 when it's detected.

 

 

Link to post
Share on other sites
18 minutes ago, Gambit21 said:

 

Yep

In this mission I've used subtitle messages to mark time so that you can compare with your stopwatch.

1 min, 2 min and I think 5 min.

 

In actual heavy mission builds (not trying to cause problems as I've done here) the TD is not nearly this severe, and might not even be noticed by many, but I still try to bring it back to 1:1 when it's detected.

 

 

 

TD occurs fairly often in PWCG campaigns.

Not too severe but enough to notice if observing that second hand on the cockpit clock.

I spent a lot of time in PWCG today.

Link to post
Share on other sites

I have been testing this benchmark today to try to understand when and why TD occurs.

 

My PC: CPU: 4790K at 4.8GHz,  RAM:16Gb DDR3 at 2933MHz, GPU:1080Ti

I have used your mission with three settings:

 

monitor 4K and with loads in CPU (Ultra, shadows high, no MSAA, etc):

minute 1: 1.05

minute 2: 2:10

minute 3: 3.16

minute 4: 4.24

minute 5: 5.33

 

monitor 4K and with loads in GPU (clouds, MSAA):

1.08,  2.14,  3.23,  4.30,  5.37

 

monitor 1080p and with loads in CPU (Ultra, shadows high, etc):

1.07,  2.14,  3.21,  4.27,  5.37  

 

So it seems that it present regardeless of settings or resolution. This is really strange.

 

Since the speed of the plane is kept constant, we could measure the distance of the travel (maybe using waypoints) and see if the time is distance/speed.

 

Just a note: To start the mission, it is not necessary to write the 3, 2, 1, Begin. Since the mission is initially paused, we can press the P (to unpause) with one hand and with the other start the stopwatch.

 

 

I am also wondering how track records are managing the time, since for the 3 minute remagen track it didn´t appear. (or we were unable to see it).

 

Take a look of the multiples runs performed at different settings:

https://forum.il2sturmovik.com/topic/56485-benchmark-for-cpuram-performance-remagen-4002-to-4005/?do=findComment&comment=861464

 

They were not more or less delayed since all of them superposed well. (the values were shifted vertically due to fps performance but not horizontally)

 

Also the Fraps stopwatch was set to 180 seconds and the info of the recorded track said 3 minutes.

 

I need to test your mission and create a recorded track of 5 minutes of real life time. Just to see if the recorded track info says 5 minutes or less.

Edited by chiliwili69
  • Thanks 1
Link to post
Share on other sites

 

Also, I remember I made a experiment to demostrate that the fps of a recorded track are very similar to the fps we have during the same play game.

https://forum.il2sturmovik.com/topic/56485-benchmark-for-cpuram-performance-remagen-4002-to-4005/?do=findComment&comment=862274

 

Both experiments took 3 minutes of real life. And the recorded track of it said it was 3 minutes. Maybe in that experiment there were not enough planes or tanks to make present the TD misterius effect.

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

 

Also, I remember I made a experiment to demostrate that the fps of a recorded track are very similar to the fps we have during the same play game.

https://forum.il2sturmovik.com/topic/56485-benchmark-for-cpuram-performance-remagen-4002-to-4005/?do=findComment&comment=862274

 

Both experiments took 3 minutes of real life. And the recorded track of it said it was 3 minutes. Maybe in that experiment there were not enough planes or tanks to make present the TD misterius effect.

 

Thanks for your work.

Let me know if you arrive at any suggested adjustments or changes to the mission.

 

P.S. Complex tanks are very hard on CPU resources. In the future it might be useful to creat a small mission with a single aircraft and a number of these tanks to isolate the CPU as much as possible.

 

I appreciate your insight.

Link to post
Share on other sites
  • 2 months later...

Thanks Gambit for this mission.

I have just tested it and i have those times (almost full details)

 

1. 1:00

2. 1:58

3. 3.00

4. 4:00

5. 4:59

 

Full details: are almost the same. 59, 1:59, 2:59, 3:59, 4:59, 5:59

 

But feeling is strange, its not fluid like in QMB or career mode without enemy units where the feeling is like you have 1000 fps.

Also it looks like there is some sort of magnet in front of the cockpit which makes the motion of the track ir heavy.

On heavy missions in career mode its even worse, definitively not fluid.

 

strange.

Link to post
Share on other sites

Exist a standard benchmark track for il-2 Sturmovik: Battle of Stalingrad only?

 

I am trying convince some local hardware channels to include in their benchmarks, one already include MSFS 2020.

 

You guys do "high end" CPU's benchmarks, but I think in more entry level ones, for "new players".

Edited by Sokol1
Link to post
Share on other sites

Weighing in on this as I have been experiencing this crazy time dilation as well. I never noticed it before, I took a month or so break and came back and now I have it. I don't know if it was an upgrade that did it, but nothing has changed on my system hardware wise. 

System Specs:
i7-6700K CPU @4GHz
16GB Ram

SSD Drive

Nvidia GTX 1660 Super

Also I will add that I am noticing an even greater time dilation when recording a track. Here's my numbers.

Not Recording a Track:

1 Min.    1:26
2 Min.    2:52

3 Min.    4:17

4 Min.    5:39

5 min.    7:06

 

While Recording a Track

 

1 Min.    1:45
2 Min.    3:22

3 Min.    4:57

4 Min.    6:34

5 min.    8:12

 

So as you can see it's quite a big slow down for me. I'm experiencing the time dilation in career mode if I have the density set to anything more than Scattered. I also experience it in PWCG campaigns no matter what I set the density to. And for the most part scripted campaigns are now unplayable, especially if I am recording a track. In multiplayer mode I don't experience any detectable time dilation.

Again I never used to have this problem. I hope whatever got changed in the upgrade gets changed back, because this in VERY frustrating!

Link to post
Share on other sites
7 hours ago, AirWolves=CutCut said:

Weighing in on this as I have been experiencing this crazy time dilation as well. I never noticed it before, I took a month or so break and came back and now I have it. I don't know if it was an upgrade that did it, but nothing has changed on my system hardware wise. 

 

Here are my numbers from today with a fairly similar system (not recording):

 

i7-6700K overclocked to 4.6GHz

16GB memory at 3200C14

970 EVO 1TB SSD

GTX 1080

2560x1440 Ultra settings (4x MSAA)

 

1 min. 1:02

2 min. 2:04

3 min. 3:07

4 min. 4:11

5 min. 5:17

 

As you can see my results have not changed since 9. November - still roughly three seconds per minute time dilation.

 

I'm not sure what's causing the huge difference between our results (barely noticeable to unplayable). Maybe it's the overclocks and faster RAM. Will do more testing with stock CPU speeds later.

Edited by Mitthrawnuruodo
Link to post
Share on other sites

i tried it on my old system i5 4670k and gtx980. 1920x1080

There is no impact of time dilation at all through the graphic settings.

even on lowest settings possible i have always:

 

1 min.    1:15

2. min     2:30...

 

its a shame that you cannot free cpu power at least with one setting.

 

and what about AI difficulty? is moderate or ace difficulty in career affecting cpu power and time dilation?

 

Edited by Voidhunger
Link to post
Share on other sites
15 hours ago, Sokol1 said:

Exist a standard benchmark track for il-2 Sturmovik: Battle of Stalingrad only?

 

I am trying convince some local hardware channels to include in their benchmarks, one already include MSFS 2020.

 

You guys do "high end" CPU's benchmarks, but I think in more entry level ones, for "new players".

 

You can try this one which is more about testing of graphics performance: 

 

Link to post
Share on other sites
On 11/11/2020 at 10:10 PM, Gambit21 said:

In actual heavy mission builds (not trying to cause problems as I've done here) the TD is not nearly this severe, and might not even be noticed by many, but I still try to bring it back to 1:1 when it's detected.

 

if you compare this testing mission with the Kuban mission intercept enemy ground attack planes above Novorosyisk town or  intercept ground attack planes (with 109D9 or Bf109) on rheinland map. Medium or high density, what is more demanding?

Link to post
Share on other sites
11 hours ago, Mitthrawnuruodo said:

 

I'm not sure what's causing the huge difference between our results (barely noticeable to unplayable). Maybe it's the overclocks and faster RAM. Will do more testing with stock CPU speeds later.

 

Maybe the faster RAM as when recording I am told it records to RAM. I'm still baffled by the fact that this never used to happen. Something must have changed in an update but who knows what.

For those of you that have been experiencing this longer that  have, have the Devs said anything about this? Or are we in the standard 1CGS cricket mode... chirp....chirp....chirp...

 

 

 

Link to post
Share on other sites
15 hours ago, AirWolves=CutCut said:

 

Maybe the faster RAM as when recording I am told it records to RAM. I'm still baffled by the fact that this never used to happen. Something must have changed in an update but who knows what.

For those of you that have been experiencing this longer that  have, have the Devs said anything about this? Or are we in the standard 1CGS cricket mode... chirp....chirp....chirp...

 

 

 

 

Nope, there was long time ago update which reduced cpu workout somewhat. (BOS/BOM release timeframe)

After that im not aware about improvements in that area, but maybe Im wrong.

 

Maybe there is nothing what can be done (unless moving to new game engine) or people are not too active in reporting this issue.

 

or they are not seeing the time dilation

 

 

 

just-takes-the-right-set-of-eyes.jpg

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

 

Nope, there was long time ago update which reduced cpu workout somewhat. (BOS/BOM release timeframe)

After that im not aware about improvements in that area, but maybe Im wrong.

 

Maybe there is nothing what can be done (unless moving to new game engine) or people are not too active in reporting this issue.

 

or they are not seeing the time dilation

 

 

 

just-takes-the-right-set-of-eyes.jpg

It is noticable, in career or whatever mission when you fly toward the objective you can tell when AI spawns or enter in proximity....it's like your plane hit denser medium than air or enters from air into a liquid.

In MP it is more in form of stutters, like radar, once someone gets close to you stutter occur for a moment or longer depends on how crowded area is.

It's always been like that!

Edited by =VARP=Ribbon
Link to post
Share on other sites
59 minutes ago, =VARP=Ribbon said:

It is noticable, in career or whatever mission when you fly toward the objective you can tell when AI spawns or enter in proximity....it's like your plane hit denser medium than air or enters from air into a liquid.

 

 

exactly, there must be something to do about it. Combat flight sims must be fluid or the immersion is really bad.

When the game is fluid in career, its completely new game. Those fights above Novorossiysk or near the Kerch on Kuban map are beautiful. Mainly on sunset, heavy clouds, glittering water and tides...beauty.

 

If the usage of the CPU and GPU are at max level, Im ok with that, our HW is too weak, but if im hitting 30% in CPU and 50% in GPU the game is not using HW potencial to the max. IMHO

Link to post
Share on other sites
4 hours ago, Voidhunger said:

 

exactly, there must be something to do about it. Combat flight sims must be fluid or the immersion is really bad.

When the game is fluid in career, its completely new game. Those fights above Novorossiysk or near the Kerch on Kuban map are beautiful. Mainly on sunset, heavy clouds, glittering water and tides...beauty.

 

If the usage of the CPU and GPU are at max level, Im ok with that, our HW is too weak, but if im hitting 30% in CPU and 50% in GPU the game is not using HW potencial to the max. IMHO

In VR i get 97% load on gpu and 30-40% on 2 out of 16 cpu threads but monitoring with fpsVR there are cpu spikes going orange/red i believe one or two cpu threads goes wild and jumps from core to core.

 

Sadly i kind of doubt there will be improvements in that field!

 

 

 

Link to post
Share on other sites

Well kids I got some more data for ya. Recently I decided that I would turn off "Navigation Markers" in my custom settings for the offline missions and careers I was flying. And low and behold it felt like things were a little better performance wise. So I decided to break out the old Rineland Benchmark Mission and see if I was indeed correct in this assumption. New data is below....
 

With Navigation Markers turned on (Labels were turned off for both scenarios)

1 Min = 1:23

2 Min = 2:47

3 Min = 4:09

4 Min = 5:30

5 Min = 6:56

 

 

With Navigation Markers turned off 
1 Min = 1:20 (-:03)

2 Min =  2:39 (-:08)

3 Min = 3:58 (-:11)

4 Min = 5:16 (-:14)

5 Min = 6:37 (-:24)

 

My conclusion is that the Navigation Markers setting does have an effect on this. And that effect seems to compound over time. Since this setting would really have nothing to do with graphic performance (Again labels were off) I think this leads to the suspicion that this is a CPU and not a GPU issue.

 

 

 

 

 

Link to post
Share on other sites
  • 3 weeks later...

Update on my situation with this time dilation nonsense. I upgraded my system yesterday so now I am running the following...

 

B450 Tomahawk Max Motherboard

Ryzen 5 3600 (6 Cores)
16GB of DDR4 3200 Ram

Nvidia 1660 Super

2TB SSD drive 

 

Here's my new numbers after another fresh install of IL2
 

Not Recording a Track:

1 Min.    1:07
2 Min.    2:14

3 Min.    3:19

4 Min.    4:27

5 min.    5:35
About 1:07 for every minute in actual time

 

While Recording a Track

1 Min.    1:23
2 Min.    2:40

3 Min.    3:55

4 Min.    5:11

5 min.    6:30

About 1:16 for every minute of actual time

(For reference here's my old numbers)
 

Not Recording a Track:

1 Min.    1:26
2 Min.    2:52

3 Min.    4:17

4 Min.    5:39

5 min.    7:06

 

While Recording a Track

1 Min.    1:45
2 Min.    3:22

3 Min.    4:57

4 Min.    6:34

5 min.    8:12

 

While these numbers are an improvement from what I was seeing with my old specs, this is still far from stellar performance in my book. I'm going to spend some time running some scripted campaigns, career mode and PWCG campaigns and see what the perceived impact on playability is. All in all though I would say this doesn't speak highly of the ability of IL2's code to take advantage of CPU power. I play DCS as well and can load the shit out of a mission with ground objects, bombers, fighters etc... and there is no noticeable slow down. In my humble opinion what I just threw at IL2 should have solved any problems that were CPU related. The only thing left is to assume that there's some code issues that aren't allowing the game to fully tap into the resources it's given.    

Link to post
Share on other sites
On 2/20/2021 at 11:01 AM, AirWolves=CutCut said:

Update on my situation with this time dilation nonsense. I upgraded my system yesterday so now I am running the following...

 

B450 Tomahawk Max Motherboard

Ryzen 5 3600 (6 Cores)
16GB of DDR4 3200 Ram

Nvidia 1660 Super

2TB SSD drive 

 

Here's my new numbers after another fresh install of IL2
 

Not Recording a Track:

1 Min.    1:07
2 Min.    2:14

3 Min.    3:19

4 Min.    4:27

5 min.    5:35
About 1:07 for every minute in actual time

 

While Recording a Track

1 Min.    1:23
2 Min.    2:40

3 Min.    3:55

4 Min.    5:11

5 min.    6:30

About 1:16 for every minute of actual time

(For reference here's my old numbers)
 

Not Recording a Track:

1 Min.    1:26
2 Min.    2:52

3 Min.    4:17

4 Min.    5:39

5 min.    7:06

 

While Recording a Track

1 Min.    1:45
2 Min.    3:22

3 Min.    4:57

4 Min.    6:34

5 min.    8:12

 

While these numbers are an improvement from what I was seeing with my old specs, this is still far from stellar performance in my book. I'm going to spend some time running some scripted campaigns, career mode and PWCG campaigns and see what the perceived impact on playability is. All in all though I would say this doesn't speak highly of the ability of IL2's code to take advantage of CPU power. I play DCS as well and can load the shit out of a mission with ground objects, bombers, fighters etc... and there is no noticeable slow down. In my humble opinion what I just threw at IL2 should have solved any problems that were CPU related. The only thing left is to assume that there's some code issues that aren't allowing the game to fully tap into the resources it's given.    

I would test it out in actual play scenarios and see what happens. Gambit mentioned above that this is particularly meant to strain the system and provoke time dilation. Real missions may hopefully have less of an issue.

Link to post
Share on other sites

To clarify things a bit.

 

Think of the game as running at 4 possible speeds. Forget "normal", just think it as having an option to run at 1, 2, 4, and 8.

On most maps, with a single aircraft running some waypoints and executing an attack against a truck you can run 1 through 8 no problem.

Yes 1 is real time, but forget that for a sec because I think it clouds the issue.

 

As soon as you start adding more aircraft or other units, you will see start seeing degradation at speed 8, with speeds 1-4 operating normally.

This begins to happen very early on in the build process...I mean just a few aircraft, map depending.

 

The more units you add, the more that degradation slides downward from 8 towards 1, until not only are you not getting 8X, but no longer 4X, then no longer 2X and then

finally you start delving into 1X territory and that starts to degrade. So TD is happening always, no matter what, in any ‘normal’ mission, you just might not notice if you're running on speed 1 (which of course you are) and it’s only affecting speed 2 and up in that instance. However TD is there just the same.

 

So TD and "time acceleration not working" are the exact same thing - we just call it one or the other depending on what speed we're looking at...but same same.

 

So the Rhineland test mission was designed to cause TD regardless of your system - the point is to give you an IL2-specific measuring stick to compare a new system with an old one, or different CPU etc. It's not designed to give you a target "0 TD" to aim for...you'll never get there with this mission. I really just built it so that when I get my 5800x system built (still waiting on the $%^&ing GPU) I can see how much it improved.

 

As I said it's possible to use one of the 'free' maps and just load it up with complex tanks as these are very CPU heavy.

This would also totally isolate the CPU.

 

 

 

 

 

 

 

Edited by Gambit21
  • Thanks 1
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...