IL-2 in-game benchmark

[chiliwili69]

Along the past 18 months ti has been useful to have a IL-2 benchmark, specially for the VR but also for 4K monitors. We have learnt what hardware/settings is good for a pleasant experience.

https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline-for-il-2-v3005/

This IL-2 benchmark was created initially by Balapan, but the problem is that sometimes new versions make old flight records incompatible. So we have to create a new track.

It would be helpful to have a betterway to do that, either with a becnhmark section or with a Flight record generated by the Dev team.

Edited September 14, 2018 by chiliwili69

[Alonzo]

Thanks for bringing this up, Chili. This would be a nice feature.

[LLv34_Flanker]

S!

 

 Bring on The Black Death vol.2!

[SharpeXB]

Running a track doesn’t put the same demand on your system as live gameplay. You can test this for yourself by just running a mission live with recorder and Fraps at the same time, then run the replay with Fraps. In the past when I tested this there was a significant difference. I haven’t done it recently. Probably the replay isn’t running active physics and AI. 

[sniperton]

6 hours ago, SharpeXB said:

Running a track doesn’t put the same demand on your system as live gameplay. You can test this for yourself by just running a mission live with recorder and Fraps at the same time, then run the replay with Fraps. In the past when I tested this there was a significant difference. I haven’t done it recently. Probably the replay isn’t running active physics and AI. 

Moreover, FPS doesn't tell everything about performance. Thing is, the AI puts a high demand on the CPU, resulting in occasional stutters that do not always affect FPS. Such stutters and slow-downs are retained in tracks, but a mission recorded on a beefy system would run just fine when replayed on a weaker system, even though that weaker system would struggle to run the very same mission live with active AI. Despite that, a test track would be great to have, but we have to bear in mind that it's only good to measure graphic performance, not overall game performance. A built-in test function would be a different story, it could also benchmark your system according to how many AI objects you can have concurrently for a smooth gameplay. But please correct me if I'm wrong.

[SharpeXB]

I imagine it might be possible to create a live but standard mission that could be used as a benchmark tool in the game. 

[chiliwili69]

6 hours ago, SharpeXB said:

a live but standard mission that could be used as a benchmark tool in the game

 

Yeah! a standard mission played with autopilot leaving the view fixed just straight ahead.

But this standard mission will need to have always all the planes in exactly the same initial position.

[Jade_Monkey]

This is the kind of stuff a smaller studio cannot afford to spend time developping.

[sniperton]

On 10/13/2018 at 8:48 PM, Jade_Monkey said:

This is the kind of stuff a smaller studio cannot afford to spend time developping.

What about a benchmark mission as suggested by SharpeXB? Could also be done by any of our talented mission makers.

[Dutch2]

On 9/26/2018 at 11:32 AM, sniperton said:

Moreover, FPS doesn't tell everything about performance. Thing is, the AI puts a high demand on the CPU, resulting in occasional stutters that do not always affect FPS. Such stutters and slow-downs are retained in tracks, but a mission recorded on a beefy system would run just fine when replayed on a weaker system, even though that weaker system would struggle to run the very same mission live with active AI. Despite that, a test track would be great to have, but we have to bear in mind that it's only good to measure graphic performance, not overall game performance. A built-in test function would be a different story, it could also benchmark your system according to how many AI objects you can have concurrently for a smooth gameplay. But please correct me if I'm wrong.

 Frame times do give you a better indication then FPS, if a wanting to judge if a game runs smooth. 

https://techreport.com/review/21516/inside-the-second-a-new-look-at-game-benchmarking

 

Edited October 19, 2018 by Dutch2

[SharpeXB]

On 10/19/2018 at 8:10 AM, sniperton said:

What about a benchmark mission as suggested by SharpeXB? Could also be done by any of our talented mission makers.

I wonder if that’s not possible because the AI will do something different every time the mission is run. And the AI is what causes some of the load. Live gameplay is more demanding than a recorded track and I figure the AI is part of the reason

Back during EA of BoS 1CGS asked us to post performance numbers in order for them to develop the presets. I’m sure if a benchmark track had been possible, they would have created one back then as it would have been really useful. 

[=RS=Stix_09]

There are third party programs that can do this with logging of data to a file  , like msi afterburner

the game does have a basic avg fps counter , but 3rd party programs , can log record the data (cpu,gpu load etc frame time, bla bla)

 

You tube video howto: Setting Up Game Benchmarking in MSI Afterburner

Edited November 6, 2018 by Stix_09

[PauloHirth]

Benchmark test are useful for select the best videocard driver!