Integrated benchmark

[GunSlingerAUS]

There is a simply massive amount of incorrect information passed through these forums about getting the best performance out of DCS, or observing weird performance issues. People are wasting inordinate amounts of time doing full system reinstalls when it's obvious the problem is a slow PC. However, the lack of a reliable and repeatable benchmark makes it next to impossible for them to realise this.

 

DCS is in dire need of an integrated benchmark, even if it's as simple as a track combined with a "benchmark" graphic pre-config. It really wouldn't be that hard to build, but it would clean up so much confusion and misinformation on this forum. VR is another area that would benefit greatly from the inclusion of a VR benchmark.

[tom_19d]

+1 — If it could also save a small text file with the results and the settings that were employed during the test (or post such data in game so it could be captured in a single screenshot) these results could easily be shared here on the forum — it could eliminate lots of speculation, arguments, and bad information.

[Rudel_chw]

.. DCS is in dire need of an integrated benchmark ...

 

 

I can think a whole lot of more pressing needs for DCS, do we really want ED devoting resources into this?

[GunSlingerAUS]

I can think a whole lot of more pressing needs for DCS, do we really want ED devoting resources into this?

 

Given that DCS performance issues are one of the top subjects on these forums, in both 2D and VR, and how simple it would be to integrate a Fraps-like feature into the game, I absolutely think this should be a huge priority. It would make it much easier to identify real hardware issues.

[freehand]

It's not rocket science to configure your own hardware for DCS.

[GunSlingerAUS]

It's not rocket science to configure your own hardware for DCS.

 

Try telling that to the hundreds of people with issues.

[obious]

+1 for this idea. Also, the notion that this shouldn’t be prioritised because there is other more important work to focus on is ridiculous.

 

I.e. imagine being able to easily measure the performance impact of the new CVN model (animated deck crew etc) with the click of a button and be given tangible results.

[Harker]

I like the idea and I've actually made several benchmarks myself, by simply creating particularly complicated missions, with a lot of units firing at each other, at dusk, with several PiP ports active at the same time (TGP, FLIR on the HUD, Maverick video) etc. Then, I can fly the (short) mission and or replay the track.

 

A simple and inexpensive approach would be for ED to create some missions like that (WW2 birds have different requirements from a modern jet with multiple sensor videos on the MFDs or low-flying helicopters) and place them in a "Benchmark" section in the main menu.

 

I'd leave it up to the user to run a monitoring software and see how their system responds, honestly.

[Shadow KT]

https://forums.eagle.ru/showthread.php?t=194244

 

From 2 years ago :)

 

To be honest... .I've just started using the Hornet "Free Flight" mission for the Persian Gulf as a benchmark

[OG84]

Integrated benchmark would be really helpful to filter out so much noise over in the performance subforums.

output of the benchmark could be a little text snippet with fps, frametime etc, which we can then post in the forum. On the other hand this would make possible performance regressions more obvious when updates come out. Not sure if that is a desired thing as a company.

[Svsmokey]

Fully agree an integrated benchmark is sorely needed .

[Py]

Not only for users, it would help ED a lot to have consistent test results they can compare and trust for debugging and comparison purposes.

 

I'm sure they know this, hopefully they will add it in the future when things are a bit less frantic... :vertag:

[Phil C6]

+1

[sirrah]

No idea how complicated this would be to implement, but it would most certainly be very helpful to both ED and its customers

 

So yeah, +1

[wju]

1+

[SharpeXB]

It’s easy enough to determine your own settings by simply adjusting them yourself. What would be useful is an indicator as to whether you’re GPU or CPU limited. That’s where most all the confusion lies. Otherwise I don’t see the point in comparing results between players. What works for me on my own system is enough for me. 
Also DCS has such a wide range of scenarios it’s hard to figure what this benchmark track would be. Another issue is that it’s possible physics and AI are more demanding in live gameplay vs a recorded track. 
But the big reason this can’t be done right now is that tracks don’t play back consistently so this whole issue cannot be addressed until that’s fixed. 

[GunSlingerAUS]

On 12/6/2020 at 4:48 AM, SharpeXB said:

It’s easy enough to determine your own settings by simply adjusting them yourself. What would be useful is an indicator as to whether you’re GPU or CPU limited. That’s where most all the confusion lies. Otherwise I don’t see the point in comparing results between players. What works for me on my own system is enough for me. 
Also DCS has such a wide range of scenarios it’s hard to figure what this benchmark track would be. Another issue is that it’s possible physics and AI are more demanding in live gameplay vs a recorded track. 
But the big reason this can’t be done right now is that tracks don’t play back consistently so this whole issue cannot be addressed until that’s fixed. 

 

Performance issues fall into the top 5 most commonly reported issues amongst all DCS users, and accurately measuring and comparing performance between different users with different PCs is currently an absolute nightmare, especially in VR. The inclusion of a benchmark feature would be an absolutely huge benefit to the community, and I don't think it'd be too much work to implement, with the caveat that replaying tracks would have to be fixed first (which the community has also been requesting for a long time).

[SharpeXB]

9 hours ago, GunSlingerAUS said:

The inclusion of a benchmark feature would be an absolutely huge benefit to the community, and I don't think it'd be too much work to implement,

Yes if it was possible to do it would clearly be beneficial. My assumption is that the game engine can’t do this due to unpredictable AI and live physics. 
 

What would also be a help is a simple description, (maybe holding your mouse cursor for a pop-up) in the graphic settings telling you whether they are primarily GPU or CPU limited. Understanding this a large part of the confusion or difficulty in getting a system set optimally.

 

Thats the trouble with a benchmark track. In typical games which are mostly GPU limited, adjusting settings is easy. But if DCS just reports a Min Max and Ave with no feedback as to why, then adjusting settings can be a crapshoot. It’s possible, for example, to be CPU bound and therefore be able to actually max a lot of settings with no impact to your FPS
 

Of course there could be two benchmark tracks. A CPU test and a GPU test 

Edited January 9, 2021 by SharpeXB