Dedicated Server / Headless client - future developement

[The Man Without Qualities 110]

This set of questions is meant for BIS devs, I am not interesteted in the opinions of the community.

So please do not spam this thread with your opinions how wrong am I with my own view.

 

Situation:

 

ArmA series is sold as "military simulation" supporting "massive battles". Reality is that involving more then 200-300 AI and/or human players results into heavy performance issues.

Massive use of headless clients along with the server solves the problem partly but results into other issues while syncing server with HCs.

 

Conclusion: "military simulation" is true for a lot of aspects of simulation, but "massive battles" is absolutely not true. Having a few normal sized motorized infantry units with supporting units simulated, and that for 2 factions exceeds already the capabilities of A3.

 

I understood from various statements by Spanels that basic design decicions in the server-client task sharing were based on the fact that during OFP developement weak computer platforms and slow internet connectivity were present. Hence, load was pushed to clients, the resulting sync issues were partly compensated with a kind of "prediction" of future positions of moving objects...but in a MilSim with many moving objects that worked not very successful, we all remember "rubberbanding" and units spawning around due to prediction/sync rematching.

 

Requirement:

 

I do not know how much work it would cause for a total redesign but I see the need of a capable server software that collects more or less just the controller input of all clients and all other computing is done on a server HW with many CPUs and a ArmA server SW able to utilize all cores.

I do not see a big benefit installing several HCs at the very same server HW and to let server and HCs communicate through the entire IP stack - even if local.

Plus having the need to keep HCs as clone from addon content and all other related issues.

 

I consider headless client as a workaround to buy time until a capable server SW is developed....but will this ever happen?

 

I would be happy if BIS devs could post here some estimations how much CPU cores + RAM would be necessary to to all that at server which is today done spread to clients but also considering CPU saving due to less sync tasking.

 

And I would like to know if a server centric approach was / is part of ongoing considerations, but skipped due to blocking issues x/y/z.

 

My personal opinion:

 

Currently I am satisfied up to happy with the amount and quality of available in-game-assets (islands/units/simulation of physics), the only big problem left over is the missing massive MP gaming experience due to limited server performance issues.

 

 

 

 

[Dedmen 2590]

This is the Arma 3 forum, from BI who make a military Action game. BIS is a completely different studio making high-profile military simulations which you seem to be talking about.

[The Man Without Qualities 110]

Well, I am talking about the game made by BIS = Bohemia Interactive Studios, not by Bohemia Interactive Simulations.

Could be that I am wrong with abbreviations, but I remembers since OFP the game manufacturer as BIS and the military simulation manufacturer as BISIM.

 

But good point! How is the same issue treated in the professional version?

[FallujahMedic -FM- 866]

Thread has been locked to prevent others from commenting. BI devs still have the ability to post.