DCS learning curve problem.

Come on jackmckay, ease up a little.

ED's been around for quite some time you know, DCS is one of the last standing combat sims of the early ninety's, you know where the rest of them went. Matt Wagner worked on some of the big sim titles in the ninety's and has a good idea what works. It's a slow and steady ride with DCS and to me there is no better or even close in this niche on the market today and we all want ED to succeed. You sound like your very passionate about this too and want ED to succeed and you certainly have ideas to throw around.

If you find faults list them in the bug sections with as much detail as possible for the ED testers to test and recreate, so they can then make "detail reports" for ED's coders etc. This is what the bug sections for, don't just complain about things in every thread with no references or proof, the engineers / coders need to recreate these things to be able to try and fix them, you know this.

ED has many programmers and I'm pretty sure they would do massive amounts of brainstorming with schematics before storming off in some direction that could cost? lots I'm betting.

It's all about balance here jackmckay, including common PC's. It's a fairly small niche and the PC power DCS needs now cuts many out of this niche if using near full graphics, so ED cannot go to far otherwise it could be very detrimental to themselves by making the market even smaller. Some of things you talk about are probably on the drawing board to code in 2035 when the common house hold PC can run it.

You have to stand back a little and look at the big picture and future and the picture looks pretty good from where I stand.

PS: Mirage thing? It plunges at near 2000 km/h, hits dense air, doesn't overheat and its hydraulic performance is still maxed and no airframe stress at all. With full missile pack. How real is that? Gas viscosity and drag force increases on high temperatures. There are more powerful planes with greater Power/Cd ratio that can't come close of 50% of that what that 'thing' can do. Its just an example of anarchy. If it's real let it be but I'm pretty sure its not.

Please show me how you do it, I can't go at 2000km/h at low/medium altitude with 4 missiles :doh:.

@Sobek

LBM method instead of NVS: No one has to make in-game real-time CFD but commit series of tests (AoA, Mach..) separately, extrapolate data and store them on rule (chart) that will be combined with atmosphere properties and kinetic data in-game. That's way efficient than doing actual complex LBM calculations on-air. Anyway, most of (90%) in-game flight model regimes are subsonic (except missiles) and error margin for LBM is under 10% on rough calculations -guaranteed, as LBM has good fluid-structure interaction solver principle. Been there - done that.

@shagrat

There was SETI as distributed job scheduler. It used available cores on idle state over network to commit calculations. There are lot of job scheduling platforms. JSON fast egg. http://dkron.io/ but there are more types available. I do have hexacore i7 utilized under 10% in DCS. Most of players have High-End PCs. To speed up network data bandwidth one should use moderate data compression.

@David

I know that DCS is almost alone in this field and that's why I'm very sensitive to this mess. That is the only sim-game I ever brought actually with 80% purchased modules (I have run Nevada couple of times). That says a lot. I can't snip into a deep code as it's not open source but I can only observe manifestations in game (usually online). Guessing principles and methods of calculations behind my intuition says something is not set right and needs re-coding, specially on physics area because model rendering is secondary in pure sim and can be switched to GPU. Silent competition is already at 2035: https://www.youtube.com/watch?v=VixkiuuFIlA.

@helljumper

I don't have M2K but I've locked an M2k at 50km hot, plunging from 5k to the deck at speed of 1950ish(km/h), fully armed, ready to engage, still very agile.. other locked up high on 15k as they climb so unreal. Viggen is the fastest thing in DCS. Make comparative tests an you will see for yourself. It's not point on M2k but diversity of applied rules of physics.

Check this out guys:

I still believe in DCS and I do want to help. This is just constructive criticism.

@shagrat

There was SETI as distributed job scheduler. It used available cores on idle state over network to commit calculations. There are lot of job scheduling platforms. JSON fast egg. http://dkron.io/ but there are more types available. I do have hexacore i7 utilized under 10% in DCS. Most of players have High-End PCs. To speed up network data bandwidth one should use moderate data compression.

That isn't the problem. The problem is the synchronization of results from the multiple threads in real-time.

Imagine a Machine-Gun firing a 10 bullet burst at a moving object in DCS.

Now we use dkron/SETI/whatever to let 10 distributed cores calculate the ballistics instead of the single core today.

While the target moves, and the cores calculate the trajectory, we now need to check back every few milliseconds with all 10 cores what the bullets positions are, compare them with the object position and register a collision... In fact we would need to wait(!) for the synchronization as asynchronous collision detection is virtually impossible.

Now that applies for mostly everything that interacts with Objects in DCS World. From bullets to weather.

The only parts that aren't time critical, seem to be internal damage effects (won't matter, if a motor block of a tank seizes even a couple seconds later than the hits), general AI route finding, maybe, logistics/warehouse inventory, and I guess dynamic weather changes (not the part that interacts with the airplane, but may be a seeding sync of clouds and general "weather" e.g. pressure changes that can be effecting the world with a few seconds difference on connected Multiplayer machines.

But there is no magic "use more cores to have more CPU power and everything is great" approach available. :dunno:

As I said jackmckay it's all good and I know your passionate, DCS is not perfect and it's fine to point things out to ED's engineer's using good constructive criticism and or discussions, there are coders and engineers that visit these forums.

Many of ED's coders would have engineering and physics background, they must love what they do, I'm sure the team has brainstormed many ideas, just like the one's you have and would like to re code many many things in DCS. But you also need to keep an eye on the big picture and not just think like an engineer when running a flight sim business.

I do admire your passion jackmckay, just remember tho this is a business too and the show must go on to pay for the upgrades / rewrites. Like the real big one from DX9 to DX11 and building and updating EDGE the new graphics engine, it's always on going. Also about those other engines, DCS is a sand box sim, it's needs to have real world conditions, real calculations ballistics, AI etc. How do you think these other engines would fair with all of that other stuff going on?

One thing I do wish ED could do, is kick the AI to a different thread, I do get it tho as to perhaps why, as shagrat pointed out.

That isn't the problem. The problem is the synchronization of results from the multiple threads in real-time.

 

This thread linked below was an interesting read tho, not sure just how good it would work offloading the mission to a second PC.

Tip for increased performance - Use second PC as server!

-

That isn't the problem. The problem is the synchronization of results from the multiple threads in real-time.

Indeed.

Distributing offline tasks like SETI to multiple workers is almost trivial. Distribute your data into batches, give each worker a batch and give the worker as much time as it needs, when it is done, you give it the next batch. Nobody cares about synchronisation in this case.

The frame time of the real time physics engine in DCS is a lot shorter than the typical network delay (one direction, mind you, not even round trip). It is impossible to keep the entire multi-client system deterministic by just streaming the command inputs.

This thread linked below was an interesting read tho, not sure just how good it would work offloading the mission to a second PC.

Tip for increased performance - Use second PC as server!

 

-

That is a way to offload some work to another PC... It is especially useful with large environments (lots of units/AI stuff going on).

On the average singleplayer mission I wouldn't buy a second PC in addition.

If you've got a spare sitting around idle and don't care much about the additional power consumption, though... Why not. :)

That is a way to offload some work to another PC... It is especially useful with large environments (lots of units/AI stuff going on).

 

On the average singleplayer mission I wouldn't buy a second PC in addition.

If you've got a spare sitting around idle and don't care much about the additional power consumption, though... Why not. :)

Yes I will perhaps give this a go at some point, I would like to see just how big a dynamic mission it could handle using moose. :thumbup:

By the way jackmckay, Here's a module for you to look at, if your interested, it's was done by a forum member CptSmiley.

It's an eye opener to what goes into these modules, even if only at the public EFM API level.

F-16 High Fidelity Flight Dynamics and Tech/Academic Initial Demonstration

Forum Post

CptSmiley, I believe now codes the FM's for RAZBAM

Some statistics:

- Lines of code: >5000

- Aerodynamic look-up tables: >24

- Breakpoints (data resolution): tables range from 5 breakpoints up to 1900 for most of them

- Data source - NASA technical papers

How accurate is the engine model?

- Extrapolated equations from the NASA manned sim

How accurate is the flight model (aerodynamics)?

- As accurate as possible with all flight data available (mostly from wind tunnels)

How accurate is the control system?

- As accurate as the NASA manned sim, using the same math model and block diagrams

- Modeling of the mechanical servo-dynamics are in there as well

RagnarDa Updated and Expanded Version (March 10, 2014) GitHub

https://github.com/RagnarDa/F-16Demo