Game Engines!

[Pyroflash]

Well, think of the competition for making a combat helicopter sim.

[asparagin]

They were supposed to be there with Crysis 2, and six months before the game released, those effects were in engine and working, but got pulled by EA, just like DX11 support.

 

Ahh EA again huh?

[ZQuickSilverZ]

I can kind of see the point about enviroment. I understand today's computers have limitations. But when you play flight sims how often do you see waterfalls, or bluffs, or waves. I would like to fly over water for instance and see the water crashing into a cliff instead of just butting up to the shore. I would like to have waves for instance because whenever we fly over water we just assume the water is calm. I think it would be cool to fly low and have a wave spash water up on the cockpit glass or maybee even have a wave hit a helo and have to adjust from the impact to stay on course.

 

Lets face it the aircraft are "there". ED proved that. The next step is to work on the enviroment.

[GGTharos]

Then 'environment' isn't going to be there until computing power increases some 10000-fold. Then maybe they can start doing something. Every time you double your view distance, you need at minimum 4x the computing power. Think of it as difference between low, medium and high scenes and tree vis radius as well.

 

You have /no/ idea what you're talking about. No idea whatsoever.

[Pyroflash]

It looks so beautiful...

 

I wish ED could jack the sim from them and release it to us. I would gladly pay for a rig that could run it and the cockpit to go with it.

[Conure]

I'm not programmer, but I think I an interject with some naive common sense. It seems to me that at the moment, whichever game I play, all four of my CPU cores are never maxed out. Further, it doesn't seem implausible to have 8x cores on a CPU, we have 6 already. Maybe we do already have 8.

 

Is it not then safe to assume that with a 6core CPU @ 4ghz + per core, or an 8 core CPU with the same spec, coupled with sufficiently programmed software that makes absolute maximum use of every single core, that some of these wishes could become fact? Let's be honest, a decent quadcore at 4ghz maxes this game with ease...Double that and can't the other 2 or 4 cores not handle all the graphics and environment physics effects going?

 

It seems that it's less a limitation of technology (as our technology can be stacked, for example SLI, multicore CPUS), and more a limitation of software programming to make full use of a computers available potential.

 

With sufficient programming capabilities, would it not be possible to divide a game as follows:

 

Core 1 - Flight dynamics and avionics

Core 2 - Ballistics

Core 3 - Friendly AI

Core 4 - Enemy AI

Core 5 - Environment processing

Core 6 - something else that needs doing...Repeat until core limit reached...

 

Apply the same process to the new multi GPU cards, one for water, one for terrain etc etc...

 

Do we really need a quantum computing for this?! From a non educated point of view, I can't understand why it isn't possible with current technology to have games that look like Crysis, are as realistic as DCS A10 and have ground level AI on a level of Shogun 2: Total War.

 

Considering all of these games currently run fine on a dual core processor...Perhaps I'm wrong, and if so...Why?

[Conure]

Oh, and obviously I understand that from a marketing point of view a Sim that only ran on those specs wouldn't be plausible, but isn't the technology there? It would be nice to see these sort of things in the near future...Games like IL2:COD, although apparently a train wreck, really show the way graphics are going for us Simmers. A10 can be absolutely beautiful at times, if not for that terrible terrain they've been using since before the Wright Brothers first took to the sky......

[asparagin]

It seems to me that at the moment, whichever game I play, all four of my CPU cores are never maxed out. Further, it doesn't seem implausible to have 8x cores on a CPU, we have 6 already. Maybe we do already have 8.

...

 

I feel your pain.

If the technology trend is multiple cores, programmers should adapt quick.

For the near future at least it looks like advances will be based on multiple cores.

[Pilotasso]

Weapons:

 

 

Would love to see how that compares to LOMAC, if it has solid multiplayer Im sold. :)

[Conure]

What a beautiful place to fly, even with Mirag 2000:

 

I can render it easily, though!

 

BTW, if DCS doesn't like realism and much details about environment, and wants to focus just on aircraft, Vertical Reality Simulations has something to share.

 

Some stuff for F/A-18E lovers:

Start up:

http://www.youtube.com/watch?v=STfD6xGD6x0

 

Weapons:

 

Regarding that top link...Wtf? That looks photo realistic....Surely no mere mortal PC can run that?

[some1]

If the technology trend is multiple cores, programmers should adapt quick.

For the near future at least it looks like advances will be based on multiple cores.

 

Programmers can adapt, but many algorithms can't. :)

[Conure]

Programmers can adapt, but many algorithms can't. :)

 

Can that be explained in layman's terms?

[SFJackBauer]

Can that be explained in layman's terms?

 

The computer is a giant minuscule calculator. Take a huge mathematical equation, and try to split it between 5 people so each people solves 1/5th of the equation, and then order them all to work simultaneously and arrive at a correct result.

 

Impossible? No, you must break the equation in parts that are independent from each other first, calculate them, and then do a final step to unite all smaller pieces to achieve a final, single result. Problem is you reach a wall where you can't split anymore parts of the equation in smaller independent pieces, therefore they must be calculated sequentially.

 

Its an analogy, but believe me it is very close to what multi-threading programming is.

[Conure]

The computer is a giant minuscule calculator. Take a huge mathematical equation, and try to split it between 5 people so each people solves 1/5th of the equation, and then order them all to work simultaneously and arrive at a correct result.

 

Impossible? No, you must break the equation in parts that are independent from each other first, calculate them, and then do a final step to unite all smaller pieces to achieve a final, single result. Problem is you reach a wall where you can't split anymore parts of the equation in smaller independent pieces, therefore they must be calculated sequentially.

 

Its an analogy, but believe me it is very close to what multi-threading programming is.

 

Thanks for explaining that to me. So, we've kind've hit a point where the MHZ(ghz) aren't raising too quickly, and a point where the algorithms are limiting what we can do...So has Moores law ceased to function? What is the future for processors?

 

I have a friend that studies Computer Science here in the UK that told me graphics cards are now being used to process "proteins and DNA" or some such thing..Apparently they're more efficient. Unfortunately I don't know how so!

[HiJack]

Would love to see how that compares to LOMAC, if it has solid multiplayer Im sold. :)

Yes it's nice but it there are no mission designer for FSX so you can create a mission with SAM and other ground units up front (not that I know of). The TacPack makes it possible to "dropp" in aircrafts while playing, like a wingman.

[ZQuickSilverZ]

HiJack what I read about Tacpak says it has ground targets. Though you are right as far as I know about there not being a mission editor. I think it is much more tactical to be able to place them where you need them (that is a little more DYNAMIC, as in dynamic campaign). It kind of makes it a flight sim / real time stratagy game. You should have just one person on each side that can place units, and they should decide before the game even starts how many they can place and when (to avoid spamming).

 

http://vrsimulations.com/tacpack.htm

 

" Not only will you be able to take out AI aircraft in single-player, you'll be able to dogfight your friends (and enemies) multiplayer, take out SAM and AAA, or bomb a hostile airfield - all in free-flight"

 

Also listed on the feature list are these

 

"A/A and A/G missiles, ballistics and guided/unguided bombs. "

 

"Drop AI into sessions (see AI menu below). Assets such as SAMs, working carriers, tankers and more may be dropped by team members into the battlefield."

 

"Ground Moving Target Indication (GMT)"

 

"Ground Moving Target Track (GMTT) "

 

"Fixed Target Track (FTT)"

 

These tell me not only will we have ground targets to shoot at........ we will have MOVING ground targets to shoot at.

 

We will also be able to do aerial refueling.

 

"When the dust settles you'll be able to call for a tanker, fuel up behind an intelligent AI refueler flying dynamic racetrack patterns, then RTB to review the action (and lick your wounds) via TacView - an amazing ACMI playback system."

 

Originally Posted by Pilotasso

"Would love to see how that compares to LOMAC, if it has solid multiplayer Im sold."

 

So far its just one aircraft. And lets not forget FSX has issues with framerates for most people wich might make combat very difficult (not to mention you just added alot more calculations). I would say wait until they come out with one more aircraft just to see how they interact with one another in multiplayer.

Edited June 11, 2011 by ZQuickSilverZ

[GGTharos]

Not in any practical way. Guess why.

[Moa]

Regarding the computation limit:

* It takes one woman 9.5 months to have a baby.

* You cannot use 9.5 women to have a baby in one month.

 

For some types of problems you can do stuff in parallel (graphics) but other types of problems (eg. carrying a baby) cannot be solved that way (eg. iterative computations that have steps that rely on other steps to be complete first, or network traffic to arrive).

 

Then there is the problem where even if you can parallelize an algorithm/computation you can't reduce the pieces too small and too spread because the overhead of communicating and coordinating between threads, cores or machines dominates the time to compute.

 

IIRC the Australian Air Force built an F-18 flight simulator that used 33 PCs working together (presumably dual-CPU each with multi-cores). For one simulated F-18. IMHO (as a developer) it is amazing what Eagle Dynamics has done with A-10C - and it runs on PCs with a lowly 4 GB RAM and on my venerable Q6600.

 

Sure that F-35 looks amazing. But you're forgetting the multi-million Defence Mapping Agency budget that did the terrain data (both acquiring it and stitching it together so it doesn't have defects). Then there is the team that wrote the software, and the team that built the simulator hardware and electronics interface. Total project cost is millions, as is the cost of a full, movable cockpit with hydraulic throw.

 

Then we get folks on here whinging at the $60 cost of the product, and then there are all those who pirate it. Flight simming is a niche market as it is, and ED do very well with what must be relatively limited budgets compared to 'mainstream' games - and that's before the 'pirates' steal copies of the game. Fortunately ED gets money through military contracts or we would have nothing.

 

Sure it is a nice dream to have Crysis level graphics in a sim, but realistically:

* no-one has a PC that can run it, and is not likely to in the future (quantum computers are good for simulating *gasp* quantum systems but are relatively crap at general purpose computing

 

* ED doesn't have the budget or time for the consumer-adaptations of the military versions to create something like Crysis (preferring instead to get aerodynamics and systems right).

 

* Even if ED made such a simulator, no-one could afford to pay what was required to cover the development costs.

 

In short, you're dreaming if you think it is possible to make DCS have a 60 km view radius and render every leaf as in Crysis while having the great systems, battle and weapons modelling of current DCS.

Edited June 12, 2011 by Moa

[combatace]

I'm not programmer, but I think I an interject with some naive common sense. It seems to me that at the moment, whichever game I play, all four of my CPU cores are never maxed out. Further, it doesn't seem implausible to have 8x cores on a CPU, we have 6 already. Maybe we do already have 8.

 

Is it not then safe to assume that with a 6core CPU @ 4ghz + per core, or an 8 core CPU with the same spec, coupled with sufficiently programmed software that makes absolute maximum use of every single core, that some of these wishes could become fact? Let's be honest, a decent quadcore at 4ghz maxes this game with ease...Double that and can't the other 2 or 4 cores not handle all the graphics and environment physics effects going?

 

It seems that it's less a limitation of technology (as our technology can be stacked, for example SLI, multicore CPUS), and more a limitation of software programming to make full use of a computers available potential.

 

With sufficient programming capabilities, would it not be possible to divide a game as follows:

 

Core 1 - Flight dynamics and avionics

Core 2 - Ballistics

Core 3 - Friendly AI

Core 4 - Enemy AI

Core 5 - Environment processing

Core 6 - something else that needs doing...Repeat until core limit reached...

 

Apply the same process to the new multi GPU cards, one for water, one for terrain etc etc...

 

Do we really need a quantum computing for this?! From a non educated point of view, I can't understand why it isn't possible with current technology to have games that look like Crysis, are as realistic as DCS A10 and have ground level AI on a level of Shogun 2: Total War.

 

Considering all of these games currently run fine on a dual core processor...Perhaps I'm wrong, and if so...Why?

 

You don't require higher cpu power you just need clustering and even dual core can do the job..

[GGTharos]

Outerra is not useable for DCS, nor is MSFS' engine.

[SFJackBauer]

Thanks for explaining that to me. So, we've kind've hit a point where the MHZ(ghz) aren't raising too quickly, and a point where the algorithms are limiting what we can do...So has Moores law ceased to function? What is the future for processors?

 

Moore's law is not a law per se, more a trend prediction that Gordon Moore did half century ago and proved accurate for the last 50 years. He himself said that it could not continue indefinitely.

 

The processors future lies in innovative manufacturing process that can put more stuff in smaller sizes without electrons crossing the barriers. However a physical barrier will be hit eventually, and hopefully at that moment a quantum computer would have been researched already.

 

I have a friend that studies Computer Science here in the UK that told me graphics cards are now being used to process "proteins and DNA" or some such thing..Apparently they're more efficient. Unfortunately I don't know how so!

 

He is 100% right. Graphics card processors are constructed specifically to work efficiently mathematical operations on large arrays of simple values - like RGB values of pixels, or XYZ coordinates of triangle vertices. Since some time ago graphics cards were made "programmable" - before their functions were completely hardcoded - to give flexibility to graphical programmers. Hence the "shaders" - programs that run on each pixel or each geometry vertex, written in a human language, and compiled and executed inside the video card, bypassing almost completely the CPU.

 

So any kind of data that falls into that same category - like long sequences of DNA which is made of basic types of proteins - can be processed more efficiently by a VGA chip than a CPU chip. The "shaders" in this case doesn't produce graphical results - only calculation results that are then used in practical ways (or not-so-practical ways, like Seti@home :)).

[Bushmanni]

The fundamental problem with wanting to have a sim with fps graphics quality is that the power available to both of the types of engines is the same but the draw distance requirements are different. The fps engine is always going to have better looking graphics due to shorter draw distance and hence more power available to eye candy. When sim engines reach the eye candy level of current fps-engines, the fps-engines have also evolved and we are still in the same situation of fps games looking much prettier than sims and non-experts whining why can't my favorite sim look like my favorite fps.

[Conure]

Thank you for all the responses guys, very interesting reading and I am now understanding the limitations in a more practical way :) +rep to all!

 

Thanks

[sobek]

I am now understanding the limitations in a more practical way

 

Props to you, that hasn't happened too often. :):thumbup:

[HiJack]

IMO ED have the best engine compared with all other flight sim's. Even FC2 is bether than all the others IMO.

 

(HJ)