Multi CPU and lua export

[AndyHill]

It'd be a great surprise to have proper multi-CPU support in Lock on, but how about at least running the lua exports in a separate thread? I've read about problems running Lock on with dual CPU computers, I hope those will be solved in the next patch or Black Shark (if they still exist that is). A couple of interesting lua projects had problems with CPU power and for example the export stuff required to run my home cockpit has about 10-15% negative impact on performance.

 

Lock on NEEDS all the power it can get and true multi-cpu support would be fantastic. It would be possible to do some pretty neat stuff with the export funcion if there was lots of power available and it might not be that difficult to implement. Hmm how about for example mission scripting through lua? Just dreaming perhaps, but it would probably be nice.

 

________________________________________________________

Lock on MUST have toe-brake functionality with differential braking

[britgliderpilot]

As I understand it, rewriting Lomac to run on different threads (and hence utilise multiple CPUs) would require a COMPLETE rewriting of the code and is hence a LONG way from economically viable.

 

You'd really need to write a new piece of code from scratch to make it work - multi-CPU exploitation isn't something you can bolt on retrospectively.

 

 

At present, the proportion of people with dual-cores is minimal anyway - and I'd venture a guess that the proportion of FC players with both a built-up cockpit requiring LUA exports AND a dual-core chip is absolutely tiny.

 

It'd be nice, no doubt about that. But I seriously, seriously doubt that it's going to happen with anything so small as a patch or addon.

 

 

The toe-brakes, on the other hand . . . . . ;)

[Starlight]

As I understand it, rewriting Lomac to run on different threads (and hence utilise multiple CPUs) would require a COMPLETE rewriting of the code and is hence a LONG way from economically viable.

 

You'd really need to write a new piece of code from scratch to make it work - multi-CPU exploitation isn't something you can bolt on retrospectively.

 

 

At present, the proportion of people with dual-cores is minimal anyway - and I'd venture a guess that the proportion of FC players with both a built-up cockpit requiring LUA exports AND a dual-core chip is absolutely tiny.

 

It'd be nice, no doubt about that. But I seriously, seriously doubt that it's going to happen with anything so small as a patch or addon.

 

 

The toe-brakes, on the other hand . . . . . ;)

 

In principle I agree with you, but IRL I see that many game-devs are releasing patches for dual-core CPUs....

[Guest IguanaKing]

I realize the limitations...but haven't we been saying all along that LOMAC wouldn't run perfectly, maxed out, on most current systems because it had future systems in mind? Why then, is the next incarnation apparently not being made capable of taking advantage of current and up-coming processor technologies? Multi-core rigs are in small numbers now, but it is clearly the wave of the immediate future. Is it true that the next version of Lock On won't even be made to take advantage of current technology which is gaining popularity? If that's true...why should we believe the "its made for future systems" argument any more?

[MonnieRock]

I realize the limitations...but haven't we been saying all along that LOMAC wouldn't run perfectly, maxed out, on most current systems because it had future systems in mind? Why then, is the next incarnation apparently not being made capable of taking advantage of current and up-coming processor technologies? Multi-core rigs are in small numbers now, but it is clearly the wave of the immediate future. Is it true that the next version of Lock On won't even be made to take advantage of current technology which is gaining popularity? If that's true...why should we believe the "its made for future systems" argument any more?

 

 

Thought this might help us IquanaKing

 

Dual Core User Poll

[Guest IguanaKing]

Thanks Monnie :icon_supe

 

I don't actually have one at the moment, but its definitely an inevitable event in the not-too-distant future. I'm thinking that it won't be long before dual-core is all that is available in new rigs, so its silly to design a sim that won't support this feature.

[MonnieRock]

IguanaKing,

 

I could not have said it better. E-tailers and E-builders will push Dual Core to the max.

 

Also, consider that the FX-57 was the last Single Core "Gamers" CPU. The next version FX-60 will be Dual Core with unlocked multipliers.

 

Processor companies have reached the limiting factor with design and we will not see much faster single cores. Their answer, More Cores.

 

 

Dual Core Users Poll

[anivanov]

so its silly to design a sim that won't support this feature.

 

i agree, but whats the price of making this sort of game?

[AndyHill]

Complete multi-CPU support is probably too much to ask although it has been done before (Doom 3 got a HUGE boost from the new patch http://www.firingsquad.com/hardware/quake_4_dual-core_performance/page4.asp). Something like that would be pretty nice for Lock on, but I'm afraid it's a bit unlikely. That's a shame, because without dual-core support Lock on is pretty much at the height of its glory; it will never have more CPU power available since CPU frequencies are unlikely to go up that much in the next few years and Lock on really could use every bit of power it can get.

 

The problem with multi thread coding is the complexity of synchronizing threads and all the communication between them. It would probably be a lot of work to include full support at this point, but the rewards would likely be fantastic (50+% framerate increase in CPU limited situations). Reading between the lines I think ED have already made their decision about that and although I might not like it it's quite understandable. Adding some support for multi cpu environments might be beneficial, however, and the best candidates should be relatively simple features that are somewhat separate from the rest of the program (to reduce communication/synchronization problems), hence the lua stuff. People using the export are pretty rare nowadays, I would think, but the situation might be different if the export was enhanced a little. Bullseye mod and different flight analyzing tools might raise interest and if there was real-time mission scripting available through lua it might even be possible to create pretty interesting scenarios and emulate a strategic AI for example.

 

The bottom line is this: Lock on needs lots and lots of CPU power. The good thing is that there's a lot of that power available, the bad thing is that Lock on won't use it. If, as it seems to be, the dev team isn't willing to go all the way with multi-cpu support I urge them to try to find any cost-effective solution available. A 10-20% increase in FPS in cPU limited situations wouldn't be that bad, I think.

 

________________________________________________________

Lock on MUST have toe-brake functionality with differential braking

[britgliderpilot]

The bottom line is this: Lock on needs lots and lots of CPU power. The good thing is that there's a lot of that power available, the bad thing is that Lock on won't use it. If, as it seems to be, the dev team isn't willing to go all the way with multi-cpu support I urge them to try to find any cost-effective solution available. A 10-20% increase in FPS in cPU limited situations wouldn't be that bad, I think.

 

Agreed.

 

The problem is that it isn't the dev team's decision to make ;)

You've heard that saying about paying the piper, right?

ED has a contract with 1C:Games. They're calling the tune.

 

 

ED is a small team, and I'm reasonably sure that to do this would require the coding chaps to work solely (or at least primarily) on adding multi-core support.

As such you'd have to halt other code developments (or at the very least, slow them down significantly) for something that only would provide an FPS increase for a minority of users.

Publisher decision - "Whoops, no more features or bugfixes for a few months, we're improving FPS for people with dual-core chips only! And because we're using programmer time on this, the cost will be passed to the customer with the addon!"

 

I just don't see how that can be economically viable. Not now, not ever.

 

 

I absolutely welcome the idea, it'd be bloody good . . . . . but I don't think there is any chance at all of it happening with a patch.

[Guest EVIL-SCOTSMAN]

Publisher decision - "Whoops, no more features or bugfixes for a few months, we're improving FPS for people with dual-core chips only! And because we're using programmer time on this, the cost will be passed to the customer with the addon!"

 

I just don't see how that can be economically viable. Not now, not ever.

 

 

Stop walking all over meh dreams BGP :mad: ;)

[AndyHill]

Yes, ED is a small team, but it's always a matter of cost/benefit ratios when deciding on new features. Dual cores may still be a minority, but in the (near) future things will change. Lock on is also VERY hard on the system and I'm pretty certain an average Lock on PC is a LOT more powerful than an average gaming PC. As far as features go I'd say a 50% increase in framerate would be pretty close to the top of the list. Especially since most imaginable features will also eat their share of the performance pie and you just can't eat anymore when the pie's already long gone.

 

Unfortunately as far as Lock on goes CPU developement practically ceased about a year ago. Things like flight model and aI enhancements will probably require a lot of CPU power that just isn't there if you can't utilize dual cores. With the Black Shark addon ED are putting the player very close to the ground war, which will give ED a real challenge, since (IMRO) ground war isn't Lock on's strong point (for example the conversation around SAMs shooting down AGMs has revealed quite a lot of room for improvement in AI modeling).

 

I don't want to sound pessimistic, but I can't help it. Lock on is nearing a dead end as far as developement goes if technological advancements can't make it run better anymore. Flight sim business is a bit more personal to me than any other gaming I do and I'm a bit worried about what will keep the ED team fed while they develop the next project after Black Shark. Sturmovik is about a year older than Lock on and thus far I've paid for three full-priced games and one add-on for the Il-2 series, Lock on has cost me one full-priced game and an add-on. There's a slight chance that the Sturmovik sequel will appear this year and there will be another add-on release for Il-2 before that. Since the Lock on sequel will probably require a complete rewrite of the engine (I can't see any new sim being published without multi-CPU support for example) it will probably be a huge effort and take lots of time. It might help if they could keep releasing add-on stuff to the previous sim while developing the new one like Oleg's team does and I think that would be easier if ED could stretch the limits of available CPU power a bit.

 

It just came to my mind that ED have done fantastic job with the export functions. There's still room for improvement, but considering how few users there are it's an amazing effort and already better than any other sim I've flown.

 

________________________________________________________

Lock on MUST have toe-brake functionality with differential braking

[4c Hajduk Veljko]

There is no excuse on not getting a support for dual core in Flaming Cliffs. Even if it would take “completely” rewriting the code. Black Shark should not be released without dual core support.

 

Flaming Cliffs ONLY competitor on the market today is Falcon 4.0. And Falcon 4.0 has a support for multiple porcessors/cores. And Falcon 4.0 works 30% faster on dual processors/cores. And Falcon 4.0 was coded almost 10 years ago.

 

Any future projects coming out of Eagle Dynamics MUST include dual core support or Eagle Dynamics will disappear from the market. Eagle Dynamics should be sending its programmers to seminars about dual core. Dual/multiple core is the future of the CPU’a and GPU’s. Software company that is not able to adapt to this new challenge, will disappear from the market scene.