I am reasonably sure that there are occasions when the memory is stretched too far and a CTD occurs.
When this happens must relate to the available memory of the users' PCs, so I would expect it to vary.

Recently Ray reported a CTD when he was intercepting a large flight of B17s with a lot of multiskins loaded.
I was able to get the same, so I have been investigating.
My system was maxed out in terms of graphics quality settings.

B-17s are four-engined bombers so each plane generates four contrails. With up to 16 different multiskins for each plane type, and large flight numbers there will be a large demand on memory which could easily be excessive and cause a CTD.
On my PC I was able to avoid the CTD with the latest exe.
If there is a "MaxSet.mpf" file in the root folder the exe reads it and gets the maximum numbers of friendly primary, friendly secondary, enemy primary and enemy secondary planes for a single mission. If it is not present the default value of 72 is used for each. I tried a "MaxSet.mpf" file with low values 10, 10, 10 and 10, and the CTD did not occur.

Another strategy that I tried was to set the SPECIAL EFFECTS value to "low" using the config graphics screen. This turns off contrails.

I would appreciate any feedback from users so that we might get a clearer picture, and develop further strategies

Jel

Busy with some serious real life issues so I haven't had time to try out the new execs but there's one very minor issues I'd like to bring up.

Recently you added a smoke trail to the ME163 rocket plane but it will only show up with the detail level set to high so if a player sets the detail level to low or medium they're going to miss this nice add-on.

Like I said, minor in the scheme of things but maybe there's something you can do.

The "MaxSet.mpf" possibility seems promising. It is a matter of trying different values.
In the recent tests I have190A, B17-G, 109G-2 a,d P38J all wearing multi-skins, and no CTD with the contrails visible and the 10,10,10,10 setting.

Have a chance this morning to try out "Dec18thOldMap(M.3dz)" with all the MPF files you sent me.



Edit - I set up an intercept with all multiskins, Me110G's and Me109G's vs B-17G's and P-38J's.

Large formations, all graphics set to medium except special effects which was set to high.

With MaxSet.mpf set to 10, 10, 10 ,10 I get a CTD at the same place as before, just as the main model comes into view.

With Max set.mpf set to 5, 5, 5, 5 the game runs fine.

However, the planes are still not showing up at middle and long distance views. ( Yes, I'm using the new exec.)

I see the furthest distance dot up until about 14000 ft, then the planes disappear and reappear at about 4500 feet.

I'll try upping each of the graphics settings one by one and run the same mission to see when I hit the memory wall.




Edit - With all settings on high and the MaxSet.mpf at 5,5,5,5 the game runs. I will try upping the MaxSet numbers.



Edit - With all settings on high and the MaxSet.mpf at 7,7,7,7 the game CTD's after about 1/3 of the main models appear.

I suspect that he key value is the third one which is the number of B17s. Try 5, 5, 7, 5.
Also try with SPECIAL EFFECTS set to low.

I may be worth adding that I have suspected for years that EAW routines are stored in the "Windows Swapfile", and that once the CTD occurs it continues to because the routine in the "Swapfile" runs. I have no definite proof, but in the early days of EAW I had routine in DOS that deleted the swapfile during boot-up when it ran my edited "config.sys". This seemed to avoid repeated problems.

Actually the more people that are able to try this and give feedback the better, with differing operating systems and memory. What works on my PC may not on someone else's.

Back with more data:

I took out all the multiskins and ran the same mission with all the graphics set to high and the MaxSet.mpf set to 10,10,10,10.

No CTD's.

I'm going to say something that might not sit well with some folks but just remember that this is my own personal opinion, not a refection on anyone's work.

Critically speaking after today's tests, I don't think multiskins, as they are presently set up, are viable in EAW. In addition, I think we're chasing a dead end by adding in the MaxSet.mpf.

I say that because it adds another level of complexity that most players are not going to fiddle with. For most casual players the game either runs or it doesn't. To expect them to run 5 to 10 missions with all sorts of different graphic settings, both the ones in the game and now a text file with more settings, to find their own personal sweet spot is asking too much.

I must stress that this is my opinion and not demeaning of any contributor's work. I hope that's coming through loud and clear.

Having said that, I do have an idea that would allow SOME variability in the graphics without adding too much of a memory load.

It is the approach I took with the B-26 multiskins that I just finished.

I used an add-on section, in this case, the B.3dz. This B is composed of only a few elements and the texture is a hi-res BMP with nothing in it but some nose art. (We could probably get away with a lo-res pcx texture because the picture can take up the entire landscape. This might reduce the memory load a bit more.)

This B.3dz is attached to the nose section with a hardpoint and in the game it looks great.

So all the modder would need to do is add one 3dz section to each side of the main fuselage with the squadron ID numbers and optionally, some nose art attached to the nose section. At most there would be three 3dz's and three textures.for each of the 15 secondary planes.

What would be lost is the varied camo skins that the Brits and Germans used but the multiskin modification might run OK if this approach was adopted.

Lastly, I'd be happy to help get functional multiskins working in EAW. I happen to like the variety they add to the game.

I can also report that with all the settings the same as before I CAN run the B-26 multiskin planes after taking out all the secondary files except for the nose art. All the planes are silver, no green and brown ones, the squadron ID's are all the same as the primary plane but the nose art is different on each plane.

I will attempt to add squadron ID's attached with a hardpoint rather than painted on a 3dz section and get back later.

Great. It is good when a problem produces some creative investigations and we find better ways of doing things

I just flew a SPAW intercept of Betties with large numbers of aircraft involved and all settings to the max

All sorts of things happening!

In terms of simplification the multiskin problem has emerged due to large numbers of four engined bombers and their contrails.
Otherwise there seems to be no problem.

It is a PITA to fly them as level bombers at 17000 feet, so they should really only be AI planes to be escorted, without the need for cockpit 3dz files.
Having them with basically the same patterns but different squad markings added at hardpoints could seriously reduce the number of files needed to be loaded.

A big YES to that sentiment. Now to spread it forum wide.

We all want EAW to work at peak performance and the only way to do that is with critical thinking.

BTW, can you check out the problem with the distance models not showing up?

To me it is the skin colours
If I intercept the default B-17s I see them clearly. I do not have a gap.

I have a lot of memory in my system, so I can't offer much help in terms of detecting limitations. However, I do agree with the sentiment that a simplicity that works for the vast majority is the most important. It is more essential that the game is maximally functional for the majority of people. In terms of gameplay, smooth and stable performance combined with consistent long-distance plane-sighting are more important than visual bonuses that matter little in the replication of combat.

Those who have more patience and know-how for increased complexity, such as multiskins, and who have systems powerful enough for these extras, could be offered the option of a Multiskin patch, which could be added either permanently or via the use of a facility such as JSGME. Those without the horsepower to run this could simply choose to ignore it.

That's my opinion, for what it's worth, gentlemen.

Thanks again for your continued efforts and generosity in sharing them.

Yeah but if a player can't run the game or has to jump through hoops to the point that they don't bother playing at all then the multipoint mod is useless


What do you mean by default?






Skyhigh,

The problem is the way EAW handles memory, not how much you have in your system.

As to your suggestion to have different levels of graphics depending on a players system, that is easily done with the file set up program that comes with EAW 1.6

I can make a basic plane folder and multiskin folder of the same plane, that's not too difficult. In fact, with a short learning curve the average player can do the same thing.

However, the problem of how EAW handles memory remains the same.





I am working on a way to have some of the variety that multiskins offer without the memory load. This is a prototype:


Notice the plane in the foreground has the designation RJ A and the one in the background is RJ B. This proves that my idea will work with only two extra files per plane, one for each side, plus one for the nose art if desired.

As stated it doesn't allow for different colored skins, just varied squadron ID's and nose art.

But................it will run with a lot less stress on a PC avoiding the need for the added complexity of an external graphics set up file.

FWIW, I plan to go ahead with this idea with the B-26 in the new plane package I'm almost done with and I'm going to take out the other multiskins just to avoid the CTD issue.

I'll lose a couple of week's worth of work but I think it's important to keep the complexity and chance for CTD's to a minimum.

Ray. I think you misunderstood me
The default ETO planeset has the original EAW B17 skin.
If I load that planeset and fly a B17 intercept in a 190A I see them as black dots which get bigger as I approach from head-on. Eventually they become recognisable planes and never disappear like your multiskin B17s do. I think it is due to the colours used in your multiskins that are making them invisible against the sky background, and not a bug in the exe.

Jel

Nope.

I thought that too.

Same thing happens to the B-26's.

I'll take a few screen shots.

Default B26 (not multiskin) as they approach my 190A









At no time did I not see an image

I've been trying the new exes with graphics maxed out and extra squads=3. I tried an intercept mission of B-17's and escort in a 109k4, with all squadrons maxed out. With all four new exes, I get ctd's very quickly (I can never remember getting a ctd in EAW before). In contrast, using the October exe, the same mission remained stable both of the times I tried it. The game did ctd both times when I exited the mission. I used the ETO Late plane-set. In all cases, framerates remained acceptable and never lower than the mid-twenties, but averaging about forty.

I tried the same mission, with the same conditions, in 1.29 without problems.

I cannot really see any noticeable visual difference with the new exe's. In none of these cases did the bombers disappear and I was able to see them right up to either the ctd or, in the other case, when battle commenced.

What exactly does the MaxSet.mpf file do? Does it limit the number of planes or reduce their graphic imprint? If the former, I think it a pity to limit the number of planes in EAW.

Planes still disappearing over here.

Screen shots tomorrow.

My old Windows XP system flies multi-skins in 1.28e and 1.29 without a problem and at very high frame rates.

http://SimHQ.com/forum/ubbthreads.php/topics/4453508/re-a-few-sceenies#Post4453508

I hope the issues with the new models, multi-skins and 1.6X get ironed out.

Thanks to SkyHigh's comment I tested the October version, which worked well.
Then I went back to my folders and found the last version that worked and it was made on November 17th.
I am now attempting to add the changes since November 17th one at a time to locate the cause of the problem.