609_Relentov Posted April 7, 2015 Posted April 7, 2015 If a mission is created with a lot of late activation objects vs objects that are activated at mission start, is the memory already allocated for all of the late activation objects when the mission is loaded, or is the memory dynamically allocated as they activate? I ask only because I'm trying to use a template mission that has - for example - AAA/SAM units at many airfields, but some are late activation (depends on certain flags, etc.) and may never get activated. If adding all of these late activation objects that may never be used during a given mission is still forcing the memory hit, then I may take them out and abandon the template mission idea (for multiple missions in a campaign, for example). Thanks...
DarkFire Posted April 7, 2015 Posted April 7, 2015 Interesting question. I'll see if I can do some testing over the next few days. I'd also like to know the answer to this. My first guess is that the additional memory necessary for the extra objects would be minimal compared to the memory used by the sim as a whole, but I'm going to test this. More to follow... System Spec: Cooler Master Cosmos C700P Black Edition case. | AMD 5950X CPU | MSI RTX-3090 GPU | 32GB HyperX Predator PC4000 RAM | | TM Warthog stick & throttle | TrackIR 5 | Samsung 980 Pro NVMe 4 SSD 1TB (boot) | Samsung 870 QVO SSD 4TB (games) | Windows 10 Pro 64-bit. Personal wish list: DCS: Su-27SM & DCS: Avro Vulcan.
609_Relentov Posted April 7, 2015 Author Posted April 7, 2015 That was quick, thanks DarkFire. I tried to test by examining the task manager while loading missions with varying numbers of late activation objects, but it was inconclusive - probably need a better way of measuring. The obvious (to me :) ) benefit is to create a fairly beefy template mission with all of the airfields set up with everything (SAMs, AAA, maintenance vehicles, etc.) except maybe the Client-flyable aircraft and statics (I don't think you can late-activate a static) - then just use a series of flags to activate the preferred airfield objects at mission start - a big time saver! In any case, I look forward to your tests... Thanks, Relent
Wrecking Crew Posted April 7, 2015 Posted April 7, 2015 My understanding is that Late Activated groups are already in the mission and on the battlefield. But, if you use a script to create/clone a group mid-mission then that is a new element. WC Visit the Hollo Pointe DCS World server -- an open server with a variety of COOP & H2H missions including Combined Arms. All released missions are available for free download, modification and public hosting, from my Wrecking Crew Projects site.
609_Relentov Posted April 7, 2015 Author Posted April 7, 2015 Thanks WC - so the hit is there, but only at mission load time. As DF stated above, it still is likely only a small amount of memory overhead relative to the amount of RAM used just to load DCS.
DarkFire Posted April 8, 2015 Posted April 8, 2015 It'll probably be Friday by the time I get to run the test due to weird shifts at work, but I've figured out how to get the Windows performance management tool to record so we'll see what happens... For interested parties, hit the start button, type "perfmon" without the quotes, hit return and when it loads have a play. System Spec: Cooler Master Cosmos C700P Black Edition case. | AMD 5950X CPU | MSI RTX-3090 GPU | 32GB HyperX Predator PC4000 RAM | | TM Warthog stick & throttle | TrackIR 5 | Samsung 980 Pro NVMe 4 SSD 1TB (boot) | Samsung 870 QVO SSD 4TB (games) | Windows 10 Pro 64-bit. Personal wish list: DCS: Su-27SM & DCS: Avro Vulcan.
SNAFU Posted April 8, 2015 Posted April 8, 2015 In my experience the late activated groups have the same impact on performance as the other groups, but only from my feeling, not from dedicated testing. You might test this by adding really a lot of units in the mission and check the FPS. For 1000 units (simply copy and paste them) you should easily compare the results. There is still the option to set the group AI to off via triggers at the same time. Didn´t have much experience with this... With AI off the units are static but visible, if you combine this with advanced triggers and make them visible, you have the similar results as late activated groups, but maybe not the performance hit. [sIGPIC][/sIGPIC] Unsere Facebook-Seite
609_Relentov Posted April 9, 2015 Author Posted April 9, 2015 Thanks DF, I look forward to your test results. Interesting SNAFU, thanks.
DarkFire Posted April 16, 2015 Posted April 16, 2015 Just a quick update: having played with the performance monitor whilst using DCS... I'm getting some very strange results that are inconsistent across runs. More to follow in a few days. System Spec: Cooler Master Cosmos C700P Black Edition case. | AMD 5950X CPU | MSI RTX-3090 GPU | 32GB HyperX Predator PC4000 RAM | | TM Warthog stick & throttle | TrackIR 5 | Samsung 980 Pro NVMe 4 SSD 1TB (boot) | Samsung 870 QVO SSD 4TB (games) | Windows 10 Pro 64-bit. Personal wish list: DCS: Su-27SM & DCS: Avro Vulcan.
EasyEB Posted April 21, 2015 Posted April 21, 2015 I have made a mission template like the ones you are talking about where I have about 900 enemy units (AA, armor convoys, truck convoys) filling out the world and giving it a feeling of beeing populated. Each group has a 33% chance of beeing activated at mission start (so sometimes there's lots and sometime there's few). The mission load time and the time after the in-cockpit briefing to actual mission start (when the mission is rolling the random and activating groups) is really long and the same goes for the after action report and such, but once the mission has actually started I don't notice any real difference. I could be wrong though, just a feeling I get.
Recommended Posts