Sundder

Hi Sheson, o.k., I've tried a number of combinations where I seem to be just short of not crashing and while using cow tamriel 0 -64 at the main menu doesn't result in a CTD, it does result in an ILS. Sort of the way Alternate Start did (though Alternate start produced a page flip sound and no load screen model).

Hi Cam, the problem I (and a few others) experience is being unable to start a new game at all or load any save when we have too many mods in our load order. You're problem sounds more like you're running out of memory. You didn't mention that you use Crash Fixes (with it's OSAllocators enabled); if you don't, you should know that it can really help with this.

Hi Sheson, I've been testing the overload condition with both a Temporary=0 and Temporary=1 setting for the output and have some results as compared with v2.14, which was the version I previously had installed (not counting 2.20). First, whatever you did since 2.14, it shows a definite and consistent improvement in the number of mods that can be added to the load order without triggering the CTD dunc and I have been experiencing (I realize I'm stepping out on a limb here since dunc hasn't actually reported back any results yet as far as v.2.21). Some mods or mod combinations that systematically CTDed Skyrim with version 2.14 and earlier now no longer do so. For example, adding just WeissadlersHamlet-3-0 to my standard load order was a sure-fire CTD with 2.14, but this now works with 2.21. Same thing as far as adding Here There Be Monsters V6 plus Here There Be Monsters - The Call Of Cthulhu V6. I can also add 2 or 3 mid-sized mods more. However...that is all that can be added before the same CTDs strike again, and in all cases countref is way, way short of 65 535 persistent references. Also, my results are the same whether I use the DynDOLOD output created with Temporary=0 or Temporary=1. Some fairly large worldspace mods that didn't load before (like Rochester https://www.nexusmods.com/skyrim/mods/52556/?) now do. Others, like Aether Suite (https://mod.dysintropi.me/aether-suite-3-4-0-operation-verthandi/) still only load when DynDOLOD (or another equally large mod) is disabled. And, BTW, Aether Suite has no worldspace, just cells with persistent and temporary references. I'm not sure what to make of all this.

I can confirm as well, I got around 8 minutes using 2.20 instead of the previous 45 min. Given that I'm comparing outputs (Temporary=0/1) this is definitely a major plus! Thanks! I'll be updating to 2.21 today and testing again.

Nope, nothing relating to DynDOLOD appears in the top left when the output file created using the Temporary=1 setting is installed. I always get my setup's standard messages related to Death Alternative and Hypothermia Plus, but I only get the successful DynDOLOD activation message when I install the output file created using the Temporary=0 setting. In the former case the "Last Message" in the Main DynDOLOD MCM window is "DynDOLOD deactivated". Main screenshot: https://i.imgur.com/53ZtqfT.png Information screenshot: https://i.imgur.com/7jk5xHL.png

Hi Sheson, o.k. I've finally got a grasp on the 2.20 initiallization problem I'm having. Bottom line: DynDOLOD 2.20 only absolutely refuses to initialize when I use the output created using the Temporary=1 setting. DynDOLOD has no problems whatsoever initializing as soon as I move to an exterior space (in this case Solsthiem) as long as I'm using the output file created using the Temporary=0 setting. I hadn't noticed earlier becuase I was just dropping the ESP files into the Data folder to get a persistant reference count using countref. When I actually install either of the output complete files, only the one created using Temporary=0 lets DynDOLOD initialize. Still, given that even that one has 9000 fewer persistent references that the previous versions did, I think I can get around to trying to "overload" the engine tomorrow and confirm whether this has actually made a difference.

I know this is going to come as a shock but I use NMM, so I ran the x64 versions directly through Windows. O.k., I'll try rolling back to my 2.14 output file and save with the 2.19 resources then upgrading again. Maybe I just didn't wait long enough when performing the clean save procedure.

Hi Sheson, I've been testing 2.20 with Temporary=0 and Temporary=1 and this is showing some great results. BTW, even with Temporary=0 I'm getting only about 56 000 persistant references for Tamriel with my standard load order, which is already about 9000 less than before. Was this expected? With Temporary=1 I get only about 33 500 persistant references for Tamriel which is a HUGE decrease. The ESP is only 1MB larger (7->8) and there aren't any additional masters that I noted in my case, so all in all a very easy trade-off. I haven't gotten around to testing whether I can overload the engine yet however, as I'm dealing with one other problem: 2.20 won't activate even though the MCM is available and seems to work o.k.. I thought I performed the clean save procedure correctly but I may have messed up while swapping various output files between 2.14 and 2.20. Right now I'm looking at rolling back to 2.14 with the save I have from then to redo to upgrade, but ​I no longer have the (core) resource files to match 2.14 (definitely my bad). I don't suppose you could drop them on mega so I can reinstall them? In the meantime I'll see if I can't recover the deleted file. Also, I had to generate textures and LOD with the 32bit versions of TexGen and DynDOLOD as the 64bit versions kept throwing a bunch of access violations.

Hi Sheson, I've upgraded to 2.20 and would like to do a run with both Temporary=0 and Temporary=1 but can't find a DynDOLOD.ini. I've found a bunch of ini's in DynDOLOD 2-20\Edit Scripts\DynDOLOD and there's c:\users\[username]\AppData\Local\Skyrim\DynDOLOD_settings.ini but no DynDOLOD.ini as such. Even a full search of the PC doesn't turn it up. Do I need to create the file then add the value or is it generated some other way?

Sure thing; here you go:

O.k., I'm getting some odd results. I've been doing further testing with DynDOLOD 2.14 installed. (I should be able to test with 2.20 within the week with and without the temp refs toggled on.) I've been swapping mods in and out and checking whether I can load or if I CDT and have found some combinations that CDT me starting at about 65 300 perm refs for Tamriel based on the latest countref.pas. Finally, I added: Jurgholm Den Erovrade Staden Dreadmyst Hollow (Cleaned) Here There Be Monsters to my load order and passed (loaded) with 65451: "Total for each parent world including all child worlds Tamriel 65451" Then I added a small house mod that has its own small worldspace, Pale Marsh Manor Lite and still managed to load. Thing is, running countref returned this: "Total for each parent world including all child worlds Tamriel 65640" Yikes! But when I checked the list of refs, Pale Marsh Manor Lite added most of these as flagged: 2 PaleMarshLite.esp for Tamriel in Tamriel 0 PaleMarshLite.esp for Tamriel in SouthfringeWorld 0 PaleMarshLite.esp for Tamriel in BluePalaceWingWorld 0 PaleMarshLite.esp for Tamriel in DLC1AncestorsGladeWorld 187 PaleMarshLite.esp for Tamriel in 00PaleMarsh (f) Still, the closer I get Tamriel perm refs to 65535 the more I fail to load. It's almost impossible passed 65350 and took some creative selection from my mod stock.

Well, now we're really getting somewhere. Here are my results... Non-CTD Load Order: Total for each parent world including child worlds with Use Sky Cell flag Tamriel 53410 CWTestHold 17 DLC2SolstheimWorld 3709 WyrmstoothWorld 3210 manny_GF_O 717 manny_GF_Desert 2133 MBITreacheryCircle 726 MBIFraudCircle 10 00MBI_GluttonyWorld 973 WhiterunWorld 84 MBI_Darren_Anger_worldspace1 577 0_MBI_Dante_ARENA1 1442 MBI_Limbo 811 BDBStabbyHideoutWorld 45 WZONESwarehouse 43 WZONESwarehouseWZA 44 ERDPGiantCave01World 44 ERDPGiantCave02World 24 WindPass_World 26 Vuldur 9 arnimaDUPLICATE003 5254 ArnimaVoid 1638 XJKislandWorld 1534 ACX_TheGatewayRuined_Worldspace 583 ACX_TheGatewayExt_Worldspace 593 ============================== Total for each parent world including all child worlds Tamriel 65228 <–Is this looking suspicious yet? CWTestHold 17 DLC2SolstheimWorld 4897 WyrmstoothWorld 4412 manny_GF_O 717 manny_GF_Desert 2613 MBITreacheryCircle 726 MBIFraudCircle 10 00MBI_GluttonyWorld 973 WhiterunWorld 84 MBI_Darren_Anger_worldspace1 577 0_MBI_Dante_ARENA1 1442 MBI_Limbo 811 BDBStabbyHideoutWorld 45 WZONESwarehouse 43 WZONESwarehouseWZA 44 ERDPGiantCave01World 44 ERDPGiantCave02World 24 WindPass_World 26 Vuldur 9 tos_SunkenTowerWorld 6 arnimaDUPLICATE003 6303 ArnimaVoid 1638 XJKislandWorld 1534 ACX_TheGatewayRuined_Worldspace 583 ACX_TheGatewayExt_Worldspace 593 ==============================93399 Grand Total CTD Load order (Same as above but with Death Mountain 2 added) Total for each parent world including child worlds with Use Sky Cell flag Tamriel 53449CWTestHold 17 DLC2SolstheimWorld 3709 WyrmstoothWorld 3210 manny_GF_O 717 manny_GF_Desert 2133 MBITreacheryCircle 726 MBIFraudCircle 10 00MBI_GluttonyWorld 973 WhiterunWorld 84 MBI_Darren_Anger_worldspace1 577 0_MBI_Dante_ARENA1 1442 MBI_Limbo 811 BDBStabbyHideoutWorld 45 WZONESwarehouse 43 WZONESwarehouseWZA 44 ERDPGiantCave01World 44 ERDPGiantCave02World 24 WindPass_World 26 Vuldur 9 arnimaDUPLICATE003 5254 ArnimaVoid 1638 XJKislandWorld 1534 ACX_TheGatewayRuined_Worldspace 583 ACX_TheGatewayExt_Worldspace 593 ==============================Total for each parent world including all child worlds Tamriel 65898 <– Yup, this crashes my Skyrim to desktop on save load or new game CWTestHold 17DLC2SolstheimWorld 4897 WyrmstoothWorld 4412 manny_GF_O 717 manny_GF_Desert 2613 MBITreacheryCircle 726 MBIFraudCircle 10 00MBI_GluttonyWorld 973 WhiterunWorld 84 MBI_Darren_Anger_worldspace1 577 0_MBI_Dante_ARENA1 1442 MBI_Limbo 811 BDBStabbyHideoutWorld 45 WZONESwarehouse 43 WZONESwarehouseWZA 44 ERDPGiantCave01World 44 ERDPGiantCave02World 24 WindPass_World 26 Vuldur 9 Sovngarde 763 tos_SunkenTowerWorld 6 arnimaDUPLICATE003 6303 ArnimaVoid 1638 XJKislandWorld 1534 ACX_TheGatewayRuined_Worldspace 583 ACX_TheGatewayExt_Worldspace 593 ==============================94832 Grand Total But wait, it gets better! Non-CTD Load Order Without DynDOLOD 2.14 (sorry, haven't gotten around to setting up 2.20 yet) Total for each parent world including child worlds with Use Sky Cell flag Tamriel 26941 CWTestHold 17DLC2SolstheimWorld 1696 WyrmstoothWorld 753 manny_GF_O 10 manny_GF_Desert 245 MBITreacheryCircle 81 MBIFraudCircle 10 00MBI_GluttonyWorld 33 WhiterunWorld 84 MBI_Darren_Anger_worldspace1 133 0_MBI_Dante_ARENA1 371 MBI_Limbo 156 BDBStabbyHideoutWorld 45 WZONESwarehouse 43 WZONESwarehouseWZA 44 ERDPGiantCave01World 44 ERDPGiantCave02World 24 WindPass_World 26 Vuldur 9 arnimaDUPLICATE003 1213 ArnimaVoid 151 XJKislandWorld 195 ACX_TheGatewayRuined_Worldspace 38 ACX_TheGatewayExt_Worldspace 19 ==============================Total for each parent world including all child worlds Tamriel 32087 <– Good grief, that's HALF of the total with DynDOLOD medium output active!!! Sheson, you done gone and broke Skyrim ;) CWTestHold 17DLC2SolstheimWorld 1925 WyrmstoothWorld 903 manny_GF_O 10 manny_GF_Desert 247 MBITreacheryCircle 81 MBIFraudCircle 10 00MBI_GluttonyWorld 33 WhiterunWorld 84 MBI_Darren_Anger_worldspace1 133 0_MBI_Dante_ARENA1 371 MBI_Limbo 156 BDBStabbyHideoutWorld 45 WZONESwarehouse 43 WZONESwarehouseWZA 44 ERDPGiantCave01World 44 ERDPGiantCave02World 24 WindPass_World 26 Vuldur 9 Sovngarde 763 tos_SunkenTowerWorld 6 arnimaDUPLICATE003 1412 ArnimaVoid 151 XJKislandWorld 195 ACX_TheGatewayRuined_Worldspace 38 ACX_TheGatewayExt_Worldspace 19 ==============================38876 Grand Total So now I know why I first ran into this when I installed DynDOLOD, almost at the tail end of my planned load order. Can't wait to see what results I get with 2.20 and the Temp refs toggle turned on!

Hi Sheson, Hi dunc001, I was just going to look you guys up as I stumbled onto a bit more info on this CTD yesterday. And wouldn't you know it, Sheson has already moved on this to try to mitigate the problem! Alright, I'll start with the additional bit I found. Like dunc, I've been "running lite" so to speak to avoid triggering this "overload" bug. Just a couple mods short so I can at least swap out quest/adventure/dungeon mods and play through them one at a time. Then it dropped Death Mountain 2 into my load order. DM2 has no scripts at all, not even TIFs, just LODS, meshes and textures, and an ESP of course. Since I hadn't had any problems with scriptless mods before, I thought this would be safe. Nope. Same overload CTD. O.k., so I pull out a couple of big worldspace mods and lo and behold DM2 loads. O.k., so I go back to the earlier config where DM2 CTDs.... Its not the LODs, meshes or textures or I would just CTD all the time. It's not scripts or fragments because it doesn't have any. Which only leaves what's in the ESP. So I have a look at the ESP with TES5Edit. I figure since the ESP is basically used to put static objects and recipes into the world, then fine, I'll just see if I can't start deleting sections of the ESP until I can get the game to load with it in the load order. I remove Worldspaces, nope. NPCs, nope, Locations, Containers, Constructible Object, all nope, nope, nope. I do these one at a time and if it still CTDs I drop in a fresh copy of the ESP and remove the next category. Then I delete the Cells. And bloody LOAD. All of a sudden I'm in my regular game. I can also start a new game. So I look at the Cells manually in TES5Edit and good lord, tens of thousands of Permanent and Temporary references, at about a 1/6 ratio. And many, many more permanent references in Cells than in Worldspaces (in worldspaces they're almost all temps). So I applied your countrefs script, Sheson [EDIT: I was writing as you were replying to dunc, Sheson; the countref.pas I used was the first one you let us have about three weeks ago. I'll get your new one now], to WorldSpaces in DM2 and get back just 16909. When applied to the Cells of the same mod I get back 56463. Again, not 64 000 or some such, but there's no question: pulling those references from the ESP lets the engine load a save or start a new game. It probably doesn't do much for Death Mountain 2, though... (Thankfully, I found that I can load Clockwork without triggering the CTD, especially given that Halloween is just around the corner :) O.k., so as soon as I get the chance, I'm going to upgrade to version 2.20 of DynDOLOD and see if I can throw a few more mods at this load order that would otherwise CTD it. (Here's hoping I don't mess up the update; the last time I did that was with v2.14...and it's been triple A+). Also, should someone like Arthmoor, or Nazenn specifically, be given a heads up on this issue given they're generally the gatekeepers of bugs, conflicts and engine problems?

Interesting. I set Skyrim to CTD, but I didn't. I got an ILS instead and had to kill the app with task manager after 15 minutes. The loading animation just kept going as normal, so did the fog in the lower left part of the screen, and the loading screen object could be rotated and zoomed normally (and there was no music), but the game never dropped into anywhere. Tried with both custom memory blocks active and disabled, for both destinations. Also Crash Fixes never produced a crash log.

Well, actually, the CTD doesn't happen right when clicking "new" or trying to load a save game. That's the other tip off that this is THE CTD (aside from the repeating crash address in each case). When a game loads or starts NORMALLY, about 3 seconds into the loading or start up sequence there is a very slight pause or "hiccup" in the loading screen animation, almost like the loading screen skips a frame and the game loading or setup continues normally (actually there's also one later when ENB come on, though it seems less of a momentary pause and more of a rendering change). When the CTD happens, it happens at that moment, the moment where that pause would normally take place. dunc initially said that he experienced the CTD right on clicking "new", but I recall he later mentioned there was actually a delay of about a second. We don't have the same hardware or exact load order so that could still be the same part of the sequence. Mind you, this is with OSAllocators enabled, but I get the same result with it disabled and using the SKSE memory patch (set to 1024); it just takes longer. Now, I know squat about the TESV loading sequence, but I'm assuming that the engine has a sequence divided up into different parts (naïvely something like reading the ESM/ESP data, reading the save file and then setting up and rendering the worldspace). And when it is set to "overload" it starts the first part of that sequence and fails at the end of that first part.

I can corroborate dunc001 numbers...to some extent. I can load, barely with 42 645 persistent records for the Tamriel worldspace and 59 309 total. I CTD at 42 679 persistent records for Tamriel worldspace and 59 343 total by adding Jurgholm.esp which contributes 34 records. Or I can leave in Jurgholm and pull out Here There Be Monsters - The Call Of Cthulhu and load with 42620 for Tamriel and 59283 Grand Total. And yes, my Skyrim.esm contributes 15 196 records to the tamriel worldspace while DynDOLOD contributes 17 811. I was kind of expecting to CTD over 65535 to be honest. 59 2xx to 59 3xx is kind of suspicious, but still pretty wide of the mark.

I'll try your countref script tonight as well, Sheson (maybe that one will run on my setup?). But zilav's comment would explain why dunc was able to omit generating Tamriel > persistent records in DynDOLOD until he produced an esp that would load. You guys are some really heavy artillery.

O.k., tried to run this late this evening. Got an error though. "Error in 'userscript' on line 4 : Identifier expected but ' ' found." This is with TES5Edit 3.1.3 ACTI is clearly there between the single quotes on line 4 in the script window. Weird!

Whoa! O.k., how exactly do i do this now? (Sorry, look, I work with devs and sometimes they'll say something to me like you just did, and then they see the kind of blank look on my face...they've gotten use to it.) So, ah, you want me to do this in the DynDOLOD generator? I dump these script lines into a plain text file, right-click the Tamriel worldspace in the DynDOLOD Generator and select ​Apply Script then I select the script text file in the open dialogue? Should I back up my output first? (BTW, I'll try to get this done this evening, but the in-laws are in town until the weekend and, eh, they require a lot of attention.)

Oh, and just one last thing, the testing I and DemongelRex have done around this bug leads me to believe that it has to be a particular type of record since I found (as did DemongleRex) that even large worldspace mods such as TESV LotR don't push you over the threshold as long as they don't have scripts (DynDOLOD does, which explains why it can tip the balance). You can otherwise put yourself right on the edge, add DynDOLOD for example (or even a small house mod with just a simple quest to get the house) and go over, or add TESV LoTR and not do so.

I can certainly see the logic in that, but it's also good you're limiting your expectations :) It's going to be a long time before Skyrim Special Edition reaches the heights of Skyrim Classic as far as community added content, and if Bethesda's statements and FO4 are any indication there's a good chance a lot of the hard coded, non-memory limitations are still going to be there. The String Limit, for example is still apparently in the FO4 save file format since the format is virtually identical to Skyrim's. But, in the end, you're right that it's wait and see; it will be a whole new bug slugfest. And I certainly hope we'll be seeing DynDOLOD on the other side, because mods are still going to need some high quality way of putting LODs into the special edition world.

Hi Sheson, you might remember me as the guy who was having z-fighting problems with the bridges around Whiterun with DynDOLOD 2.10 (truns out I was also having a couple of out-of-place floating window problems, but version 2.15 fixed all that and is working triple A+++!). Well, I'm also the one who posted about THAT issue (and initially HERE). Aside from it not occurring at the same memory address as with dunc, it does seem to be the same issue, and irony of ironies, the very first time I hit this CTD myself it was also as a result of turning on DynDOLOD. Mind you, it didn't take me long to figure out that the relationship between DynDOLOD and the CTD was purely circumstantial​: DynDOLOD just happen to be the mod that pushed my setup over the edge because it seems to contribute significantly to the "content" (FormIDs???) load that causes this CTD and ​it was the last thing I installed ​while "flirting with the edge." I had just come out of something like this with the String Count issue and the techniques I used to figure out that the String Count issue was non-mod specific also immediately showed this CTD was not directly related to DynDOLOD...which is why I never brought it to your attention. However...this test dunc did is...well I'm not quite sure what to make of it, but it seems like it could be a really useful hint as to the root cause of the problem. I'm referring to this: "Then started reducing by halfs, which took forever at 8 minutes a time to xEdit/save/load game/crash. I finally got it down to what I thought was a range of 256, then 128, 64, 32, 16...and finally no crash. So I added back in the last 8 records I'd removed to confirm the range of records containing the issue and it bloody loaded! So I added a hundred or so more records back in and it still loaded! Adding more back in I could induce the crash again but I now have no idea what's going on!" These records dunc was removing, these are Activator, Moveable Static, Tree, Static, etc. records he was removing from the DynDOLOD.esp? Until he was finally able to start a game? Right now, the only bell this rings is that bug that finally stopped the Skyrim MERP project dead in its tracks: the TESV engine couldn't deal with any activators, containers, etc. that were placed out further than 64 cells from the map origin. Not saying this is that bug of course, but maybe positioning isn't the only constraint on these FormIDs? So, anyway, I guess I'm not making a very strong case for getting any assistance with this seeing as I'm corroborating that this is NOT a DynDOLOD issue...but hey, look what can I say? My GF makes some awesome apple pie! Or apple crumble! With or without cinnamon? Strawberry rhubarb, maybe? Haven't head a peep from Meh either (can't say I blame the guy). But as I said in my r post, with Crash Fixes making heavily modded games actually very playable and tools for merging mods becoming more widespread this isn't the last we're going to see of this issue. I knew it would come to this one day. I knew! And no one believed me. (O.k., sorry!)

Interesting that someone else has/d the same issue. As a quick test, I turned off UseOSAllocators (UseOSAllocators=0) and CustomMemoryBlock in the Crash Fixes ini and it made no difference (haven't had time to recreate a DynDOLOD output with individual resource ESPs yet). The full bridge and their LOD meshes were still z-fighting. (But good grief it took over ten minutes to get in game! With OSAllocators...it's 20 seconds. 20 SECONDS. This just reminded me why crashing had become such a pain! Blessings of the Nine on you and Meh, Sheson!). With UseOSAllocators=1 and CustomMemoryBlock=1 and using 128MB blocks none of the blocks get to 100%. Also, I took the time to take a good look around Whiterun and in my case I didn't note anyother z-fighting anywhere.

The merged ESP consists of WhiterunExterior.esp + HighHrothgarWindowGlow.esp + DesyncBirdsOfPrey.esp + SolitudeExterior.esp, merged using TES5Edit 3.1.1. I named the merged ESP DynDOLOD Patches - Merged.esp. (And these are from the DynDOLOD Resources​ file, of course, not the Patches​ file as the latter has no ESPs.) The ESPs were installed along with their other assets, they were merged as DynDOLOD Patches - Merged.esp, individually disabled and DynDOLOD Patches - Merged.esp activated. I'll give DynDOLOD another spin with DynDOLOD Patches - Merged.esp disabled and the individual ESPs activated to see if that makes a difference. Thanks again!

DOh, wait....The EditorID of the Bridge LOD is DynDOLODPatchesMerged...which distinctly sounds like the ESP I created by merging the various ESPs that were installed by the DynDOLOD Patches 2.00 file. Something I hadn't done when using DynDOLOD 1.48 and 1.49. Looks like my merged DynDOLOD ESP is having side effects. Uh, my bad...I think. Does this have anything to do with the fact that one of the patches is for the Whiterun exterior?