Corrupt Saves - StrCount > 0xFFFF - CTD on load

[bcsp]

@Zelazko

I don't think there's any such mod/tool yet.

 

Another possible way of saving some string space if you know how:

I just cut 160 property strings from a follower mod with one of those body changer scripts. They're convenient alright but you'd only ever really use it once per game. So I cut it out and set it up with the default body I use. Saves strings and no configuring on a new game...

Edited April 3, 2016 by bcsp

[GoLDii3]

Using the beta 1.48 scripts with DyndoLOD cut my strings value from 35K to 31K

[zelazko]

Was wondering what alternatives to Alternate Start - Live Another Life I should use. ~600 strong count is heavy.

I am trying to come out with alternative mods for those with heavy string count e.g. replacing Warzones 2015 with Immersive Patrols and etc.

[Gawad]

Sorry about the wall of text, but this bug has been stalking me for too long...

 

Not too many alternatives to Alternate Start (pardon the pun). It' s pretty much the go-to mod for when you want to skip the Helgen intro. At least in that vein, EpitomeOfShyness's spreadsheet has a tab for mods in the same category. This is, at this point, essential I think, as we're going to have to weigh alternatives on a benefit vs. StrCount basis from here on out. Heck, if this can't be patched, strcount will eventually need to be noted as part of a mod's description.

 

To wit, since I've been fighting this bug for months I'd already made the switch from RND to iNeed (joke was on me :wink:, Frostfall to Hypothermia Plus, VioLens back to Dance of Death, Campfire to CKotNR (and back to Campfire for the build stuff from scratch features), and Convenient Horses to Immersive Horses + Caftable Horse Barding & Simple Follower Mount. I Remember thinking I was finally going to get to try Warzones and WZ: Assault Attack with my new loadout, but my savegames went insta-corrupt at the time, as they did when I finally got around to setting up DynDOLOD 1.46. (BTW, DynDOLOD worked beautifully, didn't even noticeably add to my load times and left me a little awestruck...but also instacorrupt. And I was like, "Wow, how many mod authors can claim both that, holy cow, they fixed and broke Skyrim?) By then I'd figured out that there was some kind of limit involving script load not associated with the same issues that cause latency, but I didn't know exactly what, nor did I have a way of measuring it...just guesswork: http://steamcommunity.com/app/72850/discussions/0/517142892051135426/

 

So, yeah, almost five years in and after general memory patching (Sheson) and VRAM patching (Boris) we've finally (I think) dug up the source of a third–and most insidious–Skyrim/Papyrus bottleneck. And with long lived mods that are getting heavier and more complex over time (SL, 3DNPC, and Legacy of the Dragonborn are all perfect examples) this was a time bomb just waiting to blow. Thinking back on my 2014-2015 playthrough, at least now I have an explaination as to why it got to the point that I was saving 2 and 3 times within a few seconds as a safety measure because 1 time out of ten the first save of a set would be bad. At least the 255 ESP limit was out in the open.

 

It's actually a little unnerving that after the 28000 strings gobbled up by the base game+DLCs+USLEEP, each remaining ESP only has 150 strings to take up on average over a complete playthrough. And if you want a 15k string buffer at the start, that leaves only about 90.

 

So this changes everything in terms of setting up a loadout. I'd even say it turns some things we were taking for granted on their heads.

 

In that light, I completely ditched my setup last night (based on TESV ESS analysis it was doomed anyway) and reinstalled and cleaned a fresh set of vanilla Skyrim + DLC + HiRes Texture Pack + USLEEP. I started testing just a run through Helgen to Riverwood and back and will start reinstalling installing mods one by one (based on sets/NMM profiles such as UI, Worldspaces, etc.) and testing their cumulative effect on string count as I go. I was really hoping to do a playthrough with Defeat and Submit + DAYMoYL this time around but that requires SL so I'll wait for that to get sorted out, as well as DynDOLOD 1.48 to come out of beta before I start an actual playthrough.

 

FWIW, I've added what I had so far to EpitomeOfShyness's spreadsheet (before I hit the rest button on Skyrim) and I'll be adding more as I get results, but it will probably be over a couple of weeks. I won't be reporting back on 3DNPCs or Immersive Armors, though ; I've really enjoyed these two, but they're not worth the massive hit IMHO. Time to move on to other mods...

 

So far I can confirm that the base game + DLC + HiRes Pack + USLEEP takes up 27900 at the start.

An hour later, after going through Helgen down to Riverwood and back while interacting normally with the game world my String Count was up to 27960. But I have a suspicion that this may mostly just be related to the start of minor quests such as talking to Gerdur and talking to the Jarl of Whitterun. So it may not be as bad as 60/hr indefinitely.

 

Other than that, if anyone wants to know:

 

iNeed: 750 vs RND: 245 (went back to RND after using iNeed because it was "script light")

Campfire: 650 vs. Camping Kit of the Northern Ranger: 310 vs. Camping Lite: 10 (anything below 80(?) may just be "Quantum String Fluctuations" until we start seeing a trend)

Frostfall: 640 vs Hypothermia: 175 vs Hypothermia Plus: 465

Convenient Horses: 535 vs. Immersive Horses + Craftable Horse Barding & Simple Follower Mount: 135

VioLens: 360 vs. The Dance of Death: 30 (went back to DD because I didn't need the extra features of VioLens)

 

So just with those you can go from a max of almost 3000 to a min of 635, saving up to ~2370 strings, depending on what features you can live without.

 

Anyway, I'm not counting on Beth to fix this, they stated quite clearly that the Legendary edition was the very last time they were touching Skyrim (at least as far as TESV)...and I can imagine the processes internally: bug fix request finally gets someone's attention, evaluation by management, fix approval, finding a dev to assign this to, getting the server admin to find the source backups, set up a VM on an overcrowded server, do the patch, go to testing, find out it has/may have impacts elsewhere, back to the dev...who is supposed to be billing to FO4 DLC...guys, this is NOT GOING TO HAPPEN. I'll eat a chaurus egg if it does.

Edited April 6, 2016 by Gawad

[EpitomyofShyness]

EpitomeOfShyness's spreadsheet has a tab for mods in the same category.

 

I honestly think I'm gonna combine them into a single sheet. Trying to figure out in what way exactly, and probably won't happen until this weekend, since I only have access to my desktop on Wednesday, and prefer to work on my own load order when I can. Also thanks for adding in info, every bit helps.

[Gawad]

Here's a thought. Once more strcount numbers are on a firmer footing, one place they could be really handy to have is LOOT. Along with the rest of the info in its masterlist, an estimated string count of your loadout (per mod and total) could immediately let you see how risky your loadout is becoming.

 

Still, this probably wouldn't be as easy as just plugging in numbers averaged from a bunch of modders if it's to be accurate, as from what I've understood where properties overlap in a mod order they don't get duplicated in the string count table (which may explain some of the variation between results).

 

Of course it's far too soon to implement anything like that right now, and the issue may yet get patched through SKSE, but does anyone here know if any members of the LOOT team, such as Sharlikran or Aellis, are aware of the issue?

Edited April 6, 2016 by Gawad

[BoredErica]

I estimate I would need no more than 5000 string buffer from starting string count off the cart ride for an entire playthrough. That means I doubt I will ever hit 50k strings.

[Gawad]

That may very well be true for your loadout. But loadouts vary enormously. And when someone's string count is going up overtime, how do you know which mod is pushing it up? In the first 10 to 50 hrs of a playthrough, how much is just the base game contributing? Overall, you could graph a trend for your overall (stable) loadout and get a good idea where you are going to plateau over time. Or someone who is a couple of hundred hours into a game then adds a mod and sees their strcount explode could do the same for that mod. But what do you tell someone who wants to do a playthrough using, say 200 very light to heavily scripted mods from among roughly 2000 more or less common mods? There's the rub.

 

Also, performing a playthrough can mean different things to different people. E.g., it might mean going straight for the main quest and civil war questline as fast as possible for some, completing one, most or all major branch quests to others, or even exhausting all side quests to others still.

 

Of course, one thing that can be done is to hedge your bets, so to speak, at least for certain quest mods. For example, Summerset Isle is so far reported to use up between 500 and 600 strings. But you can't even start it until you're Archmage of the Collage of Winterhold, so you might want to hold off on installing it with an initial loadout, try to account for an additional 600 strings in your buffer and see how the strings settle 20 or 30 hrs of your playthrough.

Edited April 6, 2016 by Gawad

[DavidJCobb]

And when someone's string count is going up overtime, how do you know which mod is pushing it up?

That's a good question...

 

Exactly how much is known about Papyrus save data? I haven't been able to find public documentation, but admittedly I wouldn't know where to look. If these strings exist in the save data, they must necessarily be referenced in the save data. If the data can be decoded, perhaps someone could build a utility to identify what's using strings.

[BoredErica]

That may very well be true for your loadout.

 

I estimate I would need no more than 5000 string buffer from starting string count off the cart ride for an entire playthrough. That means I doubt I will ever hit 50k strings.

 

It is true for my loadout and that is what I care about the most. Since I've made it clear in previous posts that I was testing my loadout without even releasing my mod list, it makes no sense to assume I was trying to make up a rule about string count for everybody's load order. I am aware of the basic fact that different people have different mods installed.

 

If you are unsure about the specifics of how and why my test is done, then you should've asked instead. By playthrough I mean literally do every single quest and objective in the game. I don't need to sit around and figure out the string count of every single mod because as it stands right now I don't see myself ever hitting the cap. An entire playthrough takes many days of nonstop playing. Less inclined to do it if the data doesn't affect me directly. I've managed to complete a decent chunk of the game on the new save file and extrapolated the data to try to estimate how many strings I will end up with once the save file is done. It is of course, a rough estimate. Too rough to be really useful to anybody else, but close enough to tell me all I need to know.

 

We already know how many strings the base game + DLC + USLEEP takes up give or take 1-2k strings. And if we know how many strings a finished save file takes, we just need to quickly check the strings used at the start of a vanilla game to figure out how many strings are added by the game itself absent of mods from start to finish. The answer to that should be equal to or less than 10,000 strings.

Edited April 7, 2016 by dark_wizzie