Jump to content

epu

Members
  • Posts

    27
  • Joined

  • Last visited

Posts posted by epu

  1. I gave up in October. I'm sorry to see it hasn't been addressed yet; I am the best market for this product (a patient RPG lover who grinds through and replays for content, and has worked on a game before), and I haven't revisited the title or bought add-ons. And I have a lot of patience. But I have a family, and only time to play a small number of games annually to the length of play hours they deserve.

     

    Some thoughts;

     

    The level save menu is a different problem from the level data / persistent item problem. Probably the level save menu is an easier fix; cap the number of save slots to 10. Or work harder to make an index that loads quickly async. There's probably a way to engineer it. But either way, users would never encounter long lists and multiple-minute save menu entry times (the first load of the save menu to display the saves themselves).

     

    The persistent item problem is harder, because it degrades over time; as the complexity of the save game requirements skyrockets, you still need access to all the things you've ever touched - they control the game state going forward, so their status can be updated (or used to inform what dialog is available) whenever you go visit something. I'm sure the team has already made hard tradeoffs here. Maybe they'd have to be harder tradeoffs that make the story unsatisfying (make story bottle necks, where you toss out all but a few flags, but now dialog for the people you re-interact with is no longer custom to that play-through). Or discard entire scenarios or maps once you make a decision, so they carry fewer data points down to the end-game. But would that feel real and immersive? Maybe those decisions/flags are stored in multiple text files instead of a faster database or in-memory filesystem, and speeding it up is too risky or too much engineering. Can they afford to re-engineer something as basic as saving/loading?

     

    I can appreciate that making games is hard, and the tradeoffs are hard. If we, the consumer, are paying for awesome writing and art, it might mean less money on engineering, or spending that engineering on unity bugs instead of load/save features. Maybe developers have rolled some of those engineers forward to work on another project, or to improve the engine and tools features for development, and can't fund more engineering on low priority issues on an already-shipped title. All of those things happen, and we don't get to see every piece of that.

     

    Usually we just get the game.

     

    I would ask you, developers: please, rather than sell new material, ensure your next title's load/save system is snappy and responsive. Solve the persistent data/items problem better, or make a different tradeoff that makes it so we see your entire game and experience the entire story (and as many branches of it as we are able to). Because the whole team made it, and even the end game deserves to be enjoyed for the developers who made those final encounters, and the customers who are completion-ists.

     

    I don't mean to be an apologist for the developers. Developers, I love your game, you should be proud of it.

    But these loading flaws are there. In a better world, they should be fixed.

    Beyond gently asking, it's out of my control.

     

    Thanks for reading, and sorry about the length.

    As your fan, I feel strongly about this.

  2. Time to bust out dtrace/profiler and see just what in god's green earth is going on in there.

     

    Here's a set of IO read/write timings from the main menu, to my current progress in game. That's getting into the load game menu, followed by the level load.

     

    It's the output of the command 'sudo iosnoop -p <PID> -A | ts "[ %Y-%m-%dT%H:%M:%S ]" | tee ~/iosnoop.log.0', where <PID> is the running game's pid. This is something you should be able to run on OSX, and maybe linux. Probably not on windows :(

     

    If you want to get a rough feel for what's happening, it's still a little challenging. There are file reads, and lots of file writes. Maybe all the levels you've ever played in the save file are being extracted? I'm not sure what the null file descriptors are pointing to. I'll try some other dtrace scripts.

     

    For those of you following along, all I'm doing is loading the game to main menu, opening a terminal, trying out scripts that measure system performance while loading into my newest (largest) save. I'm just poking around based on reading http://dtrace.org/blogs/brendan/2011/10/10/top-10-dtrace-scripts-for-mac-os-x/

    • Like 1
  3. I wanted to add my objective and subjective experience on a mid-2011 macbook air. Saving games in Noonfrost (which I hadn't been timing before) feels very fast in 2.0.1 steam build, compared to the 1.0.6 steam builds. Loading files from Act III (Blood Sands and Noonfrost) takes about the same average time (23 seconds and 29 seconds respectively). The load game menu continues to take about 1 minute to load with the number of saves I have (253 saves).

     

    I'm not sure if the load optimizations don't apply under OSX, or if my machine is so old that the performance bottleneck is somewhere else.

  4. Hey there,

     

    Just upgraded to the 2.0.1 steam patch, and wanted to thank you for addressing the zombie chanter skeletons. When I reload blood sands at that point, save the game, and search for skeletons in the zipped archive's .lvl segment that I am active in, I don't find any references.

     

    I wanted to share that load speed in blood sands is about the same (saving feels faster now though, but I hadn't been measuring it before).

    Previous to the update, loading the blood sands took 23 seconds on average. This is slightly up to 24 seconds (after resaving the world using the 2.0.1 game).

     

    Loading much later inside Noonfrost (after completing Maw and Elmshore) takes about 29 seconds on average.

     

    Kind regards,

     

    -e

  5. Hey team and fans,

     

    I was experiencing long load times in Blood Sands, and found multiple threads about extra chanter trap nodes in save files, and extra summoned helpers.

     

    This thread seems like it was answered in April, but it's a different save game component / subfile.

    This other thread talks about 20 hours slow loading, but I'm at 70+ hours having never completed the game yet - it's brutal going.

     

    I did some string counting inside my blood sands save file, and inside the unzipped contents, region AR_1305_Blood_Sands.lvl, I have 600+ summoned skeleton references (cre_skeleton_chanter_summon). That seems high for a single world, but I have been fighting a lot in there with Kana after slaying the head priest from the Berath vision.

     

    Is there already an internal bug tracking chanter summoned skeletons (cre_skeleton_chanter_summon)?

    Is there any attention in the upcoming patch to summoned creature cleanup / temporary object cleanup per-level-region?

     

    Kind regards,

     

    -e

  6. Checking on the status of this. I'm in Act III in Blood Sands, and my loads are starting to become intolerable. I scanned through an uncompressed save file, and I have 600+ chanter skeleton summoner references, in one blood sands save alone, without Kana or my custom adventurer chanter in my party. I'd be especially interested in these making the next patch.

     

    This is OSX 1.0.6.1619 -steam

  7. Ah. My chanter's (I have 2 of them) summoned skeletons are maybe being dragged around? If I grep the unzipped blood sands file's strings for cre_skeleton_chanter_summon, there are 624 copies of the string in just the blood sands save alone. Are these from the "If Their Bones Still Slept" chanter rime? I don't have Kana in my party currently (although I used him throughout Act I and Act II).

     

    Here's a copy of just the one blood sands save

  8. I've had this come up a few times, and the fix is just to click on the formation again. This seems to "reset" the positions for everyone.

     

    For me, seems like the trigger is when I recruit/dismiss a character. I guess the game doesn't re-apply the formation in that case and they get some sort of default position.

     

    Thanks for the workaround! This is still open in 1.0.6.0617 -steam (save and screenshots). I had swapped in 2 party members in Caed Nua also, and Sagani/Ituumak are one of the active party before the swap.

  9. I'm not sure how to repro it, but I have a save and screen shots that illustrate the problem after it occurs (formation doesn't match the companion positions). I had just swapped out 2 party members (drop Eder, Kana, add Pallegina, Hiravias) from Od Nua in the Keep, marched out of the grounds, and noticed something odd. I didn't think to test it until I got to the overland map, and figured out they weren't in formation. I haven't tested the work-around of reassigning everyone manually. I just used the other slot (slot 2) which I haven't used before, and it seemed ok.

×
×
  • Create New...