Universe Sandbox
Universe Sandbox => Universe Sandbox ² | Support & Bugs => Topic started by: smjjames on October 20, 2014, 02:31:06 PM
-
Okay, this is an odd one, whatever it is. While doing the ceres glancing collision with earth, I saw that the output log had ballooned with 'An error occured that wasn't supposed to. Contact support.' and 'An invalid object handle was used.' error. However, the first time I did that was with the particle limit at 10k (I normally have it at a reasonable 2500) and I thought maybe it was the fact that I did it with that many particles. That initial finding I already sent to you guys, along with the eventual conclusion of the sim after the output log had apparently grown to 132mb and I found that the collisions apparently broke. Though the two errors mentioned above did stop, 99% of that filesize of the one I sent you is just repeat errors. It took so long for the feedback thing to send the report that I thought I had broken it.
This time however, I tried the particle number I usually do and hit it.
To reproduce, I used the Earth and Ceres skimming as the template, upped the impactor speed to 50.5 km/s, stopped the impactor when close to impact and set the timestep to rougly real time.
Going to do a little more research on it. The default sim parameters don't produce the error, even on approx real time timestep. Actually, I'm wrong, it happens with the default sim parameters on approx real time timestep.
Tried default sim parameters and colliding at 10.00 sec/sec timestep, it didn't do the errorspam. Though the crater it makes is kind of strange.
Tried the same with the impactor speed at 50.5 km/s, the error spam doesn't happen.
Tried 50.5 km/s impactor speed with timestep at approx 5 secs, error spam doesn't happen.
Tested the earth moon grazing H fast (adjusted speed to 50.5 for testing purposes) at realtime or near realtime timestep and while it takes a bit longer, the errorspam happens too.
So, theres something about really high speed collisions while using real time or near real time timestep? I also noticed that it causes the sound to happen, despite me having master volume set to 0.
Going to test out head-on impacts.
Edit: Head-on impacts are fine, I guess off-center impacts at realtime timestep cause the physics engine to tie itself into a pretzel or something?
Edit: Heh, I set the number of nonattracting particles to 10k, did a head on collision with ceres and earth and produced the errors again.
I broke the collision of fragments with the larger body again with that last sim.
Edit: Heh, even larger bodies collision broke. I'll upload the output log for this one which utterly broke. Word of warning though, it extracts to 90mb in size.
-
"An error occurred that wasn't supposed to."
- removed -
-
Har har, real funny xriqua, no. It wasn't doing any obviously broken behavior, though it was really laggy during that period that it generated those errors and later at some point, the collisions simply stopped working.
I was definetly surprised to see that error though.
-
I meant that as a legit question: Is that what the error message really said?
-
I meant that as a legit question: Is that what the error message really said?
Yep, it's right there in the output logs.
-
I'll look into this and see if I can trace the error code. Thanks.
Update :
Looks like it's an error from FMOD. Really odd phrasing though.. hmm. What Audio device are you using? I've not seen this error in the past.
Udate 2 :
Initial research is pointing to corruption in one of the files. Can you try redownloading a clean copy of the game?
You can also try verifying the game cache first, to see if it can detect the damaged files.
https://support.steampowered.com/kb_article.php?ref=2037-QEUH-3335
-
Did you guys manage to get the one with the massive 132mb file though? Wasn't even sure if it did get through. As for audio, I'll check that and I can do an uninstall and reinstall.
Sound driver info pulled from System information, let me know if running a DXdiag and giving the output would be more useful.
Name Realtek High Definition Audio
Manufacturer Realtek
Status OK
PNP Device ID HDAUDIO\FUNC_01&VEN_10EC&DEV_0282&SUBSYS_1043119D&REV_1000\4&6824705&0&0001
Driver c:\windows\system32\drivers\rtkvhd64.sys (6.0.1.6954, 3.30 MB (3,462,616 bytes), 8/18/2013 1:02 PM)
Name NVIDIA High Definition Audio
Manufacturer NVIDIA
Status OK
PNP Device ID HDAUDIO\FUNC_01&VEN_10DE&DEV_0041&SUBSYS_10DE0E0B&REV_1001\5&32946537&0&0001
Driver c:\windows\system32\drivers\nvhda64v.sys (1.3.30.1, 192.78 KB (197,408 bytes), 2/28/2014 8:46 AM)
Name NVIDIA Virtual Audio Device (Wave Extensible) (WDM)
Manufacturer NVIDIA
Status OK
PNP Device ID ROOT\UNNAMED_DEVICE\0000
Driver c:\windows\system32\drivers\nvvad64v.sys (1.2.23.0, 39.45 KB (40,392 bytes), 8/1/2014 10:29 AM)
Cache verification says that all of them verified, I think. Going to do an uninstall/reinstall.
Edit: Did reinstall, verified cache twice, then ran the 'ceres skimming earth' collision (no alterations), stopped it near collision and ran it at approx realtime (where the timestep says 1.00 sec/sec), the error spam happened again. Going to try and lower graphics quality and see if that does anything.
Edit2: Nope, no change, if anything, it may actually be a bit laggier, and the errorlog still spams the same thing. And by lag, I mean like it's laggy when you move the camera around, even when paused.
-
If it's outputting large amounts of errors to the log, it's going to run really slow. So that in itself is pretty normal.
I'll keep doing some testing and research to see if there's some way I can reproduce this.
The drivers for your realtek are fairly old. It's probably built into your motherboard. You can check with the manufacturer of your motherboard (if it's a custom pc), or the manufacturer of the computer itself, to see if there are any updated drivers for it.
Couldn't hurt to at least check that in the meantime.
Edit / Update :
After looking into the problems, there's really two issues going on.
The first is a problem with the audio systems. Right now it's generating too many audio requests in a short time, and some drivers / audio hardware don't handle that very well. To fix this one, we're going to need to redo how the audio works and enforce better limits. This one might take a little bit till we have time to redo that system.
The second issue is a bug with the line rendering. Chris has just committed some fixes for it, so we should have that issue hopefully resolved in the next update. (This is likely the cause of your slowdown and huge log files, as this error is the one filling it up)
If you're still having issues after the Alpha 12 update, let me know.
-
Okay cool, what about the collision system completely failing later on (well, it works for a while and then at some point, it just stops working)? No idea if that's because of it freaking out due to the audio issues and the line rendering or some other issue.
Both the 132mb or so one that I sent you guys through the ingame bug report system and the second output log above are after the collision system completely breaks down.
-
Just to clarify, the main bug you're having is the error that's filling up your logs and causing a big performance issue during collision. This should be fixed in Alpha 12. (the collision bug you mention)
As far as the sound thing, it shouldn't be causing major issues. We have a planned upgrade that should take of it , as well as improve the audio in general. What we have now is just a placeholder for testing.
-
So the collisions breaking completely at some point is just a side effect of the main bug? okay.
-
Yea, I suspect that is the case. However, until we get the main error fixed, we can't be sure. We'll follow up if you're still having the problem after the update.
-
Still happening, I did the exact same setup with the ceres skimming earth sim. Sent you a feedback with 'attention C7' in it along with a screenshot (which might provide some clues). The first 58kb of the output log I sent you is normal though.
Restarted Ubox because the impact lines didn't actually turn off, though it probably doesn't impact performance. Going to see if collisions completely break down later on.
Edit: Looks like the max decals setting doesn't actually work, I tried lowering the decal limit to 10, but it's still having many thousands of decals (5610 at this paused moment), maybe that's part of the problem.
Edit2: The error spam isn't as bad as before, stopped at 2,250kb.
edit3: Somewhat over an hour in (on real time timestep) and no fragments yet, plenty of plasmafied rock (dust) though and Ceres is losing mass.
edit4: Nearly two hours in sim time, shouldn't some of the particle clumps have coalesced into fragments by now?
Edit5: 2.71 hours in, still no fragments. Also, I gave Earth it's 24 hour rotation. Going to let this run for a couple hours and see what happens.
-
Thank you for the logs, I'll look into them and see what's happening now.
In regards to heating, things don't cool down in a vacuum very quickly, as many of the major sources of heat transfer don't occur. You're only meaningfully subject to radiant heat emission at that point. Advection, conduction, and convection, really some other medium to be present for effective transfer to occur.
http://en.wikipedia.org/wiki/Heat_transfer
Specifically, this is the main factor in space and the basis of our heat simulations.
http://en.wikipedia.org/wiki/Stefan%E2%80%93Boltzmann_law
-
Well, definetly better than the simulation spamming hundreds to thousands of small fragments in an attempt to simulate shattering.
Also, apparently, after I gave the earth it's rotation, Ceres is now rolling/bouncing along at about 3 m/s. I was going to restart the simulation, but I had to go and do stuff, so, yeah, the sim will run for a while.
-
Update to it, several hours later, theres still dust flying around and ceres has been rolling around the equator not making any other marks.
-
*bump*
Any progress on it?
-
No update on this just yet. I'm working on some critical bugs at the moment in other areas. I've not forgotten about you though, and I'll get back to your report when I have some spare time. (You don't need to bump the thread ;) )