Bruce Hake Bruce Hake

Clogged Furball

Clogged Furball

Ran into a situation last night I hadn't seen before: a battle so big the game got clogged up and could not resolve damage, although it did not lock up or stop.

I was playing on a custom map with three stars, each of which had just five planets. One star was me (TEC). One was Advent. The last was Vasari. Both AIs set as Hard Random.

I had wiped out Advent and colonized all of their planets. I was ranked #1 in everything except Fleet, even though I was maxed out on fleet size. Vasari (haven't played them yet) must have some way to boost their max fleet size.

My main invasion fleet tried to attack the Vasari homeworld but got cut to shreds because they had so many strikecraft and I was short on flak frigates. So I made a second fleet with about 40 flak frigates. Then both of my fleets attacked their homeworld again at the same time. I had 9 cap ships, 7 of them at Level 10. I had 55 air wings and a total of about 200 ships. The Vasari had 50 cruiser carriers, lots of defenses, and a total of about 200 ships.  I'd say each fleet represented about 1100 fleet supply.

And the game couldn't handle it at all. For a solid 10 minutes I watched to see what would happen. Battle appeared to be taking place, but I did not lose a single ship. I kept monitoring the enemy fleet summary on the side of the planet icon when you zoom out, and it kept showing the numbers of fighters and bombers fluctuating--but never dropping below 20 each, and I never killed a single ship! When I would zoom in, things looked weird--only about five percent of the craft were in the grav well--the rest were all smeared out, green (me) and red (them) interspersed, in a long line outside the grav well (but from a zoomed out perspective, they were not in space or in phase lanes but just in orbit at the planet). At this point the game was also running painfully slow, despite a cold reboot, and was painfully slow at zooming.

Eventually I tired of this logjam and moved my main invasion fleet away. Then the game was able to resolve things, and the Vasari destroyed my fleet of flak frigates in about 30 seconds, before I had time to warp them out.

So then I ran my main invasion fleet around and wiped out the remaining Vasari planets, playing cat and mouse with their very huge defense fleet, which kept chasing me, and I won the game in just another 15 minutes or so.

But it was kind of frustrating. I had the power--I should have been able to smash their main fleet in the battle I had chosen.

I've been playing like crazy since the day this was released, and this is the first time I've had the game just plain clearly get overwhelmed by fleet size. So far I haven't had a single "minidump" or other lockup. I have noticed that the game gets a bit sluggish after a few hours sometimes, but a boot cures that.

Comments?

P.S. Another question: I noticed in that game last night that when zoomed in to see a grav well, the camera would always keep moving around, often apparently in a 360 degree motion, even when I did not even have my hand on the mouse.  I turned off "shake camera" in the options, but that didn't seem to make a difference.  Is the camera supposed to move, at all, ever, if you're not actively moving it yourself?
60,691 views 64 replies
Reply #51 Top
I reloaded an old, old save from that game, where fleets are still getting enormous and performance is beginning to degrade.

Free Image Hosting at www.ImageShack.us

My CPU speeds and associated performance with speed changes. Seems odd that CPU usage drops at 8X, but performance is the worst then.

Perhaps Sins is doing some sort of massive RAM swapping mayhem?

Notice that Core 0's performance is roughly INVERSE that of Core 2's.
Reply #52 Top
It's been about 10 years since I did any C++ programming, so my memory is beyond rusty. What I meant by "database" was that I think the program may have some kind of overrun in the internal datastructures it builds up to represent battles. Just speculation.
End of quote


That is potentially much, much worse than a memory leak. Let's hope that it isn't that and there's a 1.05 patch if that is the case.
Reply #53 Top
A memory leak only kicks in after a while, not immediately. If you're loading a saved game and the problem occurs immediately, the flaw is more fundamental.

What puzzles me is how one person gets the issue but another doesn't. Isolating the difference between their systems might give some clue as to what the underlying cause is.

Uranium - a theory on why 8x seems to use lower CPU than 1x: you're likely zoomed out more? When you're playing 1x, you're dancing around zooming in at planets and ships and watching combat closely for micromanagement purposes, and as a result, your system is performing a lot more graphics and math calculations due to the models on the screen. When you're at 8x, you're doing more things at the galactic map level when the number of graphic and pictured objects you're interacting with is a great deal less. What is the time interval on your graph?

Also, have you tried playing without the Sins+ mod, and with reducing to medium settings on graphics?

-- Retro
Reply #54 Top
Uranium - a theory on why 8x seems to use lower CPU than 1x: you're likely zoomed out more?
End of quote


Nope, I took that sample without doing a thing, except changing the speed (was all the way zoomed out at any rate). The trend is pretty much the same no matter what zoom I'm at anyway.
Reply #55 Top
I'm running 100% stock, no mods.
Reply #56 Top
It might be RAM related, I am running G-Skill DDR3 1600 RAM, which is blistering fast.
Reply #57 Top
As I said, run a 9-star 9-AI game with large fleets, so how the works out for you.
Reply #58 Top
I'll give it a shot and post results.
Reply #59 Top
I thought about the issue last night and thought about some of my prior games in this...I also realized something else. Though it seems unrelated, AI modding (at least for me) in Red Alert 2 and Yuri's Revenge experience quite a bit of slowdown as well (though it is virtually unnotceable with today's PC's). By complicating the AI triggers a bit, Stardock may have greatened the memory usage severely. That's another possibility, but brings us all no closer to remedying the problem. The question (if this is the case) is: horrible AI or performance? Also, depending on the way the game is loading files and ships, (this involves an understanding of how RAM is used) the game could have a rough time finding a proper fit for everything. Also, an internal error (often caused by and causing unusual things happening at the same time) could have caused more issues. Just throwing out a few ideas.
Reply #60 Top
I'm the original poster in this thread. As I wrote today in a new thred entitled /General/The Battle of Gethsemane, I found the autoreplay for the clogged furball situation, and I think it's fascinating to watch.

However, I don't know (1) how to find the autoreplay file on my harddrive, and (2) how to post it here. So far nobody's answered those questions in that thread, or in a short post on this I made in the /Technical forum. Please somebody let me know so I can post it. I'm really curious for feedback on it.
Reply #61 Top
Your most recent thread about replays has answers now ;)
Reply #62 Top
Thanks, Annatar11! You're very helpful.

You can find my very interesting "clogged furball" situation on the following replay:

http://rapidshare.com/files/107836397/Furball.zip.html

It's only about a 400K file.

Watch what is happening on the planet Gethsemane and its neighbor (in the red Vasari solar system).

Use 8X with everything revealed. The really interesting stuff starts after a few minutes. Watch what is happening with all the fleets at Gethsemane, especially after my second fleet with lots of flak frigates turns up. Also notice what happens after I get tired of the standoff, jump to the neighboring planet, and jump right back.

There are several threads on this apparent issue with Sins 1.04: /General/Clogged Furball, /General/The Battle of Gethsemane, and several related threads in the /Technical forum.

I went to a lot of trouble to figure out how to post this, because I think the situation is quite fascinating. Curious for feedback.
Reply #63 Top
Looking at this now... curious if it is the same problem with Malice or not.
Reply #64 Top
Nope, my battle is TEC vs. Vasari, so no Malice involved.

I'm sure happy to see somebody checking this out! Please post what you think.