The pause generated in the first frame of the game by feature place spawning features is long enough to generate a false hang detection, making the rest of the game < 1FPS and unplayable.
Since hang detection was added, any maps that have a lot of these features have major problems with my machine. There must be an alternative to spawning them all at once in the first game frame?
Major Issue with feature placers and hang detection
Moderator: Moderators
Re: Major Issue with feature placers and hang detection
Come back with better evidence that it happens all features instantaneously and this is why your shit is crashing..
Re: Major Issue with feature placers and hang detection
It happens on forebodings evorts maps with the huge numbers of trees.
Maps like comet catcher etc do not have this issue. It is only maps with lots and lots of features, maps that use lua to spawn them
As I said, if the time to spawn features > hang timeout == crashlog + pathetic framerate
Attached is an infolog showing the problem, and the multiple crashes that happen afterwards it causes
Maps like comet catcher etc do not have this issue. It is only maps with lots and lots of features, maps that use lua to spawn them
As I said, if the time to spawn features > hang timeout == crashlog + pathetic framerate
Attached is an infolog showing the problem, and the multiple crashes that happen afterwards it causes
- Attachments
-
- infolog.txt
- (52.78 KiB) Downloaded 119 times
Last edited by AF on 16 Jan 2011, 16:49, edited 1 time in total.
Re: Major Issue with feature placers and hang detection
AF is right, when a Lua gadget takes more than the hang timeout (default 10s IIRC), then Spring goes into hang detect mod which makes it all the worse.
Don't act so rudely defensive, Smoth, the problem is not your shit but the anti-hang thing.
Don't act so rudely defensive, Smoth, the problem is not your shit but the anti-hang thing.
Re: Major Issue with feature placers and hang detection
How come we don't have a stage of the loading screen to add the features in?
May I also add that although Shard is running in the game I gave the example for, shards operations in frame 1 are near nonexistant. All shard has done is setup a lua VM in the Init call the engine makes prior to frame 1, which takes an almost imperceptible amount of time. Widgets running are the standard Evo RTS widgets.
This issue is also dependent on the load placed on the machine, which in turn makes spawning the features slower. If I run at 1024x768 with no other programs open its fine. If I run at 1680x1050 my native resolution, with a browser and photoshop open it happens 100% reliably due to the increased time taken to load.
My machine is getting on a bit these days but theres people with far worse than me. You can see my specs in the infolog.txt above
May I also add that although Shard is running in the game I gave the example for, shards operations in frame 1 are near nonexistant. All shard has done is setup a lua VM in the Init call the engine makes prior to frame 1, which takes an almost imperceptible amount of time. Widgets running are the standard Evo RTS widgets.
This issue is also dependent on the load placed on the machine, which in turn makes spawning the features slower. If I run at 1024x768 with no other programs open its fine. If I run at 1680x1050 my native resolution, with a browser and photoshop open it happens 100% reliably due to the increased time taken to load.
My machine is getting on a bit these days but theres people with far worse than me. You can see my specs in the infolog.txt above
Re: Major Issue with feature placers and hang detection
Ok, that map was using my old spawer which does frame 1.
NOW does this happen on grts maps, specifically the new ones?
I cannot do anything during the loading stage.
NOW does this happen on grts maps, specifically the new ones?
I cannot do anything during the loading stage.
- Forboding Angel
- Evolution RTS Developer
- Posts: 14673
- Joined: 17 Nov 2005, 02:43
Re: Major Issue with feature placers and hang detection
^^ This. Those maps are using very old versions of feature placer spawner.smoth wrote:Ok, that map was using my old spawer which does frame 1.
Re: Major Issue with feature placers and hang detection
it is important to see if the others do the same shit.
I suspect it is the one size fits all nature of the hang detection. I think it is flawed and am trying to see if I am wrong in this. I do not know enough about it yet but like always I am willing to start looking into it.
I suspect it is the one size fits all nature of the hang detection. I think it is flawed and am trying to see if I am wrong in this. I do not know enough about it yet but like always I am willing to start looking into it.
- Forboding Angel
- Evolution RTS Developer
- Posts: 14673
- Joined: 17 Nov 2005, 02:43
Re: Major Issue with feature placers and hang detection
I object to tests like this. For one, you run it on a laptop and even then it's a damn MAC running bootcamp. How the hell is that in any way supposed to be "Fast"?AF wrote:with a browser and photoshop open
Re: Major Issue with feature placers and hang detection
Its fast because a lot of my uni courseworks built on ym mac in windows 7 vastly outperform them running on my desktop. Windows on a modern Mac isnt like a virtual machine or a stunted platform, its actually quite fast once its setup.Forboding Angel wrote:I object to tests like this. For one, you run it on a laptop and even then it's a damn MAC running bootcamp. How the hell is that in any way supposed to be "Fast"?AF wrote:with a browser and photoshop open
The test you object to was ran on a custom built desktop tower, not an Apple machine.