0003602: Lagspike when unit try to cross map's ocean for QTPFS

View Issue Details [ Jump to Notes ]

[ Issue History ] [ Print ]

Project

Tags

No tags attached.

Checked infolog.txt for Errors

Attached Files

Relationships

has duplicate

0003594

closed

Kloot

lag spike when ordering (alot of) units into unpassable terrain (qtpfs & standard).

Relationships

Notes
~0010156 msafwan (reporter) 2013-03-22 06:13 Last edited: 2013-03-22 08:33 View 2 revisions	I have idea. Maybe the square should get bigger when look ahead?! EDIT: I have trouble using Spring.RequestPath() for qtpfs it always return NIL

~0010162 Kloot (developer) 2013-03-22 13:51 Last edited: 2013-03-22 13:55 View 2 revisions	Spring.RequestPath only works in synced gadgets for QTPFS. The rest is a duplicate of 0003594, and the ONLY reason the standard pathfinder does not create lagspikes in this situation is because it uses a HACK to limit the squares it explores to a circle centered between origin and destination which prevents it from finding valid paths in many other cases. I won't add such hacks to QTPFS.

~0010167 msafwan (reporter) 2013-03-23 01:43 Last edited: 2013-03-23 01:54 View 6 revisions	Kloot, I know qtpfs can be really cool, but can't it simply limit the square just to prevent lagspike? I think lagspike is really bad :( it is more bad than unit behaving stupidly (especially vs chicken) because when it consume 90% CPU the entire game halt, and then, no more game. Also, can we have Spring.RequestPath() for unsynced? because really need this for AI. For example: to check for location to send Constructor to build stuff (not to build underwater or in a blocked area), and for making widget that use Teleportation enterance/exit. (I would love to use qtpfs) Also, organically, no one can look ahead and see infinite resolution. The further we look, the coarser the squares (like driving car: our thumb is bigger than a big truck at distance)

~0010168 Kloot (developer) 2013-03-23 03:00	"can't it simply limit the square just to prevent lagspike?" it can but it simply won't. any such limit WILL mean cases are created where QTPFS will FAIL to find a path where one does exist and then people will report that as a major bug instead. I know this because there have been such complaints about the standard pathfinder before. "Also, can we have Spring.RequestPath() for unsynced?" maybe in the future, supporting that needs a lot of extra memory.

~0010172 msafwan (reporter) 2013-03-23 23:14 Last edited: 2013-03-23 23:15 View 2 revisions	Hi, kloot I think its really bad that qtpfs require a lot of memory. I think, why not tweak standard pathfinding instead? :P I saw that Standard already can obstacle avoidance at short range (for example it knows to go left/right when you build terraform wall quickly) and also is low cost at long range (ie: when try to cross island). I cannot understand why its bad anymore. Maybe it just had tiny bug or something that need tiny tweak???

~0010175 Kloot (developer) 2013-03-24 00:11	"I cannot understand why its bad anymore" You cannot, but I can, ok? Don't you think that maybe IF the standard pathfinder only had some "tiny bugs" I would have simply "tweaked standard" instead of writing QTPFS? "and also is low cost at long range (ie: when try to cross island)." Again: THE LOW COST IN THAT CASE IS BECAUSE IT USES DESIGN HACKS.

~0010284 msafwan (reporter) 2013-03-27 01:29 Last edited: 2013-03-27 01:33 View 4 revisions	Hi Kloot, I have discovered another source of lag. This apply to Spring94.1 Unit stuck (can't move) and it lag the entire game down. Both player has 100% CPU. Here is the Demo for Spring 94.1, at map Small Divide, at minute 9.13 the lag start when 5 unit stuck on the hill. https://skydrive.live.com/?cid=E7F297F4F70F2EF0&id=E7F297F4F70F2EF0%21121 (pls forgive the large empty square UI. The file is the empty square on left) This lag the Demo too.

Notes

Date Modified	Username	Field	Change
Issue History
2013-03-22 05:48	msafwan	New Issue
2013-03-22 06:13	msafwan	Note Added: 0010156
2013-03-22 08:33	msafwan	Note Edited: 0010156	View Revisions
2013-03-22 13:50	Kloot	Relationship added	has duplicate 0003594
2013-03-22 13:51	Kloot	Note Added: 0010162
2013-03-22 13:55	Kloot	Note Edited: 0010162	View Revisions
2013-03-22 13:56	Kloot	Status	new => closed
2013-03-22 13:56	Kloot	Assigned To	=> Kloot
2013-03-22 13:56	Kloot	Resolution	open => won't fix
2013-03-23 01:43	msafwan	Note Added: 0010167
2013-03-23 01:43	msafwan	Status	closed => feedback
2013-03-23 01:43	msafwan	Resolution	won't fix => reopened
2013-03-23 01:44	msafwan	Note Edited: 0010167	View Revisions
2013-03-23 01:44	msafwan	Note Edited: 0010167	View Revisions
2013-03-23 01:48	msafwan	Note Edited: 0010167	View Revisions
2013-03-23 01:54	msafwan	Note Edited: 0010167	View Revisions
2013-03-23 01:54	msafwan	Note Edited: 0010167	View Revisions
2013-03-23 03:00	Kloot	Note Added: 0010168
2013-03-23 23:14	msafwan	Note Added: 0010172
2013-03-23 23:14	msafwan	Status	feedback => assigned
2013-03-23 23:15	msafwan	Note Edited: 0010172	View Revisions
2013-03-24 00:11	Kloot	Note Added: 0010175
2013-03-27 01:29	msafwan	Note Added: 0010284
2013-03-27 01:30	msafwan	Note Edited: 0010284	View Revisions
2013-03-27 01:31	msafwan	Note Edited: 0010284	View Revisions
2013-03-27 01:33	msafwan	Note Edited: 0010284	View Revisions

Issue History