Loading...
  OR  Zero-K Name:    Password:   
Title: [A] Teams All Welcome
Host: Nobody
Game version: Zero-K v1.8.4.4
Engine version: 104.0.1-1477-g8ecf38a
Battle ID: 868211
Started: 4 years ago
Duration: 33 minutes
Players: 30
Bots: False
Mission: False
Rating: Casual
Watch Replay Now
Manual download

Team 1 Lost
Chance of victory: 62.7%
XP gained: 43
ILranknimor died in 33 minutes
CArankSippio died in 33 minutes
DErankAdminmojjj died in 24 minutes
ITrankThomas1 died in 33 minutes
GBrankddaboqepp died in 32 minutes
EErankNorthChileanG died in 33 minutes
ATrankRomux died in 33 minutes
BErankAmpul died in 33 minutes
USrankArchangel died in 33 minutes
USrankHellaciouss died in 33 minutes
DErankarchbishop died in 33 minutes
GBrankPlayerOne died in 33 minutes
DErankSnowlob died in 33 minutes
USranktravilnmic died in 33 minutes
ITrankDanShingen died in 33 minutes
Team 2 Won!
Chance of victory: 37.3%
XP gained: 53
unknownrankTinySpider
USranksomekid
unknownrankScolopaendra died in 11 minutes
FRrankHelwor
GRrankTheLyraki
FRrankOrientar
ATrankxxRezipro
ITrankMakT
USrankchaplol
PLrankKnightOfNee
USrankVivianite
IQrankAroura96 died in 33 minutes
USrankDapperDan
FRrankSIM
FRrankDelwendur




Preview
Filter:    Player:  
sort
This replay should be analyzed for what caused massive performance drops around 24 minute mark.

Unsure if it was graphical or server was dying but this is the first game I have had to "catch up" and not much was even happening. Played much larger games without issues previously. I'm guessing it's flea uncloak area detection being so large, or just fleas in general. When a flea blob moved performance tanked really bad.
+0 / -0

4 years ago
+300 windgens in map was already built in 14 min mark. Many players started already lagginfg at that point. Some ven faster. I have feeling that GBrankddaboqepp started to supply even Russia with this green windgen power.
Sadly there was singu in lower righ corner and until 14 its still wasnt connected to grid. Player started build second singu near seems for connecting purposes.
+0 / -0


4 years ago
I don't see anything in particular. 1700 units is a lot of units.

Optimise pathfinding, construction priority, terraform and LUPS. Moving the handling of move commands back into the engine could help. The 'lowest' hanging fruit here could be looking inside LUS and trying to improve it, since I don't think that has been tried before and lua coroutines could be expensive.
+0 / -0


4 years ago
You could analyse the replay and give some precise times at which performance drops unexpectedly. Then I would look into it again.
+1 / -0
4 years ago
I've pinpointed it to exact time of 23:57. Though I do not know what caused it.

Nearly everyone in game has their CPU and ping spike. I'm assuming ping in this situation indicates that their simulation is behind, but the cpu spike is pretty clear.
+1 / -0