[Beta 18b] Creep stalls

[Guest Sancdar]

Quark still makes creeps pause a bit, very noticeable with 10 quarks hitting the same creep.

[Karawasa]

I tested this with a batch of 10 and then a batch of 20. 20 did cause noticeable stuttering (but not game breaking), 10 not so much (could be computer dependent though).

This is an important point of clarification, is it 10 quark towers attacking or 10 quark towers attacking the same creep? Also, any other 0.66 speed tower that uses the DDS should suffer from this problem (Impulse, Tidal, Flamethrower etc.). Anyone mind confirming?

[dvdlesher]

I'll try, I need that tower for my build in forum draft anyway

[jolin012]

I never felt it even with for example impulses+well lvl 3 but did feel it with quark and flamethrower had a super fast attackspeed before and never lagged them anyway hmm..

[Guest Sancdar]

This is an important point of clarification, is it 10 quark towers attacking or 10 quark towers attacking the same creep?

With 10 attacking I think that all creeps were slowed about the same, so it was harder to notice. With 10 attacking the same creep, he'd come to a stop and cause a major traffic jam.

[Karawasa]

So how much a particular creep stalls is dependent on how many towers are attacking it. I wonder what would be causing this. I would have rather expected all the creeps to become fucked up, like it is when there are too many (i.e. a drag on the cpu). But instead only the creep attacked gets affected...This may give us some insight into the problem, though there is a good chance it simply is not able to be optimized further.

[Cisz]

This allready happened in 3.0, and I have a feeling that this is about collision. Even though the "pushback" force of the used effect is set to zero, it still seems to do a tiny push, when the hits come to often. As if the game engine would go "ok, I push it, and after that I check if the push force is zero... ups, has been zero, to late now".

[Guest Sancdar]

Wouldn't it push them sideways instead of making them stand still then? And attacking from behind would make them move faster?

[Cisz]

I assume they get pushed with a vector of zero length.

[dvdlesher]

I think the stalls usually happen on ronald wave, anyone agree?

Also, think about this:

-20 or more arrow tower attacking the same creep, but can't even stall the creep

-20 or more elemental tower (quark, impulse) attacking the same creep can cause the creep to stall, why?

Btw, about my test, quark without waterfall buff still can cause creep stall, not very bad though, just a bit (Strangely enough, the stall happens on wave 61 only). The replay is...

-->My own forumdraft(#4 forumdraft) replay.

[Karawasa]

All of this is related to the DDS (Damage Detection System) that is the foundation for the abilities of a lot of combination towers. Too many calculations (i.e. attacks) alongside too much shit going on (Ronald, end game tower mass) leads to CPU drag (at least I think it is the CPU) and thus these undesirable effects.

Anyone around for the first private beta cycle for the PB may remember when creeps would stall with a lot of towers attacking, regardless of tower (since emjlr3's DDS registered any tower attack, while cohadar's is more efficient and also registers only specific ones).

[Cisz]

Kara - is there a dds in eletd 3.0?

If not, why can I stall with fast towers on that ver?

[Karawasa]

There is no such thing, and why you can is beyond me...perhaps CPU load like too many graphical projectiles and calculations to make per second (but from Blizzard's end not mine).

[jolin012]

but how can one's cpu affect the game and not just teh player? i mean we've seen this in multiplayer -what is actually happening then? the creeps stall for one player but not the otehrs?

or is teh host has a bad cpu it happens? was just thinking this might be a way to find a clue of why it happens, I mean i doubt it but i just thought it'd be worth saying anyway.

[Cisz]

I really think it's a tft bug.

[dvdlesher]

I really think it's a tft bug.

Have you ever confirmed this with other games on TFT?

[Cisz]

I don't play other games.

[dvdlesher]

Ah-ha, so you didn't play other games, eh?

Well, based on my experience with other games on TFT, I never seen "creep stall" actually happens except on EleTD

[Guest Sancdar]

I've only seen it as caused by too many moving units for a single player, not by too many attacks to one unit. Given that 3.0 glacier could cause it, I assume that the problem exists to some degree in TFT; I don't know how that situation relates to this one.