These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE General Discussion

Forum Index

EVE Forums » EVE Communication Center » EVE General Discussion » Increasing Server Tick Frequency to 2Hz ... Good, Bad, ...

Topic is locked indefinitely.

First pagePrevious page234

Increasing Server Tick Frequency to 2Hz ... Good, Bad, Ugly?
Author

Previous Topic Next Topic

seth Hendar

I love you miners

Likes received: 521

#61 - 2014-02-10 14:27:03 UTC | Edited by: seth Hendar

Jamwara DelCalicoe Ashley wrote:

I noticed that when my modules were preheated the 1tick-delay was not noticeable at all ... could this be because all of the input was processed at the same time rather than getting target lock and then applying module effect?

... I'm not really sure of the right way to word that ^ ...

just checked, and whther a point is preheated or not, it take at least 2 ticks to actually apply, 1 for lock if lock < 1s (in fact, minimum lock time is 1sec as per the srver tick), then next tick, point apply.

wich mean, anything warping under 2 sec is uncatchable, and under 3 sec almost incatchable at a gate (because of the reaction time when he decloack).

as now, any ceptor / shuttle / pod (and some fast frig) cannot be catched unless smartbomb / bomb / bubble if the first order given is a "warp to" order (and provided they are still cloacked when order is issued)

it is so bad that such fast entitys often show up on overview being 200+ km away @ 3500+ m/s (i.e. already in warp)

and yet i have a very good connection, with a 16ms +- ping, i can't imagine the nightmare for AUS or NZ ppl for example, when their ping adds to the reaction time and stuff

just for giggle, i tested, and a cynabal flown by a pilot with a full low grade nomad, shield dps fit witha t1 polycarbon rig and a nanofiber align in 1.97s..... so even some cruisers can achieve it Roll

Quote:

[Cynabal, test]

4x 425mm AutoCannon II (Barrage M)
Medium Energy Neutralizer II

2x Large Shield Extender II
Adaptive Invulnerability Field II
Warp Disruptor II
10MN Microwarpdrive II

Nanofiber Internal Structure II
2x Gyrostabilizer II
Tracking Enhancer II
Damage Control II

Medium Polycarbon Engine Housing I
2x Medium Core Defense Field Extender I

Jamwara DelCalicoe Ashley

New Eden Tech Support

Likes received: 428

#62 - 2014-02-10 18:55:27 UTC

seth Hendar wrote:

Jamwara DelCalicoe Ashley wrote:

Quote:

KAAAAAAHHNN!!!!!!!!

Thank me later :D

Paul Panala

School of Applied Knowledge

Caldari State

Likes received: 218

#63 - 2014-02-10 20:08:14 UTC

Mallak Azaria wrote:

Jamwara DelCalicoe Ashley wrote:

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

What is the barrier to increasing server response time to every half second instead of every one second?

Nullsec.

The servers already struggle over 1000 pilots with a 1Hz tick, exponentially increasing client polling simply won't help

That's a product of design. What network infrastructure would be needed to make it work?

Likely nothing that currently exists.

No! It is silly to think a game like Eve requires more power than anything else the world has ever seen. It is just a matter of having enough power. The question is weather it is cost effective for CCP to do so.

CCP should consider moving their data center to Amazon to take advantage of spot pricing on AWS servers. That way instead of buying a ton of servers that are overkill 90% of the time, you only pay for server time as needed. They could do some really cool stuff like automatically bring additional nodes online based on the number of players in an area.

I know that would be a huge undertaking, but if Amazon works for Netflix, I am sure it could work for CCP.

BrundleMeth

Black Parrot

Sylvanas Super mercenary

Likes received: 522

#64 - 2014-02-10 20:28:51 UTC

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

What is the barrier to increasing server response time to every half second instead of every one second?

Nullsec.

The servers already struggle over 1000 pilots with a 1Hz tick, exponentially increasing client polling simply won't help

That's a product of design. What network infrastructure would be needed to make it work?

Network performance isn't the issue. Raw number of calculations per second is. When you double the polling rate the absolute number of calculations increases to the square.

So say you have 400 pilots scapping, and say (totally made up number) that it requires 100,000 calculations a second. This is firing solution, ship vectors, bumping, refits, damage calculations, bonuses going up and down, session calculations, drawing models and what not

100,000 square is 10 billion, its not a linear progression.

Yes, but if you take the number of quamtum calculations and multiply it by a finite number of idioms, then the resulting equation is equal to the number of server ticks multiplied by the quantum capacitors ability to keep pace with the full cycles of the square root of the hypotinuse. And all this depends on the processing power of the current hardware minus the overhead of the cycles lost in heat generating calculations of the parallel equations...

Jamwara DelCalicoe Ashley

New Eden Tech Support

Likes received: 428

#65 - 2014-02-11 11:45:29 UTC

BrundleMeth wrote:

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

What is the barrier to increasing server response time to every half second instead of every one second?

Nullsec.

The servers already struggle over 1000 pilots with a 1Hz tick, exponentially increasing client polling simply won't help

That's a product of design. What network infrastructure would be needed to make it work?

I wish that I could agree with you ... but I have no idea what you just said.

Thank me later :D


seth Hendar I love you miners Likes received: 521	#66 - 2014-02-11 13:53:25 UTC 1 Jamwara DelCalicoe Ashley wrote: KAAAAAAHHNN!!!!!!!! wut?

Jamwara DelCalicoe Ashley

New Eden Tech Support

Likes received: 428

#67 - 2014-02-11 14:04:17 UTC | Edited by: Jamwara DelCalicoe Ashley

seth Hendar wrote:

Jamwara DelCalicoe Ashley wrote:

KAAAAAAHHNN!!!!!!!!

wut?

That was a Star Trek reference to the adversarial relationship between JT Kirk and Kahn. The underlying implication is that, like Kirk, I have too many feelz and need to vent my frustration with not having found a way around 2-tick clicking associated with a 1Hz cycle.

*also, thank you for testing my hypothesis!

Thank me later :D

seth Hendar

I love you miners

Likes received: 521

#68 - 2014-02-11 14:16:22 UTC | Edited by: seth Hendar

Jamwara DelCalicoe Ashley wrote:

seth Hendar wrote:

Jamwara DelCalicoe Ashley wrote:

KAAAAAAHHNN!!!!!!!!

wut?

ha ok, got it (, too late tho, thx coffee yeah \o/ )

np, i'm struggling with this frustration too for, like i said, more than a year, hence why i try anything that could solve the issue, unfortunately, only CCP can do, and they are not motivated about it it seems......

also, while increasing server tick would solve the problem, there is another way to solve THIS lock problem (won't solve the uncloacking issue encountered by 4000+ m/s fig passing throught tho).

back a year and half / two years ago, it was possible to catch such fast things on gates with fast (very fast in fact) locks.

reason is simple, some delay where introduced since.

3 majorchanges i have identified: before, a preacivated mod was applyed on SAME tick lock was achieved, meaning you could lock and point in 1 tick.
also a "delay" was introduced in overview refresh, it is way more laggy now(meaning almost any time, when the tgt appears on overview, it his decloacked since certain time, wich is a big % of align for fastys).

and alignement time seems decreased, at least for pods and shuttle.

factor in all 3, and bazinga the window in witch they are vulnerable is, in practice, non existent anymore (in fact, way to short for any ship to exploit it due to actual game mechanics)

ps: ARGHH forum ate my post AGAIN Evil

Plastic Psycho

Necro-Economics

Likes received: 1,345

#69 - 2014-02-11 17:49:10 UTC

Jamwara DelCalicoe Ashley wrote:

Mallak Azaria wrote:

Jamwara DelCalicoe Ashley wrote:

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

What is the barrier to increasing server response time to every half second instead of every one second?

Nullsec.

The servers already struggle over 1000 pilots with a 1Hz tick, exponentially increasing client polling simply won't help

That's a product of design. What network infrastructure would be needed to make it work?

Likely nothing that currently exists.

You really think so?

Meet TQ.
TQ is a Beast. To double the polling rate, you'd need *four* such beasts. How much are you willing to pay per sub to support such a monster?

Frostys Virpio

KarmaFleet

Goonswarm Federation

Likes received: 3,464

#70 - 2014-02-11 18:01:08 UTC

Plastic Psycho wrote:

Meet TQ.
TQ is a Beast. To double the polling rate, you'd need *four* such beasts. How much are you willing to pay per sub to support such a monster?

More like "How much are you willing to pay your subs to finance the R&D required to design/manufacture CPU going at 4 time the current wall." since we currently can't offload the process to other cores so adding a load balancing infrastructure is bound to be just as fruitless.

CCP was awoxed a long time ago when Intel and AMD quit the GHz race.

Plastic Psycho

Necro-Economics

Likes received: 1,345

#71 - 2014-02-11 18:27:36 UTC

Frostys Virpio wrote:

Plastic Psycho wrote:

Meet TQ.
TQ is a Beast. To double the polling rate, you'd need *four* such beasts. How much are you willing to pay per sub to support such a monster?

Fair point.

seth Hendar

I love you miners

Likes received: 521

#72 - 2014-02-13 14:05:11 UTC | Edited by: seth Hendar

Plastic Psycho wrote:

Meet TQ.
TQ is a Beast. To double the polling rate, you'd need *four* such beasts. How much are you willing to pay per sub to support such a monster?

while i agree the cluster is a really powerfull beast, and it would cost a freakin lot to upgrade it, remember that it is under used badly because of the base code not being able to use several core or span throught several blades.

this is imao where the work need to be done, granted it might not be easy, but it is required if eve wants to keep evolving.

the current situation is really not good, with tidi kicking all over the place, if the code would allow dynamic server ressources allocation, eve online would be really badass (yeah, even more badass, can you picture that?).

and then, massive battles like asakai and alike would probably not even require TIDI, out of the few seconds required for the load to be spreaded across the hardware.....

of course, under such conditions, 2hz server tick would be easier to implement (well, it is easy, it's just that it's not a good idea given the curent situation)

/me dreams of a cuda / opencl server side code for eve online THAT would be.....incredibly FAST


Plastic Psycho Necro-Economics Likes received: 1,345	#73 - 2014-02-13 16:59:05 UTC Code first, then we'll talk about changing the polling rate. Cleaning up the code is, in itself, a massive challenge that costs a lot of money... They do it as they can.

Rainbow Dash

The Scope

Gallente Federation

Likes received: 156

#74 - 2014-02-14 22:20:34 UTC

Paul Panala wrote:

Mallak Azaria wrote:

Jamwara DelCalicoe Ashley wrote:

Onictus wrote:

Jamwara DelCalicoe Ashley wrote:

What is the barrier to increasing server response time to every half second instead of every one second?

Nullsec.

The servers already struggle over 1000 pilots with a 1Hz tick, exponentially increasing client polling simply won't help

That's a product of design. What network infrastructure would be needed to make it work?

Likely nothing that currently exists.

Now we get to play a fun game: is this guy joking or an idiot?

First pagePrevious page234