These forums have been archived and are now read-only.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

 
  • Topic is locked indefinitely.
 

Lag, TiDi, 6-VT and you...

First post
Author
Onictus
Federal Navy Academy
Gallente Federation
#181 - 2013-07-30 21:28:00 UTC
Maximillian German wrote:
Onictus wrote:
Maximillian German wrote:
[
which is of course countered in eve by slowing the client update rate as not to overload the server. Thus, you have TiDi.


Shocked


Did I get something wrong? Or is eve slowing down the request rate?



No you nailed it.

The server slows down the client updates so that it has more time to process the queue.
Murk Paradox
Ministry of War
Amarr Empire
#182 - 2013-07-30 21:29:46 UTC
Onictus wrote:
Murk Paradox wrote:


Eh? So your solution is to simply double talk the fact that time dilation is lag created by ccp simply because there is no other way of doing it?




Its finally sinking in.

Its not hardware, you aren't getting anything better, the Eve cluster architecture is a super computer.

So that leaves software/middleware optimization, and like I said real time dynamic loadsharing isn't a trivial undertaking, and anything done to the engine itself is a monumental undertaking.

Murk Paradox wrote:



Just you saying "no" doesn't dispute the fact.

You asked for it, and got it. The burden has been met.

.


When you have your facts right you can tell me that anything has been met.


I don't know of any other way to prove my work unless you want a bunch of links.

Not proving me wrong is by default making me right. This thread has already shown my work in above mentioned links.

Beyond that, time dilation is still lag.

This post has been signed by Murk Paradox and no other accounts, alternate or otherwise. Any other post claiming to be this holder's is subject to being banned at the discretion of the GM Team as it would violate the TOS in regards to impersonation. Signed, Murk Paradox. In triplicate.

Murk Paradox
Ministry of War
Amarr Empire
#183 - 2013-07-30 21:33:46 UTC
Onictus wrote:
Murk Paradox wrote:

But that is not in any sort of direct conflict with time dilation being lag. So again, I'm trying to figure out if you're agreeing with me or arguing, because we are on the same page.


So I see we are back to any form of performance change is lag.

Latency is not lag.
TiDi is not lag.

Lag is what happens when the the interrupt queue exceeds the client update rate.



Uh, latency is a measurement of time to define speed. It is the measurement of lag. Time dilation increases lag to increase stability and prevent a disconnect.

If you want to say TiDi does not increase stability that's on you, but it sure as **** increases lag, that's the point. It slows time down. That's what lag is since the client (see you, me players) are not pressing buttons in slow motion.

This post has been signed by Murk Paradox and no other accounts, alternate or otherwise. Any other post claiming to be this holder's is subject to being banned at the discretion of the GM Team as it would violate the TOS in regards to impersonation. Signed, Murk Paradox. In triplicate.

Tippia
Sunshine and Lollipops
#184 - 2013-07-30 21:34:33 UTC  |  Edited by: Tippia
Murk Paradox wrote:
I don't know of any other way to prove my work unless you want a bunch of links.
So link it.

Quote:
Not proving me wrong is by default making me right.
No. You are wrong by default until you've proven yourself right. So far, you have not. Not proving you wrong still makes you wrong, with the only possible side-effect that the other side might be wrong as well.

Quote:
Uh, latency is a measurement of time to define speed.
No. Latency is a measurement of time to define routing and processing delays. TiDi increases the processing delay on the server side and tells the client to wait a bit longer for the result. Whether or not this creates lag depends on how much the client spams the server with new junk to process. TiDi ensures that the client doesn't do that.
Murk Paradox
Ministry of War
Amarr Empire
#185 - 2013-07-30 21:39:57 UTC
Maximillian German wrote:
On topic: Murk, can you take a break from your lag discussion and answer the question of the OP? In case you have forgotten, here it is:

How as paying EvE users did we come to accept UI razzle dazzle and gimmicks instead of a true remedy to slowdowns?


I don't understand what you mean by razzle dazzle in lieu of slowdowns.

This post has been signed by Murk Paradox and no other accounts, alternate or otherwise. Any other post claiming to be this holder's is subject to being banned at the discretion of the GM Team as it would violate the TOS in regards to impersonation. Signed, Murk Paradox. In triplicate.

Murk Paradox
Ministry of War
Amarr Empire
#186 - 2013-07-30 21:47:56 UTC  |  Edited by: Murk Paradox
Tippia wrote:
Murk Paradox wrote:
I don't know of any other way to prove my work unless you want a bunch of links.
So link it.

Quote:
Not proving me wrong is by default making me right.
No. You are wrong by default until you've proven yourself right. So far, you have not. Not proving you wrong still makes you wrong, with the only possible side-effect that the other side might be wrong as well.

Quote:
Uh, latency is a measurement of time to define speed.
No. Latency is a measurement of time to define routing and processing delays. TiDi increases the processing delay on the server side and tells the client to wait a bit longer for the result. Whether or not this creates lag depends on how much the client spams the server with new junk to process. TiDi ensures that the client doesn't do that.



Routing and processing delays impact latency, not define it. Well, the routing does. But latency is a measurement of time ie- speed.

And I do not have to prove myself right, again, as the threads have the links already. Find them yourself if you do not believe me.

Time dilation is still lag regardless. So any increase in time due to tidi is an increase of lag, not a decrease, as you have tried to say as you think tidi "removes" lag. It removes instability if it removes anything.

It increases lag.

This post has been signed by Murk Paradox and no other accounts, alternate or otherwise. Any other post claiming to be this holder's is subject to being banned at the discretion of the GM Team as it would violate the TOS in regards to impersonation. Signed, Murk Paradox. In triplicate.

Maximillian German
Task Force Coalition
#187 - 2013-07-30 21:50:43 UTC
Murk Paradox wrote:
Maximillian German wrote:
On topic: Murk, can you take a break from your lag discussion and answer the question of the OP? In case you have forgotten, here it is:

How as paying EvE users did we come to accept UI razzle dazzle and gimmicks instead of a true remedy to slowdowns?


I don't understand what you mean by razzle dazzle in lieu of slowdowns.


According to the OP, CCP is giving us UI updates instead of/to distract us from slowdown problems. Can't make it much clearer than that.
Tippia
Sunshine and Lollipops
#188 - 2013-07-30 21:55:00 UTC
Murk Paradox wrote:
Routing and processing delays impact latency, not define it.
Routing and processing delays are latency. And no, measurements of time are measurements of time, not speed. To have a speed, you need to measure something over time.

If you absolutely want to call latency a measurement of something other than time, it is of the somewhat abstract concept of (routing) distance.

Quote:
And I do not have to prove myself right
…if you want to remain wrong. if you want to be right, though, you have to prove that you are. If the thread contains the links that does this, you can do it trivially.

Quote:
Time dilation is still lag regardless.
…except that all it does is increases the processing delay on the server side and tells the client to wait a bit longer for the result. Whether or not this creates lag depends on how much the client spams the server with new junk to process. TiDi ensures that the client doesn't do that, which removes one of the principal causes of lag.
Maximillian German
Task Force Coalition
#189 - 2013-07-30 22:00:12 UTC
Murk Paradox wrote:
Tippia wrote:
Murk Paradox wrote:
I don't know of any other way to prove my work unless you want a bunch of links.
So link it.

Quote:
Not proving me wrong is by default making me right.
No. You are wrong by default until you've proven yourself right. So far, you have not. Not proving you wrong still makes you wrong, with the only possible side-effect that the other side might be wrong as well.

Quote:
Uh, latency is a measurement of time to define speed.
No. Latency is a measurement of time to define routing and processing delays. TiDi increases the processing delay on the server side and tells the client to wait a bit longer for the result. Whether or not this creates lag depends on how much the client spams the server with new junk to process. TiDi ensures that the client doesn't do that.



Routing and processing delays impact latency, not define it. Well, the routing does. But latency is a measurement of time ie- speed.

And I do not have to prove myself right, again, as the threads have the links already. Find them yourself if you do not believe me.

Time dilation is still lag regardless. So any increase in time due to tidi is an increase of lag, not a decrease, as you have tried to say as you think tidi "removes" lag. It removes instability if it removes anything.

It increases lag.


no, you are not right just because you say so.

Anyway, a quick point of clarification. Are you saying that you would rather have the cluster the way it was before TiDi was implemented? Or will you concede that TiDi is an improvement over the previous state of the cluster?
Vincent Athena
Photosynth
#190 - 2013-07-30 22:20:03 UTC
There is a definable difference between lag and TiDi.

Eve runs on one second ticks. If all is well you can expect that when you turn on a gun it will fire in one or maybe two ticks. When TiDi starts cutting in those ticks are no longer 1 second. At 50% TiDi they are two seconds long, and you can expect your gun to fire 2 to 4 seconds after turning it on.

At 10% TiDi those ticks are 10 seconds long. You would expect your guns to fire in 10 to 20 seconds after activation. If they take longer, say 5 minutes, then we have entered the realm of lag. TiDi allows more players to fight without lag. If it were not for TiDi issues with guns not turning on or off would occur with 1000 player fights.

Now here is the big important difference: With TiDi, but no lag, everyone will see the same response to module activation: One to two ticks of however long. With lag the delay between activating a gun and having it fire becomes much longer and more random.

TiDi slows the game for everyone equally and the game still works in a predictable manner. Lag slows it in different and random amounts for different people and the game works in an unpredictable manner.

Apparently in the 6V battle there was both TiDi and lag. it would be nice to not have either, but CCP has yet to figure out how to do that.

Know a Frozen fan? Check this out

Frozen fanfiction

Ka'Narlist
Dreddit
Test Alliance Please Ignore
#191 - 2013-07-30 22:33:42 UTC
Stopped reading after the first couple of posts but want to add my opinion anyways.


Sure TiDi is much better as random lags where the people first on grid or in system can shoot the ones still loading and its what makes such huge fights possible.

But it can't be the solution, only a step to the solution. Because fighting in max. TiDi sucks, it really sucks like hell and it drags fights which would be resolved under normal circumstances in half an hour over hours and hours where you have to sit there waiting that your mod cycled down to click it and wait for another 10 or 15 minutes to click it again.

Yes hardware has its limits but if one server node can't handle such a huge fight than two should do it or a whole group of them. If Eve currently can't let the ongoings in a system or on a grid get handled by a servercluster this should be the direction to develop it further. I know it will not be easy or quick or cheap but imho that is the direction ccp has to take if they really want to brag about being able to handle huge space fights.

I really hope ccp doesn't consider TiDi as their end solution, but is constantly looking for better solutions.
Dersen Lowery
The Scope
#192 - 2013-07-30 23:47:26 UTC
Ka'Narlist wrote:
Yes hardware has its limits but if one server node can't handle such a huge fight than two should do it or a whole group of them. If Eve currently can't let the ongoings in a system or on a grid get handled by a servercluster this should be the direction to develop it further. I know it will not be easy or quick or cheap but imho that is the direction ccp has to take if they really want to brag about being able to handle huge space fights.

I really hope ccp doesn't consider TiDi as their end solution, but is constantly looking for better solutions.


Good news: They are, and they have been since Incursion or thereabouts. Unfortunately, you're right that it will not be easy or quick or cheap to get multiple nodes to do the work that is currently done by one, but they're working on it as we speak.

Proud founder and member of the Belligerent Desirables.

I voted in CSM X!

Victoria Sin
Doomheim
#193 - 2013-07-31 00:04:17 UTC
Dersen Lowery wrote:

Good news: They are, and they have been since Incursion or thereabouts. Unfortunately, you're right that it will not be easy or quick or cheap to get multiple nodes to do the work that is currently done by one, but they're working on it as we speak.


Not sure how they're going to do that. Spreading a load like a fleet fight across threads would be non-trivial because it's not obvious how to partition the set. If you're sharing memory you have to also be careful you aren't stalling cache lines too, otherwise your 8 cores will only be slightly faster than 1. Not sure if Stackless Python allows you to take care of details like this.
Ranger 1
Ranger Corp
Vae. Victis.
#194 - 2013-07-31 00:09:59 UTC
Victoria Sin wrote:
Dersen Lowery wrote:

Good news: They are, and they have been since Incursion or thereabouts. Unfortunately, you're right that it will not be easy or quick or cheap to get multiple nodes to do the work that is currently done by one, but they're working on it as we speak.


Not sure how they're going to do that. Spreading a load like a fleet fight across threads would be non-trivial because it's not obvious how to partition the set. If you're sharing memory you have to also be careful you aren't stalling cache lines too, otherwise your 8 cores will only be slightly faster than 1. Not sure if Stackless Python allows you to take care of details like this.

Yep, that's the reason why a better solution hasn't already been implemented. However, that hasn't stopped them from working on it.

For those who continue to dismiss Tidi as unhelpful, perhaps you should try a few other games that allow a 4000 person battle and give us a report on how it went. Blink

View the latest EVE Online developments and other game related news and gameplay by visiting Ranger 1 Presents: Virtual Realms.

Invisusira
Escalated.
OnlyFleets.
#195 - 2013-07-31 00:15:01 UTC
...sorry, did you have some sort of magical wizard server cluster capable of handling 4000+ simotaneous mmo client server connections? Or were you just bitching because you keep losing SBs?
Tippia
Sunshine and Lollipops
#196 - 2013-07-31 10:59:23 UTC
Awww… how cute!
The OP is so upset about being wrong that he has to dec the guilty! Lol
Tialano Utrigas
Running with Dogs
Northern Coalition.
#197 - 2013-07-31 11:17:49 UTC
To those of you who say they didn't mind fighting with 4000 guys under 10% Tidi:

YOU WERE CLEARLY NOT F*CKING THERE!

10% is the lowest the indicator goes but I can assure you it was waaaay worse than 10%. If it ahd stayed at 10% it would have been bearable. It was no where near it.

Weapons not firing
Modules not reloading
Modules not deactivating
3 - 4 minutes for a frigate to lock a Dread!
40 minutes to jump in system

So please can everyone stop talking sh*t!

The fight was enjoyable for its own reasons, primarily the "I was there" factor. The lag, however, was not.

I DO however appreciate the lengths that CCP had to go to make this fight actually happen and most of us appreciate that unless we asked NASA to borrow some equipment for a day, that this was to be expected.
Frostys Virpio
State War Academy
Caldari State
#198 - 2013-07-31 12:11:38 UTC
Tialano Utrigas wrote:
To those of you who say they didn't mind fighting with 4000 guys under 10% Tidi:

YOU WERE CLEARLY NOT F*CKING THERE!

10% is the lowest the indicator goes but I can assure you it was waaaay worse than 10%. If it ahd stayed at 10% it would have been bearable. It was no where near it.

Weapons not firing
Modules not reloading
Modules not deactivating
3 - 4 minutes for a frigate to lock a Dread!
40 minutes to jump in system

So please can everyone stop talking sh*t!

The fight was enjoyable for its own reasons, primarily the "I was there" factor. The lag, however, was not.

I DO however appreciate the lengths that CCP had to go to make this fight actually happen and most of us appreciate that unless we asked NASA to borrow some equipment for a day, that this was to be expected.


NASA's computer are probably threaded behemot by now so it would not help all that much. The process that require a slowdown of the game to remain responding to command, if at a really slower rate, is not made to be divided on more processing unit. We are playing a 10 yo game. Removing some of the limitation linked to this will not happen by snapping our fingers.
Jenn aSide
Soul Machines
The Initiative.
#199 - 2013-07-31 12:48:29 UTC
Feyd Rautha Harkonnen wrote:
I

Can someone explain this please? It wasn't so long ago people were burning Jita because of micro-transactions, yet now with increasing server suckage the response is 'meh'?


Your problem is twofold.

First: You seem to not understand how completely terrible fleet fights were pre-TiDi. Those of us who experienced that ("I was there" for the Fall of C-J6 lol) understand that TiDi is a literal walk in the park compared to what is was. There is no "increasing server suckage", it's better now than ever AS EVIDENCED BY THE RECENT 4000 SHIP FIGHT, the fact is that your perceptions are what's flawed here.

Second: We are realistic, we know CCP can't just wave a wand and make latency go away. We understand that CCP is (in allowing us such fights) doing what no one else can come close to and that it's going to be imperfect. And we don't believe in being utter ingrates by saying to them "you guys suck because this amazing thing you are giving us isn't as amazing as we want it!".

So yea, i get what you're saying we should all riot in Jita against the game company giving us the most incredible experience in modern gaming. You sir are Brilliant.
Malcanis
Vanishing Point.
The Initiative.
#200 - 2013-07-31 13:01:29 UTC
Murk Paradox wrote:

Time dilation is still lag regardless. So any increase in time due to tidi is an increase of lag, not a decrease, as you have tried to say as you think tidi "removes" lag. It removes instability if it removes anything.

It increases lag.


TiDi does not increase lag. That's like saying traffic lights cause traffic jams.

"Just remember later that I warned against any change to jump ranges or fatigue. You earned whats coming."

Grath Telkin, 11.10.2016