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

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.



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

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.

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.

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?

So link it.

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.

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.

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

Routing and processing delays impact latency, not define it.

And I do not have to prove myself right

Time dilation is still lag regardless.

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.

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.

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.

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.

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'?

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.

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.

The EVE engine has two steps: 1) calculate everything; 2) update everyone involved with the results of the calculations. The hope is that 1 and 2 can run on different cores, with 1 calling 2 asynchronously, because step 2 can take a surprisingly long time.

But there was not just single Tidi (slow motion of time) there was lag too.
I was there. My guns needed 15 minutes activation times in 10% Tidi. 10% tidi = 10x slowing, and a matematics know 1second x 10=10second and 10seconds is not equal with 15 minutes.
The gun cycles stucked or didn't work. The damage stoppped many times when we got lagspikes.
I made 10 kills within 3h45min in a 4k battles. (Pathetic thing)
The battle was almost unplayable for 4 hours long.

Oh and i have a opinion from TiDi. Tidi = bullsh*t
Paid for lag.

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.

That doesn't have any part of what I've been saying. So whereas you might be clear in your communications to the OP, I am not him.

Therefore I cannot answer your questions for him, because well, I don't share that view.

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?

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.