These forums have been archived and are now read-only.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

Fix the TiDi please

Author
TWHC Assistant
#161 - 2012-05-13 10:45:31 UTC
Lord Zim wrote:
Wait, what? My client has to wait for a response because of TiDi? Since when?

No, you have to wait. TiDi slows down time.

Lord Zim wrote:
That has nothing to do with lag. Slowing down the server-side simulation reduces lag.

This is what it has to do with lag.
Lord Zim
Gallente Federation
#162 - 2012-05-13 10:52:50 UTC
TWHC Assistant wrote:
Lord Zim wrote:
Wait, what? My client has to wait for a response because of TiDi? Since when?

No, you have to wait. TiDi slows down time.

So what you're saying is, a warp taking longer is "lag"?

TWHC Assistant wrote:
Lord Zim wrote:
That has nothing to do with lag. Slowing down the server-side simulation reduces lag.

This is what it has to do with lag.

So tidi is bad because it reduces lag.

Huh.

Cyno's lit, bridge is up, but one pilot won't be jumping home.

RIP Vile Rat

Francisco Bizzaro
#163 - 2012-05-13 10:52:54 UTC
TWHC Assistant wrote:
Lord Zim wrote:
Wait, what? My client has to wait for a response because of TiDi? Since when?

No, you have to wait. TiDi slows down time.

Lord Zim wrote:
That has nothing to do with lag. Slowing down the server-side simulation reduces lag.

This is what it has to do with lag.

It's not lag. It's a re-definition of the Eve clock tick.

Imagine that when they designed Eve, they had decided the Eve clock should tick 10 times slower than it currently does. Everything would happen 10 times as slowly, but you wouldn't call it lag. It's just that the servers would be working at a very relaxed pace and under-utilized most of the time. With TiDi, they've added the ability to re-define time dynamically.
TWHC Assistant
#164 - 2012-05-13 10:55:52 UTC  |  Edited by: TWHC Assistant
Onictus wrote:
Its not a warning, its a response.

Most warnings are a response.

You may wonder what a road sign for a train crossing is a response to. It is a response to people not wanting to get others killed by trains and so they put up a sign to warn you of crossing trains. It does not mean that everyone will get killed who crosses the track. You may see a no-parking sign as a punishment, but it really is a warning for you not to park there. You can try to park there for a short time and may not get a fine. Same with TiDi. It warns you about the node getting close to its limit. You are not really in control of who else enters a system and so may decide to leave it.

It is only smart to take it as a warning. You may not be part of a fight in which case you would only do others, who are fighting, a favour by leaving the system.
Ptraci
3 R Corporation
#165 - 2012-05-13 10:58:36 UTC
Gorenaire wrote:
was travelling from Caldari space through Gallente and TiDi just makes the game absolutely unplayable , its takes 5 minutes to jump from 1 system to another

TiDi @ 10 for at least a dozen system i went through, it takes ages to move anywhere

To fix battle lag in 0.0, people nowhere near those systems have to lag to hell as well ?? thats the point of this solution ?


There is a fix of course it involves putting every single system on its own blade server, however I don't think you will enjoy the $500/month subscription fee.
Lord Zim
Gallente Federation
#166 - 2012-05-13 10:58:49 UTC  |  Edited by: Lord Zim
TWHC Assistant wrote:
Onictus wrote:
Its not a warning, its a response.

Most warnings are a response.

You may wonder what a road sign for a train crossing is a response to. It is a response to people not wanting to get others killed by trains and so they put up a sign to warn you of crossing trains. It does not mean that everyone will get killed who crosses the track. You may see a no-parking sign as a punishment, but it really is a warning for you not to park there. You can try to park there for a short time and may not get a fine. Same with TiDi. It warns you about the node getting close to its limit. You are not really in control of who else enters a system and so may decide to leave it.

It is only smart to take it as a warning. You may not be part of a fight in which case you would only do others, who are fighting, a favour by leaving the system.

So the gates to a system with TiDi enabled tells me that I'm about to jump into a system which is on the same node as a system where there's a big fight going on?

Cyno's lit, bridge is up, but one pilot won't be jumping home.

RIP Vile Rat

Onictus
Federal Navy Academy
Gallente Federation
#167 - 2012-05-13 11:06:59 UTC
Lord Zim wrote:
TWHC Assistant wrote:
Onictus wrote:
Its not a warning, its a response.

Most warnings are a response.

You may wonder what a road sign for a train crossing is a response to. It is a response to people not wanting to get others killed by trains and so they put up a sign to warn you of crossing trains. It does not mean that everyone will get killed who crosses the track. You may see a no-parking sign as a punishment, but it really is a warning for you not to park there. You can try to park there for a short time and may not get a fine. Same with TiDi. It warns you about the node getting close to its limit. You are not really in control of who else enters a system and so may decide to leave it.

It is only smart to take it as a warning. You may not be part of a fight in which case you would only do others, who are fighting, a favour by leaving the system.

So the gates to a system with TiDi enabled tells me that I'm about to jump into a system which is on the same node as a system where there's a big fight going on?



No it means nothing. The system may be 10% TiDi and the fight is on the other side of the map.

Last time I caught a random TiDi I was in Khanid or something and it was the Goons and Raiden fighting in Tenal. The fact that there is TiDI just means there is a fight happening, the warning is a local count at like 300. If there is a fight next door the system you are sitting in isn't necessarily affected. The systems on a node are not arranged to be connected.
TWHC Assistant
#168 - 2012-05-13 11:08:29 UTC
Lord Zim wrote:
So what you're saying is, a warp taking longer is "lag"?

Sort of. It is the same, but it is not identical. To get from A to B during real lag with everything taking 30 minutes to respond is the same as looking at the warp animation during TiDi for 30 minutes. It still takes 30 minutes to get from A to B.

Instead of having a real lag, caused by the node's inability to process all requests are the clients being slowed down in order to send less requests. The difference between a real lag and an artificial lag is simply a technical one where the real lag is caused by hardware limits and the artificial lag is caused by a limiting software.

TWHC Assistant wrote:
So tidi is bad because it reduces lag.

No. It is better than a disconnect, but it is only a work-around and when players do not take it as a warning is it of little use. Better would be to have faster hardware.
Francisco Bizzaro
#169 - 2012-05-13 11:16:01 UTC  |  Edited by: Francisco Bizzaro
TWHC Assistant wrote:
Lord Zim wrote:
So what you're saying is, a warp taking longer is "lag"?

Sort of. It is the same, but it is not identical. To get from A to B during real lag with everything taking 30 minutes to respond is the same as looking at the warp animation during TiDi for 30 minutes. It still takes 30 minutes to get from A to B.

Instead of having a real lag, caused by the node's inability to process all requests are the clients being slowed down in order to send less requests. The difference between a real lag and an artificial lag is simply a technical one where the real lag is caused by hardware limits and the artificial lag is caused by a limiting software.

Lag only has a meaning in reference to the software, because the software defines the clock tick which defines when events are judged to be "simultaneous". As long as the processor can do what it needs to do within that timeframe, there is no lag.

If it can't, then the operation queue *lags* behind the clock.

If, however, you fix the clock so that it always stays ahead of the operation queue, you have eliminated lag - the operations can be processed "simultaneously" within the clock resolution of the simulation.

Lag has nothing to do with slowing down. It has everything to do with the processor falling behind the clock so that the idea of simultaneity gets lost and things get out of whack in an unpredictable way.
Onictus
Federal Navy Academy
Gallente Federation
#170 - 2012-05-13 11:16:31 UTC
TWHC Assistant wrote:


TWHC Assistant wrote:
So tidi is bad because it reduces lag.

No. It is better than a disconnect, but it is only a work-around and when players do not take it as a warning is it of little use. Better would be to have faster hardware.


There IS no better hardware, I work in the industry, and I will tell you right now, that HS22 blades are beasts, their only weak point is the northbridge which causes a bottleneck getting the 100GE interfaces ...that is literally it. These things are dual 8 core xenons with like 12gigs of ram per proc. These things cost 8 grand a pop JUST for the blade and 14 go in a chassis.

Each chassis racks in at about $20,000 without power supplies or network switches. The onboard switches run 12 large and each chassis can handle 8 of various types, from simple ethernet to 100GigE.

The "super node" is THE FASTEST commercially available enterprise level server on the market.

The only way to cluster performance would be single server every system, which besides being RIDICULOUSLY expensive and I mean US military ridiculously expensive, the power bill alone for running that many servers would likely eclipse what the company makes a month.
Lord Zim
Gallente Federation
#171 - 2012-05-13 11:17:43 UTC
TWHC Assistant wrote:
Lord Zim wrote:
So what you're saying is, a warp taking longer is "lag"?

Sort of. It is the same, but it is not identical. To get from A to B during real lag with everything taking 30 minutes to respond is the same as looking at the warp animation during TiDi for 30 minutes. It still takes 30 minutes to get from A to B.

Yeah, that's not lag, that's just slowing down time, much like changing the speedlimit of a highway to actually increase flow/throughput isn't lag.

TWHC Assistant wrote:
Instead of having a real lag, caused by the node's inability to process all requests are the clients being slowed down in order to send less requests. The difference between a real lag and an artificial lag is simply a technical one where the real lag is caused by hardware limits and the artificial lag is caused by a limiting software.

The clients are able to send just as many requests as they'd like, the difference is that they're being processed in a more timely manner instead of being queued up, so the server can tell the client "yes, you've told me that already, stop nagging".

TWHC Assistant wrote:
No. It is better than a disconnect, but it is only a work-around and when players do not take it as a warning is it of little use. Better would be to have faster hardware.

Stop harping on about it being a warning system. It isn't a warning system, I do not get a warning before I jump in to a system that the node the system on is heavily loaded, so it's neither a warning system, nor is it particularly early.

You harping on about "well it would be better if the players took it as a warning to commit fewer people so the experience would be better" shows that you're hilariously out of touch with what EVE is all about. What's next, are you going to demand that CCP add in code which kicks in whenever TiDi kicks in, and randomly moves people from both (or all 3, or 4, or 5) sides of a fight until we get down to a sufficient number of people so as to not cause TiDi, because there's a chance that some poor, poor hisec whiner might have to spend a bit more time warping from point to point?

And the bit about "just throw faster hardware at it" comment shows that you have literally no clue how to actually do scalable software.

Cyno's lit, bridge is up, but one pilot won't be jumping home.

RIP Vile Rat

kakmonstret
Domain Mining and Trading Corp
#172 - 2012-05-13 11:21:49 UTC  |  Edited by: kakmonstret
Uhhmm fun thread....

I might have misunderstood this completely but isn't the idea behind TiDi to no slow down request resolving but slowdown the simulation.

So in essence the server uses more time to respond to request instead of recalculating the location of all ships and such things?

So all your requests are handled in time as they should. To do this it uses less time to update the simulation calculating positions, movements and events(such as missile explosions).

To do this the server simple does the simulation calculation less frequent. At normal time the server runs the simulation calculation 1 time every second I think. With TiDi it does this less and less often up to some point.

The result is that all the things you see goes slower but you requests are handled in time.

I think I said the same as most other people but maybe with a few different words, hope it helps someone....
Onictus
Federal Navy Academy
Gallente Federation
#173 - 2012-05-13 11:25:31 UTC
kakmonstret wrote:
Uhhmm fun thread....

I might have misunderstood this completely but isn't the idea behind TiDi to no slow down request resolving but slowdown the simulation.

So in essence the server uses more time to respond to request instead of recalculating the location of all ships and such things?

So all your requests are handled in time as they should. To do this it uses less time to update the simulation calculating positions, movements and events(such as missile explosions).

To do this the server simple does the simulation calculation less frequent. At normal time the server runs the simulation calculation 1 time every second I think. With TiDi it does this less and less often up to some point.

The result is that all the things you see goes slower but you requests are handled in time.

I think I said the same as most other people but maybe with a few different words, hope ti helps someone....



More or less.

At full speed the server updates the clients once a second. The procs themselves are aren't changing speed, they are running WFO, what IS changing is how fast they are updating the clients. So the Client goes slower, but everything actually happens, opposed to things being bumped out of queued processes and not happening at all or exhibiting random undesirable behaviors.
TWHC Assistant
#174 - 2012-05-13 11:26:05 UTC
Lord Zim wrote:
So the gates to a system with TiDi enabled tells me that I'm about to jump into a system which is on the same node as a system where there's a big fight going on?

Possibly, but I do not now all the details and have to deduct from what I see. What I see is that not all of EVE gets to see the message and so it has to be restricted to some systems. OP has seen TiDi on a travel route, which at the time had 50-80 players in them. The map did not show a high count of destroyed ships for these systems but in two nearby systems.
Tippia
Sunshine and Lollipops
#175 - 2012-05-13 11:35:38 UTC
TWHC Assistant wrote:
OP has seen TiDi on a travel route, which at the time had 50-80 players in them. The map did not show a high count of destroyed ships for these systems but in two nearby systems.
…because they were on the same node and thus subject to the same TiDi.
Lord Zim
Gallente Federation
#176 - 2012-05-13 11:38:57 UTC  |  Edited by: Lord Zim
TWHC Assistant wrote:
Lord Zim wrote:
So the gates to a system with TiDi enabled tells me that I'm about to jump into a system which is on the same node as a system where there's a big fight going on?

Possibly, but I do not now all the details and have to deduct from what I see. What I see is that not all of EVE gets to see the message and so it has to be restricted to some systems. OP has seen TiDi on a travel route, which at the time had 50-80 players in them. The map did not show a high count of destroyed ships for these systems but in two nearby systems.

Exactly, you don't know most (or maybe even any) of the details, and your deductions are incorrect, inaccurate and wrong.

When tidi kicks in, chances are the fight is either getting underway, or is just starting. Tidi then stays on (with varying levels of dilation) for as long as it is required. There are no indications about what's going to happen whatsoever until the fight does start, so it's not a warning, and it certainly isn't early. What it is, is a graphical feedback to the users why things are suddenly going slower than they might otherwise expect. The only function the tidi progress thingy serves is to soothe the nerves of raging fat neckbeards who might otherwise foam at the mouth at CCP for not making their game work properly, much like they would if they were downloading a file and they got no indication as to the speed of the download or how much was left.

It certainly isn't a "well guys, it looks like you might be putting some strain on a server, better choose who are worthy to stay and decide the outcome of this random battle #423478234. Please be honourable." warning, which you seems to think it is, for some reason.

Cyno's lit, bridge is up, but one pilot won't be jumping home.

RIP Vile Rat

Zagdul
Federal Navy Academy
Gallente Federation
#177 - 2012-05-13 11:56:00 UTC  |  Edited by: Zagdul
@TWHC Assistant

Lets use a traffic analogy with magnets and science.

Pretend a car = packet or a request.

Scenerio one year ago:

Thousands of cars travel down a 5 lane freeway and right before the bridge to get over the river, that freeway would bottleneck into 3 lanes.

A year ago, Tranquility had no speed limits on the freeway leading up to that bottleneck. So, when the cars got to that bottleneck, they'd crash. When they crashed, some NEVER got over the bridge. Those cars that never got over the bridge represent lag and forgotten requests.

Those cars... got lagged.

Today:

CCP decided to implement a speed limit, however not just any speed limit, they installed awesome magnets of science under the road that force all the cars on the road to slow down. The strength of the magnets are based on the traffic on the road many miles (or kilometers) before the bridge. The more traffic, the stronger the magnets get.

Now, all the cars get to the bridge in an orderly, spread out fashion and everyone gets to their destination. Without being left behind. Without lag.

Dual Pane idea: Click!

CCP Please Implement

HVAC Repairman
Hedion University
Amarr Empire
#178 - 2012-05-13 12:14:58 UTC
TiDi is the best feature EVE has added in the last few years and everyone on team gridlock will be getting a case of their favorite domestc or imported beer on me whenever I'm in Iceland.*




* I'm never going to be in Iceland
Takseen
Federal Defense Union
Gallente Federation
#179 - 2012-05-13 12:20:03 UTC
Gorenaire wrote:
Shpenat wrote:


There is nothing CCP can do about this situation right now. There is no code to allow node loading on the fly. To move the systems to different nodes they would have to take down the entire cluster. That is not an option.



Well if TiDi is supposed to solve lag problem in a system due to lots of people in it and it creates tons of lag in a dozen others systems, what kind of solution is that ?


Prior to TiDi, the highsec systems linked to null/low/wherever the fights were causing the lag, just wouldn't have loaded at all, or would be intermittently laggy, which I find much more annoying.

Fredfredbug4
The Scope
Gallente Federation
#180 - 2012-05-13 12:28:36 UTC
Would you rather be slowed down by the game going in slow motion at a smooth framerate, or play in real time at 1 FPS and have to restart the client every 5 minutes?

Watch_ Fred Fred Frederation_ and stop [u]cryptozoologist[/u]! Fight against the brutal genocide of fictional creatures across New Eden! Is that a metaphor? Probably not, but the fru-fru- people will sure love it!