LIVE shroud testing event to compare with lama s&b vs twf

[Drfirewater79]

Welcome Turbine fanboi's and haters alike

welcome to Shroud-a-polusa : proof is in the testing (sarlona server)


on saturday june 19th we are throwing a back to back test of lag on live servers

first at 6pm EST a s&b only event ... attempting to see how bad the lag is on live servers in an all s&b shroud ....

then directly following WE DO IT AGAIN .... this time ALL TWF .... its time to prove once and for all that shroud lag is not just caused by twf but overall dps.



Please note that for balanced testing we would prefer that all testers have either tier III shroud weapons OR silver/good dr bypassing weapons for both runs.

we would love it if we could get a stealth Dev to join (send me a PM no need for people to know if or who the dev is but i think its important for them to see and hear peoples reactions preferably a DEV who wont take it personal if there is alot of negativity tossed around which could be expected since this is live not lama and lag is expected in both runs)

hope also that we can grab at least one of the Lama testers so they can compare our lag with the lag they found on lama servers as well.

if your interested please note below if your interests lie in

s&b only

or

twf only

or

both ...


if you wish list what weapons you use in part 4 and 5

so that i can make sure we are using similar numbers of dr bypass and shroud weapons in both runs to keep the runs equal for testing purposes.

Thank you and see ya at the event.

sign ups also found at DDOcast forums ...http://s7.zetaboards.com/Ddocast/topic/8302029/1/

[ghettoGenius]

Good luck with your test. I hope you will consider the possibility that various classes contribute to lag in different ways. For instance, classes with many extra damage modifiers such as pallys/rogues/barbs almost certainly add more dps lag than those that dont do additional damage. I'm thinking the best way to isolate this variability is to have same or similar group makeups in those runs. Just a thought.

Edit: Shouldve listed monks as well as they seem to add more additional die rolls.

[grodon9999]

Which live server are you testing on? Count me in with Ghallanda, my ranger has a Madstone Shield.

[smatt]

hmm, I salute the idea to try and test..... But, I'm trying to figure out where you got the idea or read that all Shroud lag is caused by TWF? Perhaps it has been said that various aspects of TWF are causing a good portion of it, but exactly what aspects? This is the kind of pure propaganda tha causes so many people to be misinformed. Talk radio/ Cable News blah blah.....

Sooo.. you would have to gather a wide variety of TWF classes, documenting exactly what feats/enhancements they had... Then run many mnay mnay and a few more many runs to actually have any kind of results..... I think Eladrin said, that perhaps something with monks was a possible major culprit... More likely though it's a cumulative affect from many sources...

Good luck though...... Anything might help

[grodon9999]

hmm, I salute the idea to try and test..... But, I'm trying to figure out where you got the idea or read that all Shroud lag is caused by TWF? Perhaps it has been said that various aspects of TWF are causing a good portion of it, but exactly what aspects? This is the kind of pure propaganda tha causes so many people to be misinformed. Talk radio/ Cable News blah blah.....

Eladrin did say this in the thread where the proposed TWF changes were first presented.

[Shade]

The lag encountered in large raids IS caused by twf/monks directly. This is a known fact and your test will just go to further show that.

Personally I've seen hundreds of test cases and the results are conclusive.

In the 100+ VoN6 epics i've done:
Everytime there was bad lag: There was at least 3 TWF
Everytime there was insane lag: At least 4 twf
Worst lag ever: 6 twf, couple monks
Everytime there was little to no lag: Almost everyone was THF, with maybe 1 TWF
Zero lag: Everytime everyone was THF

Same goes for tower.

The most people attacking with the more effects going at once, the more likely you are to lag.

That's not to say THF or SnB can't cause lag, they can. They just don't get as many effects per second so they contribute allot less towards hitting the "event horizon" that causes it.
IE: If you have 4 TWF attacking and no lag, and 2 more THF start attacking, the lag could very well start up.

But turbines solution of nerfing everyone is not the way to fix this.

As if people recall, years ago when the shroud first came out and the servers weren't very busy.. You could have 10 TWF attacking with full t3 greensteel and no lag. As less people playing meant we had more bandwidth available per raidgroup.

[ghettoGenius]

I tend to agree with Shade on this, in the sense that making changes to the combat system seems like the wrong direction to go. In every scaling RPG I've ever played (meaning characters get stronger as they level up and hence more calculations are made in combat as they progress), the lag gets worse as you level. Sometimes its dps lag, sometimes its network latency, sometimes its graphics lag due to so many visual effects.

However, in all cases performance was improved by upgrading hardware not software. It would probably be more expensive, but I think Turbine should at least consider throwing another hamster or two into the server wheel to beef up performance. The additional cpu power and network capability would no doubt help, and this offers a path to not only reduce lag, but keep most traditional players (myself included) happy by not deviating too much from the D&D ruleset as well as free up devs to work on content additions.

[Drfirewater79]

Good luck with your test. I hope you will consider the possibility that various classes contribute to lag in different ways. For instance, classes with many extra damage modifiers such as pallys/rogues/barbs almost certainly add more dps lag than those that dont do additional damage. I'm thinking the best way to isolate this variability is to have same or similar group makeups in those runs. Just a thought.

Oh good point i will try to keep it balanced once i see what people are interested ... two weeks to figure it out and sure i am gonna get more response on DDOcast forums then live ... closer community there.

Which live server are you testing on? Count me in with Ghallanda, my ranger has a Madstone Shield.

Sorry grodon .... though i mentioned sarlona ... wish i had lvl 20's on multiple servers but grinding new gear or paying for transfers so i could have similar builds on multiple servers just wasnt in my cards

i edited the post to state sarlona server

hmm, I salute the idea to try and test..... But, I'm trying to figure out where you got the idea or read that all Shroud lag is caused by TWF? Perhaps it has been said that various aspects of TWF are causing a good portion of it, but exactly what aspects? This is the kind of pure propaganda tha causes so many people to be misinformed. Talk radio/ Cable News blah blah.....

Sooo.. you would have to gather a wide variety of TWF classes, documenting exactly what feats/enhancements they had... Then run many mnay mnay and a few more many runs to actually have any kind of results..... I think Eladrin said, that perhaps something with monks was a possible major culprit... More likely though it's a cumulative affect from many sources...

Good luck though...... Anything might help

The nerf to TWF was because of multiple physics checks stating that the mutliple physics checks are to blame for the lag ....

also the 10% alacracy was taken out because it requires the same multiple physics checks BUT FASTER.

so to squash the propaganda i want to do this test ... which will prove once and for all if twf is the culprit .... this is why we are doing it on live servers not on lama

this way with current game stats ... we see if s&b has less lag or same lag as TWF

if it has the same lag as TWF then it proves that the new TWF nerf is not needed and as some testers have pointed out from lama test ... is actually worse then live.

HOWEVER

if lag is obviously worse in the TWF party then it proves the latter ... that the nerf is needed and then people can stop fighting about it.

I have noticed my mail box have letter after letter of friends not in my guild stating they are leaving the game .... after trying out lama server and reading the lama forums.

I think this test will help them see that the changes are NEEDED ... but if it proves other wise ... then it tells turbine that they should delay these changes and work on a better way to fix the lag on overpopulated servers.

If you ask me my opinion over populated servers are the major issue ... and the fact that turbine will not open new servers allowing free server transfers to new servers for the sake of reducing populations of individual servers with high traffic .... this fact is the reason we have massive lag that we never had pre mod 9.

it makes no sense to open a new server and not allow transfers cause the over populated servers are not just over populated with new players but over populated with vets from the last server merge.

people who would have to beable to move there whole guild or friends list to a new server before it would be a viable option.

if turbine opens two new servers allowing free server transfers to the two new servers for 1 weekend they would notice a huge dip in dps and server lag per server as troubled servers have three - four guilds worth of players changing for the hope of better lag conditions.

[stainer]

The lag encountered in large raids IS caused by twf/monks directly. This is a known fact and your test will just go to further show that.
<...snip>

Shade,

There is something about the lag problem and how blame is laid that bothers me. This is a personal observation I have had, that I can not explain. I am hoping that you, or someone else, could give me some insight on what might be causing it.

I have a capped monk (dex/wis), and a nearly capped, lvl 19, barbarian (THF) among my characters. When leveling each of them I have solo'd in GH and the vale. My monk flys through each of the areas without an issue. My barb, on the other hand, will have times when the game completely freezes. This usually occurs during large fights. I just keep the attack button pushed in, and hope for the best. Sooner or later, my computer "catches up" with the server and I get back on sync. The delay can last (estimating) from 45 secs to 2 or 3 mins. The only difference between the two runs is I have a pocket cleric with the barb. I am at a loss to explain why this is happening. If you follow the logic that is laid out in the forums, the situation should be reversed and my monk should be the one lagging out. Any ideas?

[grodon9999]

As less people playing meant we had more bandwidth available per raidgroup.

My money is on turbine throttling back bandwidth since F2P. no way to prove it but it fits.

[smatt]

My money is on turbine throttling back bandwidth since F2P. no way to prove it but it fits.

[Scraap]

Shade,

There is something about the lag problem and how blame is laid that bothers me. This is a personal observation I have had, that I can not explain. I am hoping that you, or someone else, could give me some insight on what might be causing it.

I have a capped monk (dex/wis), and a nearly capped, lvl 19, barbarian (THF) among my characters. When leveling each of them I have solo'd in GH and the vale. My monk flys through each of the areas without an issue. My barb, on the other hand, will have times when the game completely freezes. This usually occurs during large fights. I just keep the attack button pushed in, and hope for the best. Sooner or later, my computer "catches up" with the server and I get back on sync. The delay can last (estimating) from 45 secs to 2 or 3 mins. The only difference between the two runs is I have a pocket cleric with the barb. I am at a loss to explain why this is happening. If you follow the logic that is laid out in the forums, the situation should be reversed and my monk should be the one lagging out. Any ideas?

If the surmise is correct, and it's due to additional transmission of damage type+ damage number for each and every effect proced, large fights with glancing blows on upwards of say... 5 enemies? more? Could easily meet or exceed 1v1 monk blows in terms of pipeline usage, I should think, since those as well are getting transmitted all at once.

[Drfirewater79]

My money is on turbine throttling back bandwidth since F2P. no way to prove it but it fits.

If that was the case then wouldnt it make more sense to throttle up or simply purchase more bandwidth space rather then completely destroy the combat system which is DDO's biggest selling feature to date?

Shade, '

I agree with you ... to me its conclusive and the PCS checks makes sense too ..

however there seems to be alot of people (not to say i am not included in that group) who hear about the tests done on lama and lag being worse regardless of the changes made.

if lag is worse then common opinion is dont make the change ....

but if you can prove with hard evidence that S&b (not high dps THF) has no lag while full TWF creates unmesurable lag it silences the critics right?

personally i dont know why turbine hasnt put this type of thing together on there own to prove there point ...

and i am kinda sad that turbine hasnt stepped up and said they are willing to monitor the event for data collecting purposes ..


if i thought we could survive with a full monk group doing shroud to show the scale from a monk side .... i would run that too ... maybe that is what i will do tonight on sarlona

lvl 20 monks only + 2 full healers and 1 full caster

[systemstate]

The problem I see with your test is that you will be unable to account for other groups running quests on the same server(s) while testing is being done.

According to Turbine, they feel that lag is, in part, being created by the system having to run so many simultaneous calculations. If this is true, then other groups running quests on the server will skew your results. You will not be able to get any meaningful data from this test because your group won't be the only one on the server(s) during the time of the test. Also, you would also have to rule out lag introduced by each player's ISP.

If the theory of DPS creating lag on the server is true, then other groups- even solo runs should be experiencing the lag simultaneously with the players who are generating the DPS lag.

The only group who can successfully validate where the lag is coming from is Turbine. And they can only do this by running their tests on a dedicated test server from clients on the same network, with no other users connected.

Your test, while well-intended, will only add to further speculation with not enough information.

[stainer]

If the surmise is correct, and it's due to additional transmission of damage type+ damage number for each and every effect proced, large fights with glancing blows on upwards of say... 5 enemies? more? Could easily meet or exceed 1v1 monk blows in terms of pipeline usage, I should think, since those as well are getting transmitted all at once.

I would agree, but this goes against the arguments that other, more experienced (than me) players have submitted. It would point to a conclusion that lag is complicated and could have many contributions.

I have also noticed if one person is dc'g that lag is increased significantly. This is especially noticable in raids.

I dunno if the answer is as simple as people would make it out to be.

[Drfirewater79]

First off i want to thank the people on this thread for being understanding and constructive in your critisim (sorry bad spelling)

but unfortunately the other thread has caused me to no longer want to help this community out by attempting a fun and interesting LIVE server way to push some of these topics out into the open with field data ...

I will still be alowing sign ups for those interested on DDOcast.com forums but will no longer be involved in the debate ...

some things being considered are also a all monk + healers and a caster run to test monks part in this lag issue

and a THF run for seperate high dps low PCS calculations for balancing issues ...

please feel free to continue the discussion and hope you all have a great day.

[Vanshilar]

Actually what you could consider doing is have two runs, consisting of:

1) TWF/monks/whatever using weapons with next to none weapon effects (say, just a regular +5 khopesh)
2) TWF/monks/whatever using weapons loaded up with as many weapon effects as possible (say, +1 shocking burst khopesh of pure good, ToD rings for monks, etc.)

Note that the example weapons I use here don't bypass DR, I don't think bypassing DR matters for the purposes of the test, although of course it makes the fight shorter. However, you may actually want the test to be longer, to accentuate the total load that the server takes. Going by Eladrin's bucket model, you want the bucket to fill up as much as possible so the time it takes the water to dribble out is also as long as possible, to magnify the effects of DPS lag under whatever conditions you use.

Note that you may have to differentiate between overall DPS and number of damage effects per second. That is to say, one big hit every second could have the same overall DPS as many little hits every second, but the latter may cause more lag than the former. I've already shown that weapons of similar DPS (regular sword of shadows vs lightning 2 falchion) contribute significantly different amounts of network bandwidth even though they have similar DPS here; since they have similar DPS, the difference is presumably in that one has no additional damage effects while the other has a bunch of them. In fact, I would guess that it's possible that low DPS with lots of damage effects may actually be more laggy than high DPS with few damage effects.

Or another way to test this is using THFers of similar DPS to TWFers/monks, and having one round of mostly THFers and another of mostly TWFers/monks. THF has less damage effects per second because of the pitiful proc rate of glancing blows, so testing this would also keep the DPS roughly the same.

Basically in all these cases the overall DPS is kept roughly the same. I'm looking at the issue more from a number of damage effects per second perspective, rather than an overall DPS perspective, so I'll be doing tests along those lines. The key piece of information that I'm missing right now is whether or not a bandwidth cap exists, and whether or not it coincides with the onset of DPS lag as well as whether or not it coincides with the end of DPS lag. I'll try to get a raid in when I'm less busy to test this out hopefully sometime today or tomorrow.

[smatt]

Eladrin did say this in the thread where the proposed TWF changes were first presented.

No what he said is that they thought it was a big contributer, and one that would also address other issues they had going forward in the future. But he didn't say it was the only cause.... There are so many varying causes for lag.... Exactly which problem is the most prevailant at any one time or in anyone party or in any one quest isn't neccesarily all that easy to track down. Go back through the Dev trackers and read through ALL of his posts on this subject.. Then take them as a whole, instead of keying in on 1 particular statement. He says a lot more indirectly than he does directly...


I'm all for players throwing out ideas to help the Devs come up with what might be better solutions.... Gathering as much live server info as possible would likely help them. But it has to be pretty detailed info... And the players certailyl wouldn't be able to come to any conclusions themselves.. But it might provide some valuable data for the Devs to take a look at. screen shots of party/server etc.. They could pull the data for those toons and recreate the party... Then run it on their in house servers IF they thought it would help them track a a particular problem.

Again,the only problem I see are too sure of themselves players thinking that one of the se runs is proof of anything .

[REALb0r3d]

I have a monk I'd like to bring along. +4 metalline wraps or if you want more efffects I have some +3 metalline of enfeebling wraps, shocking burst ring, and when we get to the 2nd shrine I can upgrade one of my rings to holyburst.

I think I have 775 TP and haven't bought the vale pack on Lamma yet.

[Thrudh]

It would point to a conclusion that lag is complicated and could have many contributions.

LOL... yeah, that couldn't be true, could it?