Combat Feedback for Update 5.

[R0cksteady]

So it probably won't have any requirements.

Does that mean there will be no way to get a fighter to 100% offhand for a fighter, or will GTWF get you there now, instead of needing STWF for it?

[Borror0]

Changing specific instances comes much later, because it usually gets tacked onto design cycles 2-3 updates away. At least in my experience.

Of course. Doing it the other way around would be counterproductive.

If there is a global problem that affects several instances, it's best to apply a global fix. Then, if there are any instances where the problem still exist you investigate to see what the rest problem is. Otherwise, you could waste energy fixing stuff that would be fixed by your global change anyhow. Additionally, if the global fix does not seem to fix anything during early testing, it follows that your assumptions about the problem were wrong and that the fix does not work.

If the global fix addresses most instance, then cool. Perhaps each time the cause of the problem is specific to that instance but, if the fix work, it does mean that collision detection did put a lot of stress on the server. If it does not work, then the fix won't go through and they'll try to fix the problem differently.

[REALb0r3d]

See you're saying it now. Its the collision detection huh?

Then skip the nerf, work the collision detection.

[Borror0]

See you're saying it now. Its the collision detection huh?

You're saying that like if Turbine didn't recognize that openly in the OP.

Then skip the nerf, work the collision detection.

Doing one beneficial change does not prevent you from doing a second one. Perhaps you don't think that the nerf is beneficial, but Eladrin does. Telling stuff like this won't sway him. He wants to fix the lag and improve the balance between TWF and THF. You're telling him to give up one of his goals, without providing any argument as to why.

[Natashaelle]

There was NEVER dps lag in Europe ...... We Europeans have never saw a dps lag ever before than in USA . Ask anyone but him and he will tell you the same thing (meaning the person that you quoted ) .

You know, I'm not lying about this.

In an earlier post, Razcar wrote a report about what the dps lag is like in US. Exactly the dps lag as he describes it there can, and does, occur on the European servers, although from Razcar's descriptions and thanx to his clarifications it now seems clear to me that the problem there is not so severe as it is in US.

The symptoms of dps lag as he has described them are identical to symptoms that I have seen on multiple occasions on the European servers.

[bobbryan2]

Of course. Doing it the other way around would be counterproductive.

If there is a global problem that affects several instances, it's best to apply a global fix. Then, if there are any instances where the problem still exist you investigate to see what the rest problem is. Otherwise, you could waste energy fixing stuff that would be fixed by your global change anyhow. Additionally, if the global fix does not seem to fix anything during early testing, it follows that your assumptions about the problem were wrong and that the fix does not work.

If the global fix addresses most instance, then cool. Perhaps each time the cause of the problem is specific to that instance but, if the fix work, it does mean that collision detection did put a lot of stress on the server. If it does not work, then the fix won't go through and they'll try to fix the problem differently.

Then by your logic... it would be more beneficial to do the collision detection and see how that works before doing a complete combat rebalancing.

Here's my problem with this whole thing. I see them as basically using lag as an excuse to do a combat rebalancing that they want to do anyway. But worse than that, I am not even convinced that these changes will greatly affect the 3-4 instances that people have so much problems in.

Maybe I'm wrong on both counts. But even then, it seems like we're putting in so many fixes at the same time that you'll never see which were effective, and which were completely unnecessary.

And the real problem with testing is that there isn't a big enough testing population to adequately stress test these systems. That's a greater problem with Lamania in general... that things like 'lag' aren't useful to look at on Lamania for the same reason that none of the devs can see lag on their own internal servers. It took them months to check and see what the problem with Shroud Part 5 was, because they were mostly testing on internal servers.

That means that the first "testing" it really gets will be once it's completely implemented... and then it gets messy trying to change it.

So understanding those limitations, I like smaller fixes, one by one.

[Natashaelle]

I know its off topic but there is NO DPS LAG in Europe or ever was . Don't believe what that guy says ...... ask any other European but him .....
(I posted twice cause i cannot stand people misinforming)

I'm not particularly keen on people accusing me of being a liar.

As far as I know, you have not played in Europe for several months, so that you are therefore very uninformed concerning the state of the game over there at the moment.

Are you really just going to sit there and pretend that there's never been any "Shroud lag" on the European servers ? Oh, that's right -- you are. That seems to be because you appear to be have an agenda to encourage the idea that this is just a server/hardware problem, rather than the causes of this lag as the *devs* have analysed them being accurate.

Well I'm sorry, but *I* cannot stand misinformation -- I can confirm that lag can occur in Europe in the same places and with the same symptoms despite those servers being extremely underpopulated. I have accepted Razcar's *constructive* input nonetheless, that the problem when it does occur is more severe on the US servers than on the European ones.

Oh, and please take note of the following contribution :

DPS lag has been reproduced on Lamannia several times. One of those is even documented on these forums if you don't believe me.

The Lammania server has an *even tinier* population than the European servers btw.

Just some extra feedback, I have been forced to adapt my tactics to deal with dps lag, during Shroud part 4 especially, to ensure that I don't suddenly just keel over dead for no apparent reason. The healers who run Shroud on Europe these days, the experienced ones anyway, don't wait for people's red bars to go down before they start spamming heals ; because if they do so, we get party wipe, due to the DPS lag.

PS. As why there wasn't ever any dps lag in Europe even when we had big population ..the conclusion is yours ........

Oh, and WHEN exactly was it that Europe had a "big population" ? During Beta and just after the game launched maybe ? But I thought that the Shroud was only released with Mod6, when the original four servers had already been merged into two, due to the low populations ?

[wax_on_wax_off]

I'm sure the two points that i wish to make have been mentioned but i can't read through 101 pages of posts just to find out.

1. Part of the skill of playing a TWF paladin is timing your smites/sacrifices to when you are doing double hits, if you take out that player skill then it makes the game worse.

2. TWF should not equal THF close to exactly. THF is easier to equip for, does more damage to groups, and now the feats can be applied by a sword and board character with a barstard sword or dwarven war axe (i believe?). This makes THF more versatile while TWF is more specialised at quickly killing a single foe and as such should have a somewhat superior damage output.

Regards the first point, i understand the need to reduce lag but from what i read in the post couldn't you leave the attack chains as they are while removing the physics detection on the second attack?

Well thats my 2 cents, tyvm, i hope the game keeps getting better and better.

[Meetch1972]

One of the other problems with random generation is entropy. Another reason why pre-generation is often used - smooth out the random load peaks.

The other option is dynamic dps/hp scaling. Most lag seems to occur when attacking large fixed HP objects - ie. bosses or portals. Why not dynamically half the number of attack calculations during these periods - just start dup attack 1/4, 2/5, and 3/6 in the default chain and 1/4/7, 2/5/8, 3/6/9 in the tempest III chain.

It's likely the statistical result will be the same over a 2-10 min combat event.

This comes down to optimisations by equivalence, whether there is an opportunity to exploit it, and whether the player notices the pattern early enough to take advantage of it. If such a thing were contained within the 1 round, it would probably work, but it cannot work without potential for exploit if noticeable patterns occur. If in the above example I saw my first three hits were particularly good (or even get a vorpal strike), I might be able to spin and push the remaining hits onto 2 other chosen mobs that I want to put a finishing move on, knowing it would happen.

Developers do need to be, and I'm sure have been(!), careful to avoid predictability resulting from pseudo-randomness which is algorithmic. eg if with a Pally high on Zeal that instead of having a 10% chance to proc double-strike I discover that every 10th hit is a double strike, and I know in the same series of 10 combat rounds that 1 and 6 are the 2 off-hand procs that won't happen (80%), I might be able to time my Exalted Smite III to coincide with it, guaranteeing 3xsmite strikes practically every time I use it. If there's anything other than a random value presented to each desired roll, then I can start using something approaching card counting type techniques to figuring out when I'm going to proc off-hand and proc the double-strike. A 10% chance to proc should never EVER mean a guarantee that it'll happen once within the next 10 rounds (and not for the remainder).

Generating or simulating entropy is indeed the bane of random number generation. Might I suggest a random number cache of potentially very high values like random(something close to max int) and leeching the values out by a series of modulus and division operations until the result of multiplying all the random numbers generated by it would be higher than the close to max int used. You can maximise the use of entropy that way.

Just thinkin'...

That is of course if random number generation is actually a factor in performance, which I think some are saying is not the case.

[TechNoFear]

I am a bit concerned with less collision detection when fighting a moving mob, especially when affected by some slowing effect (Defender stance, Sleet storm etc) or high latency (>300 ms).

Not overly happy that my Smites etc (landing with both hands) are now based on random chance rather than my skill.

Glad to see the balancing of Tempest 1 vs Tempest 2/3 and TWF vs THF.

It all comes down to the way it plays and how fast any bugs / adjustments are rolled out (ie for the speed effect items not 'fixed' in this pass).

I don't know for you, but I would vote against adding such a feature to a game. Catering to hardcore players at the expense of casual players is rarely a good idea, especially with a mechanism that is both counterintuitive and looks goofy.

I also think twitching may have a performance impact.

Turbine are reducing the frequency of collision detection calculations and possibly are going to cache data (the locations of the player/mobs) to increase performance (no need for new calc if neither has moved).

As twitching moves the player (requiring a new collision detection) you would lose any performance increase.

That is of course if random number generation is actually a factor in performance, which I think some are saying is not the case.

You can get true random number generators (TRNGs) and some very good software RNGs (as opposes to the big list of numbers in older implementations of widely useds IDEs / APIs).

I assume that the numbers are not generated as required ('Just In Time') but cached.

I used a list of random values and the processing thread removed a value from the list as required (for an OCR system to read ore car numberplates based on digital images captured as the train passes). The list has a thread monitoring it and generating new values, ensuring there is always more elements than are used under max load.

[R0cksteady]

ok, so now that fighters have no way to compete with two handed barbarian DPS, and with more of a cost for TWF (a bunch of points in Dex that would be better for putting into Strength/Con/intelligence, a few feats and double the cost for getting weapons), are GS weapons going to be able to get deconstructed so I don't have a bunch of useless kopeshes? Not that it matters, I reincarnated my fighter and I'm not going to bother leveling it up anymore. But I could use those ingredients on other toons.

[Borror0]

Then by your logic... it would be more beneficial to do the collision detection and see how that works before doing a complete combat rebalancing.

I would hope that they did test internally, at least a little, before posting about it on public forums. You're right, though. It would be better PR to load the non-nerf version on Lamannia first to get clear feedback on the lag/feel and, once they are certain it works, toss in the nerf.

And the real problem with testing is that there isn't a big enough testing population to adequately stress test these systems.

DPS lag has been reproduced on Lamannia several times. One of those is even documented on these forums if you don't believe me.

[Natashaelle]

It's well documented that Internet in Europe is far better than Internet in USA. In Europe most countries have more average bandwidth available per user than USA ISP's. If the servers were the same and the server populations were similar then the differences would be the ISP's Turbine uses in each location.

This could be another explanation, apart from populations comparison, why the dps lag issues are generally less severe in Europe than US ; although people can have some absolute rubbish bandwidth in Europe too.

[Natashaelle]

Does that mean there will be no way to get a fighter to 100% offhand for a fighter, or will GTWF get you there now, instead of needing STWF for it?

It might be a good idea to make an alternative 2WF focused Fighter Capstone, if people wanted it ? Make a couple more for 2HF and S&B while they're about it ?

[bobbryan2]

DPS lag has been reproduced on Lamannia several times. One of those is even documented on these forums if you don't believe me.

Well, what we call DPS lag at any rate. And I didn't say it's impossible to spot lag on Lamannia. I said Lamannia isn't a good indicator of how things will run on live servers, because it's not stressed to the limit.

Basically... a positive lag test, like those, means something.

A negative lag test wouldn't necessarily mean anything.

[R0cksteady]

It might be a good idea to make an alternative 2WF focused Fighter Capstone, if people wanted it ? Make a couple more for 2HF and S&B while they're about it ?

Well as it stands, 110%/80% is a 15% nerf to DPS. A TWF fighter is already barely able to keep up with barbarian THF, which was dumb since you have to put your dex up to 17 and you still end up with a lower to-hit, even if you take OTWF. So basically, unless they put put a capstone to at least bring it up to 110%/100% there is no point to ever making a fighter for DPS. Which is really the only good roll for a fighter, since S&B is **** and Barbs are better for two handers. Pretty much breaks fighters.

So until they realize that if anything, TWF needed a bit of a boost to seem worth the extra cost over THF, fighters are useless.

Hopefully I get my greensteel ingredients back at the least.

[Borror0]

Basically... a positive lag test, like those, means something.

A negative lag test wouldn't necessarily mean anything.

The procedure is ridiculously simple:
1. Reproduce the problem internally
2. Implement fix
3. Try to reproduce the problem, again, on the same server with the same characters

If the problem disappeared on the second test, then it's safe to conclude that the fix did reduce significantly the load on the server. Perhaps there would still be lag on the live servers, but it does not take away from the fact that one cause of lag was addressed.

[TechNoFear]

Hopefully I get my greensteel ingredients back at the least.

I do not think GS deconstruction will be released anytime soon, and when it does I assume it will be as expensive as LR/GR.

[R0cksteady]

I do not think GS deconstruction will be released anytime soon, and when it does I assume it will be as expensive as LR/GR.

Well, why wouldn't they at least open it for the people who's builds are getting ruined?

[Wizzly_Bear]

Well, why wouldn't they at least open it for the people who's builds are getting ruined?

Because they never have. They don't compensate for ruined characters based on major game changes. It sucks, but its been that way for at least as long as I've been pummeling Stormreach's denizens since back in Mar of '07.