Update 25 Combat Adjustments

Alienoutlaw

looking for to getting some sustain back, changes to block look promising, i'm not sure how to take the set proc adjustment statement just yet and a little confused as to what is ment by cost as set proc's have no resource cost at the moment, but so far so good, no point asking for this or that as the patch is already done (PTS is just a PR exercise)

Alienoutlaw

JumpmanLane wrote: »

ZOS_BrianWheeler wrote: »

Hello all!

We’d like to take some time to go over our goals for the next Update coming to ESO, and some of the more noteworthy changes you’ll see.

For Update 25, the Combat Team’s primary focus has been on performance improvements, particularly related to Block and item set proc conditions. While making these adjustments, we are also bringing any item sets affected into combat standards. As a reminder, the term “standards” means we are simply using consistent formulas that determine how much an ability should cost and how effective it should be based on various factors such as being an Area Effect, having an Over-Time effect, complexity of execution, etc. This process will continue in future updates for item sets.

Block is an ability that everyone can use at any time, which means it can create a lot of traffic for the server to handle. In our continued efforts to improve performance, we are adjusting how the ability works under the hood. You will still be able to do everything you used to with Block. What you’ll notice is animations will blend a bit differently out of a “block cancel”, but the effectiveness of an ability will remain and function the same; the ability will still fire when “block cancelled” but you won’t have any “dead time” before the global combat cooldown ends. This should also get rid of a few health desync issues when getting killing blows with abilities that have been triggered after a block cancel. By refactoring Block much the same way we did with Sprint in the previous Update, we expect the usage of that ability to be more performant.

Next, we’ve been focusing on bug fixes and conditional logic (item set and ability procs). Many abilities and item sets have to step through several requirements to fire their effect beyond their standard line of sight, cost, range, etc. From these changes, the conditions for procs will now work more reliably with their stated requirements. For example, Scathing Mage will no longer erroneously trigger from the first tick of any Damage over Time, and Reflective Scales will no longer work versus channeled effects such as Radiant Destruction. While making these fixes, many of the item sets affected have also been updated to fall in line with our item set or ability standards to ensure their overall power remains closer to their intended design.

Finally, we are addressing Magicka/Stamina sustain, especially in encounters where longer fights or prolonged damage in specific areas are required. With that in mind, we are reducing the cost of Area of Effect Damage over Time (AE DoT) abilities to reduce the pressure on your Magicka/Stamina while utilizing a rotation involving an AE DoT. This does result in a revised standard between Single Target DoTs and AE DoTs which we will be evaluating with the live game and on the PTS.

Again, our focus for U25 is primarily on performance updates and bug fixing which you’ll see in more detail when we publish the patch notes. We encourage everyone to hop on the PTS when we publish U25 to run through the adjustments and give us feedback!

What does no dead time after a skill is cancelled with block mean? Say I block cancel the animation on a whip. Does that mean the next whip will be faster minus the dead time.

Usually the animation for a whip looks like a flame shooting straight down a sword or staff. Sometimes it can seem just to be a little puff.

Block casting whips to make them faster now is pointless because of GCD. Would your tweak make them actually faster?

im going to assume it means they have increased the animation time for block to match the GCD of skills so there is no dead or static time

JanTanhide

Could just invest in a more powerful Server to handle the player load.

relentless_turnip

JanTanhide wrote: »

Could just invest in a more powerful Server to handle the player load.

It isn't a hardware issue.... with the population at the height it is they would already have one of the most impressive servers available. It is most likely the complexity and the quantity of code that is sent back and forth.

I think what they are doing is addressing that by reworking skills, sets etc...
We will be reinstalling the game on the next patch with presumably a significantly smaller file.

Starlock

Joy_Division wrote: »

Starlock wrote: »

Casterial wrote: »

From PVE point, removing block cancelling will be a disaster. All PVE tanks naturally block cancel all day long. Imagine that you need to wait for full heroic slash animation or chain animation... tanks will be killed while waiting for animation to end and since majority of damage in latest veteran content one shots if not blocked, no amount of healing can help with that.

Oh, I'm aware - I mostly tank myself. It's why I mentioned in the original comment that I don't expect this change to happen and it would likely be better suited for an ESO2. There would need to be a lot of rebalancing and retooling of game content to adjust for the shift.

Also, to be clear I'm not talking about removing block canceling but making it so abilities do not fire when block canceling. Basically, if your try your heroic slash and then block cancel, your attack doesn't actually connect, do damage, or have any of its effects. Your block otherwise goes off as normal. What this would mean for gameplay is more strategic timing of blocking, because if you activate an ability and then cancel it with a block, that ability doesn't fire. Different abilities might have different standards for whether or not they would be interrupted by blocking. Some abilities I could see being allowed to be cast while blocking, while others would cancel their effects if interrupted by a block.

To put a D&D nerd perspective on it - if an ability only requires vocal components (e.g., vigor), it can be block-casted and block canceling doesn't stop it from firing. If an ability requires somatic components (e.g., heroic slash) it can't be block casted, and block canceling will stop it from firing. It just makes way more sense to me than the super weird, unintuitive system we have now. That block canceled abilities can still connect is easily the most bizarre thing about the ESO combat system to me.

And your suggestion, which is please get rid of animation canceling with different wording, would be a frustrating disaster.

*sigh*

If there's one thing we can count on with discussions on this topic, it's constant misrepresentation. Even, apparently, from players who were once supposed to accurately represent player concerns.

Shantu wrote: »

Yeah, I could go either way on the issue. On the one hand, why design animation into a game that can be cancelled? On the other, with a little practice it's not that hard to learn. At any rate, you're probably right; it's a little late in the game to start changing things now.

It *is* probably too late, which is something I mentioned in my first post about this in this thread. Along with a bunch of other nuances and caveats in other posts that folks seem content to ignore. But hey, I'm used to strawperson tactics. I'm just going to move along and shrug now...

RMerlin

nafensoriel wrote: »

Honestly, the obsession with server upgrades is extremely out of date. It was true in the late 90s. It isn't true now. With VM and containerized development speed between servers and speed between storage is far more important than the raw CPU speed. CPU speeds, in general, are not advancing all that quickly. A 6700k from 2015 is barely 15% slower single-core wise to a brand new state of the art ryzen. No one in software brute forces tasks anymore. We found out there are serious hard stops in doing that.

When I talk about hardware upgrade, I don`t just talk just about CPU clock speed or IPC. Core density, bus bandwidth, RAM, I/O storage - all of these can have an impact on performance. Is the data stored on 15K RPM SAS? SSD? What about data caching? Bandwidth between the server instances? How is the database load spread across multiple server instances?

Considering that the game launched five years ago, there has been a lot of hardware improvement since 2014, and not just about raw CPU speed. Since the game launched in 2014, I assume a portion of their hardware infrastructure is older than that, probably dating back to the last development days. Since we never noticed any sudden performance increase, I suspect that there has been very little improvements brought to the actual hardware infrastructure (or at least any meaningful one).

I don`t say that everything can be resolved just by upgrading hardware, but it`s one piece of the puzzle. And so far the optimizations have barely kept up with the load increases introduced over the years by the addition of many new skills, item procs, game mechanics, etc... While they mentionned those software optimizations over the years, I don't recall them ever mentioning a single hardware upgrade.

So far, the results tend to indicate that the issue cannot be tackled only from the software optimization side.

Tommy_The_Gun

So PTS notes next week huh ? Another judgment day I guess. Oh well...

katorga

Kalante wrote: »

They better not mess up animation canceling with block now. Animation canceling has been ruined so much due to their "revisions". Combat has become so clunky. Also remove ult cast times its a terrible change that nobody likes.

Block cancelling is going to get wrecked. Just from the description, you know it is.

OrderoftheDarkness

I hope that someday I will see a well-developed classes weaving and not just a shell for using dizzy-exe, etc, or rather some classes already have it, but not all.
I stopped playing because of that. I don't like being forced to use abilities that are more effective than class ones.

Aikar

RMerlin wrote: »

nafensoriel wrote: »

Honestly, the obsession with server upgrades is extremely out of date. It was true in the late 90s. It isn't true now. With VM and containerized development speed between servers and speed between storage is far more important than the raw CPU speed. CPU speeds, in general, are not advancing all that quickly. A 6700k from 2015 is barely 15% slower single-core wise to a brand new state of the art ryzen. No one in software brute forces tasks anymore. We found out there are serious hard stops in doing that.

When I talk about hardware upgrade, I don`t just talk just about CPU clock speed or IPC. Core density, bus bandwidth, RAM, I/O storage - all of these can have an impact on performance. Is the data stored on 15K RPM SAS? SSD? What about data caching? Bandwidth between the server instances? How is the database load spread across multiple server instances?

Considering that the game launched five years ago, there has been a lot of hardware improvement since 2014, and not just about raw CPU speed. Since the game launched in 2014, I assume a portion of their hardware infrastructure is older than that, probably dating back to the last development days. Since we never noticed any sudden performance increase, I suspect that there has been very little improvements brought to the actual hardware infrastructure (or at least any meaningful one).

I don`t say that everything can be resolved just by upgrading hardware, but it`s one piece of the puzzle. And so far the optimizations have barely kept up with the load increases introduced over the years by the addition of many new skills, item procs, game mechanics, etc... While they mentionned those software optimizations over the years, I don't recall them ever mentioning a single hardware upgrade.

So far, the results tend to indicate that the issue cannot be tackled only from the software optimization side.

This game doesn't run on a single server... It's less to do with server spec upgrades, and everything to do with ensuring code is scalable and that adding MORE servers distributes the load evenly.

Hardware improvements increases bang for your buck per server count, but you reach a point where you can't vertically scale any more and it's more cost effective to scale horizontally.

With zone instancing, this is a clear sign of horizontal scaling. If an instance is constantly overwhelmed, maybe they are tuning too many players to same instance.

They can implement less players per instance, moving solo players in between instances if one becomes overload, etc.

Dungeons are easy, separate instance per dungeon.

Then question how many instances are ran per physical machine? If small instances lag, then they are over crowding the host. Add more servers, less instances per host.

Sandman929 wrote: »

So...you're not going to limit healing to group members? Block is the big performance hit huh? Ok then.

In relation to above reply, group healing really shouldn't be a huge impact. aoe heals will mainly occur in group content such as dungeons and trials.

Well, in those, instance members === group members. zero impact change to do that for the main place a group heal might be used.

For group heals in overworld, 40 players vs 4 players is not a huge iteration cost to do distance based lookups.

Plus, they could do regional segmentation of an instance (this chunk of the zone has 8 players, this chunk has 4) and only inspect nearby chunks for distance based lookups. Now we are down to 4-8 nearby players for distance based checks for aoe calculations.

Then, consider this is only talking about healing.... Healing is going to be a trivial percentage compared to AoE calculations for damage.

Totally pointless area to try to optimize.

Now block? Block is a request from client that is going to happen per player, that is going to be happening a lot with people using it to animation cancel, pvp etc. Then factor in state changes sending packets to other clients showing you are blocking, changes for the game to detect you are blocking, and yes, block has significantly more impact to server load than group healing.....

Elsonso

RMerlin wrote: »

Considering that the game launched five years ago, there has been a lot of hardware improvement since 2014, and not just about raw CPU speed. Since the game launched in 2014, I assume a portion of their hardware infrastructure is older than that, probably dating back to the last development days. Since we never noticed any sudden performance increase, I suspect that there has been very little improvements brought to the actual hardware infrastructure (or at least any meaningful one).

Based on statements that Firor said prior to Zenimax Online Studios, my expectation is that ZOS leases the hardware. If that is the case, then I also expect that it has been upgraded since launch. Probably more than once.

Additionally, we know that they added server capacity on multiple megaservers in early 2019.

Banana

I didn't think wings had much left to NERF.

RMerlin

Elsonso wrote: »

Additionally, we know that they added server capacity on multiple megaservers in early 2019.

Now that you mention it, I had forgotten about the EU server upgrades last year.

Somber97866

I left for 2 months, cancelled my eso subscription and I'm now back to see if you all have learned anything from the past year. Depending on how this pts and new update go will depend on me leaving permanently! It's up to you!

Kuratius

Something like block shouldn't be a huge performance cost. If it is, then it's probably due to the underlying language being inefficient (they probably coded combat and skill checks in some proprietary scripting language that is not great compared to soemthing like C++)

Another option would be to a) let clients make block calculations themselves and just tell the server the result
b) assuming cpu load isn't the issue, then adding better network prioritization/sorting might help.

Somber97866

Kuratius you seem to know what your talking about! Will you please help out the group of ppl that created my fav game before we spend 6 months with glitches and trying to get them acknowledge there is a problem! No offense my wonderful ZOS works! Then spend 6 more months while they get it ironed out?!? PLEASE!!!!!!
PLEEEEEEEEEEEEEAAAAAAAAASE!

coletas

how do we kill invisible enemies now? lol XD

Deathlord92

Please please please buff stamblade give major fracture back to surprise attack fighting tanky builds is a complete nightmare and stun back on incapacitating strike I’m begging you 🙏

TwiceBornStar

ESO is definitely running smoother on my 2K 144hz screen than it did before the update, and combat seems slightly more responsive too. Just a tad more fluid, but noticable! FPS still drops after a while, but far less and far less frequent. I guess that was the whole point of this exercise, so I'd say mission accomplished! Cheers!

Muskrap

TwiceBornStar wrote: »

ESO is definitely running smoother on my 2K 144hz screen than it did before the update, and combat seems slightly more responsive too. Just a tad more fluid, but noticable! FPS still drops after a while, but far less and far less frequent. I guess that was the whole point of this exercise, so I'd say mission accomplished! Cheers!

The game is made directly worse from this block change. You getting stuck in animations is killing this game's combat and is lowering the skill cap. They're talking about desyncs but more people are getting desynced by shock damage and ele weapon/ overload. Imagine changing the games mechanics before actually fixing any of their abilities