Caith discussing the Scheduler for Everquest 2

This is a screenshot from Discord of Caith discussing the Scheduler.

[/IMG]

 

Caith discussing auto attack.

 

dmg proc's are the new autoattacks!

 

I know a few people who need to read this....especially when they are complaining about lag. /sigh

 

This explanation from Caith doesn't make any sense.

A long time ago, we were told it was changed to be the following way to help with server lag:

1: When an ability is cast, it should only cast (sent to the server) if the client thinks it is available (not on cooldown, stunned, charmed, etc).
2: If something is already casting, it is added to the ability queue (all client side).
3: If the ability queue already has something to cast next, that is thrown out and the new cast request is added to the queue (all client side).
4: The ability in the queue is only processed (starting at step 1) once the client receives confirmation that the previous ability was handled.

This above should be the case whether it is done from a casting macro or hitting 6 keys in quick succession. If they want to do a last check on the server to make sure an ability is castable (i.e. to prevent exploits) that's fine but there's no reasonable reason for the server to be dealing with more than one ability from each player at a time.

Blaming server lag on players hitting too many abilities at the same time is just deflecting from the real problem that is causing lag (whether that is known or even solvable).

 

For the server, abilities is not just the buttons we press. Each autoattack hit (including flurries/multiattacks/AE auto), each chance for a proc to go off, each dot tick, every attack done by pets and dumbfire pets. It all counts towards the things needed to be done serverside.

Also, not sure why he sais mob casting doesn't matter much, as those constant AE dots in raids are the mob casting a single ability on every single member of the raid at the same time, every 2 to 4 seconds.

 

Sounds like they need ... drumroll.. more threads!

its really a non excuse she brings up.

It's not as easy as just setting a flag, but assuming (big assumption) that they didn't do anything really stupid it's not that hard either. Like.. Half the players of the world are processed on thread 1 and half are processed on thread 2, and all of a sudden you have less server lag.

I wrote a threadpool last year, so I just throw work at it and whichever thread that is available first takes care of it. DBG could do the exact same thing and everyone would notice a huge network improvement.

If I worked at DBG and didnt want the button smashing simulator every single class is now I would do the following;

Optimize my thread pool a bit and just have every thread throw work at the pool as it popped up, then I would run that on both the server and the client for a nice speedup. Obviously more threads on the client side wouldn't do anything to help your lag, but **** having singlethreaded programs in 2016.

 

If writing a thread pool last year makes you an expert, then you are truly blessed.

The solution will of course be not as trivial as that, because (off the top pf my head) you have to synchronize the numbers you do on the mob, e.g. to tell when the mob is dead, who gets the kill pull, when a damage threshhold is exceeded, and a dozen things more.
Whatever maths are done from any individual spell you cast, their computing time will negligible in comparison to what it takes to synchronize 24+ threads.

This is coming from someone who works with thread pools every week, and does not consider himself an expert on the matter.

 

a 64 bit client would solve a LOT of this

 

No it wouldn't, because combat calculations are done server-side.

 

shhh

 

At no point did I claim to be an expert.

What I was trying to point out is that rather than doing the hard work, they decided to take the easy way out and make auto attack and procs a lesser part of EQ2, after 13 years of auto attacking being the vast majority of our damage together with various procs. Did players ask for this? I personally dont believe so.

What players wanted were lag free raids and instead of investing and optimizing their code they decided, screw it, make EQ2 a button smashing simulator (aka every class is now an illy on crack) and lets get rid of auto attack, BUT, we keep the 9 stats decicated to auto attacks which now does 4-5% of a characters damage.

Caith claiming its the only way to do it is not true and that is what I pointed out.

 

Let's be clear about what Caith was saying when he was discussing auto attack.

He was mocking what I had said earlier in the discord chat, and imitating what someone else said to me at the same time. It probably came off a bit more dismissive than Caith would have said in a more official way.

however, the general direction behind what was said i.e. that autoattack will be fading in priority seems to hold true.

in terms of the stats, it is brutally obvious to me and a few other people that there are some serious statflation issues here with EQ2, and that a gear reset should have happened this expansion. However, i think the developers were seriously overworked and had issues during this expansion development (it's LUCLIN! nope, KUNARK!).

Stat and ability consolidation needs to happen for Everquest 2.

 

Not sure how I got pulled in here, but I've always said 'abilities' should count for more than 'just swinging your sword'. You can see why it's more desirable as well since it puts more focus on button mashing than just sitting there letting autoattack do the work.

 

No, but you made a concrete suggestion on how to fix lag in EQ2, insinuating to know better than the EQ2 devs. But the solution was rubbish to begin with.
But I understand that it was a rant... I often rant too. And I agree that, whatever the cause and solution to the server side lag issues are and were, they were never addressed with the appropriate urgency.

 

Yes, but not going to.

 

out you! out i say!!