Jump to content

It's Not The Netcode, It's The Latency.


103 replies to this topic

#41 Taizan

    Com Guard

  • PipPipPipPipPipPipPipPip
  • 1,692 posts
  • LocationGalatea (NRW)

Posted 18 January 2013 - 08:50 AM

View Postzmeul, on 18 January 2013 - 08:44 AM, said:

it's close to 2 months from the moment they even publicly admitted the issue
also they say it's their no1 priority, yeah right :P

Two months really isn't long to completely rewrite netcode for a server / client based shooter with server side authentication and server/client side prediction, test it and make sure it doesn't break anything else..

About Netcode being top priority. Here you go.

#42 FringeAggressor

    Member

  • PipPipPip
  • 63 posts

Posted 18 January 2013 - 08:51 AM

If it wasn't the netcode, PGI wouldn't have speed-capped mechs capable of builds in excess of 130kph, because players with good latency would remain unaffected.

/thread

#43 zmeul

    Member

  • PipPipPipPipPipPip
  • 397 posts
  • LocationBuzau, Romania

Posted 18 January 2013 - 09:08 AM

View PostTaizan, on 18 January 2013 - 08:50 AM, said:

Two months really isn't long to completely rewrite netcode for a server / client

2 months is the time frame they publicly spoke about it, not the time frame the issue existed
as for what Paul said, do you actually want me to believe him?! while they spiting out premium mechs and new chassis like crazy

they sure have at least twice as many people invested in other areas of the game than they have for netcode - this is not assigning top priority to netcode!

#44 CygnusX7

    Member

  • PipPipPipPipPipPipPipPip
  • 1,803 posts
  • LocationA desolate moon circling a desolate planet

Posted 18 January 2013 - 09:14 AM

Its mostly about stability but if the hit boxes are ahead of a mech depending on latency then something is wrong with the net code. Are hit boxes not part of the equation and only the mech translated to different position? Odd for sure.

Edited by CygnusX7, 18 January 2013 - 09:27 AM.


#45 Kaspirikay

    Member

  • PipPipPipPipPipPipPipPipPip
  • Giant Helper
  • 2,050 posts

Posted 18 January 2013 - 09:18 AM

With a ping of 280, i'll still get killed by lasers of players whose ping is less than 50.

#46 Gaan Cathal

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 2,108 posts

Posted 18 January 2013 - 09:21 AM

View Postzmeul, on 18 January 2013 - 09:08 AM, said:

they sure have at least twice as many people invested in other areas of the game than they have for netcode - this is not assigning top priority to netcode!


You're right, all those artists should get on helping with the netcode. With all that in-depth knowledge of the Cryengine3 code that they have. Because they're computer artists, right?

#47 Darkmoose

    Member

  • PipPipPipPipPipPip
  • 210 posts
  • Google+: Link
  • LocationSTL MO

Posted 18 January 2013 - 09:24 AM

Would more local servers fix this, an NA Server, an Asia Server, and a Europe Server.

#48 zmeul

    Member

  • PipPipPipPipPipPip
  • 397 posts
  • LocationBuzau, Romania

Posted 18 January 2013 - 09:27 AM

View PostGaan Cathal, on 18 January 2013 - 09:21 AM, said:

You're right, all those artists should get on helping with the netcode.

artists ..
do you even understand concept of programming? do you actually believe artists are the only people involved in creating a new chassis?
LOL .. just lol

when Paul said what he said, I now understand why a lot of people actually believed him
this is sad

Edited by zmeul, 18 January 2013 - 09:29 AM.


#49 Gaan Cathal

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 2,108 posts

Posted 18 January 2013 - 09:36 AM

View Postzmeul, on 18 January 2013 - 09:27 AM, said:

artists ..
do you even understand concept of programming? do you actually believe artists are the only people involved in creating a new chassis?
LOL .. just lol

when Paul said what he said, I now understand why a lot of people actually believed him
this is sad


It was a slightly facetious example, I'll grant, but throwing every programmer they have at the problem won't solve it any faster. There will be duplication of work if nothing else, which is inefficient, and if they stop putting out other content then the forums will be just as full of bile and vitriol about the lack of maps/mechs/weapon balance as they are currently about the netcode (and the lack of maps/mechs/weapon balance), so even if throwing every coder at the problem had a marginal effect on time-to-completion, if would be a stupid decision.

They will have a few coders who are assigned to the netcode issue, who will be in regular communication with Crytek on the issue, since the issue is with a Crytek product that PGI are licencing. That's how this programming works on a corporate scale, the same way everything else does.The odds are, they have a support agreement with Crytek as part of the licencing deal such that a large portion of the work on netcode is joint.

#50 Bishop Steiner

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • The Hammer
  • The Hammer
  • 47,187 posts
  • Locationclimbing Mt Tryhard, one smoldering Meta-Mech corpse at a time

Posted 18 January 2013 - 09:39 AM

View PostGaan Cathal, on 18 January 2013 - 09:36 AM, said:


It was a slightly facetious example, I'll grant, but throwing every programmer they have at the problem won't solve it any faster. There will be duplication of work if nothing else, which is inefficient, and if they stop putting out other content then the forums will be just as full of bile and vitriol about the lack of maps/mechs/weapon balance as they are currently about the netcode (and the lack of maps/mechs/weapon balance), so even if throwing every coder at the problem had a marginal effect on time-to-completion, if would be a stupid decision.

They will have a few coders who are assigned to the netcode issue, who will be in regular communication with Crytek on the issue, since the issue is with a Crytek product that PGI are licencing. That's how this programming works on a corporate scale, the same way everything else does.The odds are, they have a support agreement with Crytek as part of the licencing deal such that a large portion of the work on netcode is joint.


stop....

making....

sense.

It only confuses and angers the QQ brigade.

#51 zmeul

    Member

  • PipPipPipPipPipPip
  • 397 posts
  • LocationBuzau, Romania

Posted 18 January 2013 - 09:49 AM

View PostGaan Cathal, on 18 January 2013 - 09:36 AM, said:

There will be duplication of work if nothing else, which is inefficient

I'm sorry, why do you assume they work like monkeys and they do not divide the workload to groups - isn' that what efficient programming teams do?
should I also assume you do not believe them to be efficient ?!?!?!!? :P

programming has various steps to complete, not all steps have to handled by specialized network coders

back in CBT, when someone asked why the lack of new mech chassis, PGI said they need about 1month to create a new one from scratch; but if you look at the roadmap, that time has been shorten - how? they got more people assigned to this

why not same priority to netcode?

Edited by zmeul, 18 January 2013 - 09:49 AM.


#52 KyBr2K

    Member

  • PipPipPip
  • The Covert
  • The Covert
  • 87 posts
  • LocationCockpit of a DWF-B

Posted 18 January 2013 - 09:56 AM

OI! Why on earth would you use Norway as an example? We actually get "decent" ping to Canada for crying out loud! 125ms isn't great, but 60% of the other people I see get ~300 for Valhalla's sake..

#53 Gaan Cathal

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 2,108 posts

Posted 18 January 2013 - 10:07 AM

View Postzmeul, on 18 January 2013 - 09:49 AM, said:

I'm sorry, why do you assume they work like monkeys and they do not divide the workload to groups - isn' that what efficient programming teams do?
should I also assume you do not believe them to be efficient ?!?!?!!? :P


That's exactly what I'm saying they do do.One group is currently working on the netcode, others on maps, mechs, weapon tweaks, fixing whatever got broke in the last test build.

View Postzmeul, on 18 January 2013 - 09:49 AM, said:

programming has various steps to complete, not all steps have to handled by specialized network coders


There's a bottleneck though, that non-specialist work can only be completed at a rate determined by the rate at which the specialist work progresses. There's no point having a non-specialist sitting around waiting for the netcode guy to get done, so he is put to work on coding mech chassis, or new and interesting ways to make night vision appalling.


View Postzmeul, on 18 January 2013 - 09:49 AM, said:

[color=#959595]why not same priority to netcode?[/color]


See, the Devs have said that netcode is no.1 priority. If that is the case, that means that they have added enough members to the team that increasing it further won't increase the rate of progress (at this juncture). Unless you have evidence that they've not done this, beyond the fact that the completion date isn't 'yesterday' then I fail to see any basis for accusing them of lying.

#54 Soda Popinsky

    Member

  • PipPipPipPipPip
  • 169 posts

Posted 18 January 2013 - 10:27 AM

From what I've gathered, with the tape roll back system, you fire and send that packet to the server. The server "rolls back tape" to when you actually fired (based on your ping), and determines if there was a target there.

Is that all there is too it? Because if that's true, you will still be firing at a "ghost" since the target will be rendered ping time after what the server sent you. For example, if you have a 250 ms latency, you fire at a mech that was at a spot 250 ms ago according to the server.

You send that fire signal to the server, taking an additional 250ms (500ms total round trip) which rolls it back, 250ms. So now you're at 250ms behind still. I guess it's better than 500ms behind. But shouldn't that still require leading the target.


Unlike most FPS, where one can zig-zag with the mouse, the mechs tend to move in a constant straight line, or arc. I wonder if that is used in anyway in the netcode to optimize hits, as it's much more predictable than the zig-zagging you can do in other games. Of course that doesn't help with the torso twisting, which can throw off any per-polygon hit detection (assuming there is per polygon hit detection).



I don't know how applicable this is, but I mainly play a niche WWII FPS, called "Red Orchestra 2." Due to the high proportion of overseas players (they love it in Russia), netcode was a real issue. A player in the community took it upon himself to write up a client-side hit detection mutator (Unreal Engine), that was such a hit, the Devs made it official. They use VAC/Punkbuster, so cheating hasn't been an issue (plus the smaller player base tends to be more mature, and less cheaty). That may be a way to go, assuming cheating can be handled.

#55 ImagineDwagons

    Member

  • PipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 81 posts
  • LocationKiev

Posted 18 January 2013 - 10:42 AM

View PostFergrim, on 18 January 2013 - 05:49 AM, said:

In a usual game that I join, I see at least a third of the latencies in the range of 180-200. Lag shielded mechs are not shielded by the netcode, they are shielded by the fact that their pilot is in freaking Norway!

I have 140 ping average and before 8th of January patch I was able to hit a light mech with unguided SRM on my light mech running 150 km/h speed. After 8th of January patch I can't hit a **** i can't even lock LRM on moving target normally due to huge responce time and warping mechs around me.

#56 zmeul

    Member

  • PipPipPipPipPipPip
  • 397 posts
  • LocationBuzau, Romania

Posted 18 January 2013 - 10:59 AM

View PostGaan Cathal, on 18 January 2013 - 10:07 AM, said:

the Devs have said that netcode is no.1 priority

and yet we have other new stuff in each patch, except netcode
just look at the patch notes since OBT - a lot of stuff added, very few of the issues fixed


as Soda Popinsky said, their "roll back the tape" to the moment of the shot - it's bogus! because the "shielded" mech will still have the lead compared to the moment of the shot; he explained it better

unless they artificially add same amount of lag to everyone in the match as the highest ping player, I do not see how they can "predict" movement
how this works: example, if the highest ping player is @200ms, everyone in the match will have their actions delayed to 200ms
it is similar in concept with what CCP did for EVE with "time dilation"

Edited by zmeul, 18 January 2013 - 11:07 AM.


#57 ReD3y3

    Member

  • PipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 480 posts

Posted 18 January 2013 - 11:04 AM

Well the lag is worse since the most recent patch.

There was a time when the hit detection was pretty damn good.

Right before the Cicada patch I could hit what I was shooting at.

So it WAS working. They broke it somehow.

#58 silentD11

    Member

  • PipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 816 posts
  • LocationWashington DC

Posted 18 January 2013 - 11:07 AM

View PostKyBr2K, on 18 January 2013 - 09:56 AM, said:

OI! Why on earth would you use Norway as an example? We actually get "decent" ping to Canada for crying out loud! 125ms isn't great, but 60% of the other people I see get ~300 for Valhalla's sake..


I think different people have different ideas of what a "good" ping is. I ping sub 40 from home with a single person on the LAN, pretty much the same with the usual 2-3 and it's a fairly different experience than playing at 100+. Stuff is still off, stuff still warps, but it's a lot less dramatic. Generally when it is dramatic it's because someone has a really high ping. One of the guys I play with a lot is from Aus and his ping is always stratospheric.

The netcode is certainly problematic and does need to be fixed. But it isn't as if people were deliberately lagging their flag runners in games like Quake and UT to make sure they had lag shields even back then and low pingers have always had an easier time making their shots.

#59 CancR

    Member

  • PipPipPipPipPipPipPip
  • 766 posts

Posted 18 January 2013 - 11:33 AM

It's also a deep lack of skill. People are either not leading at all or leading way to far ahead and then blaming lag shield rather then themselves. As a light that is hitting other lights and killing them using nothing but lasers (6 MLs on a jenner f) it does take a bit of skill to learn where to aim the shots but you don't learn this by complaining, you learn by practice.

#60 ReD3y3

    Member

  • PipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 480 posts

Posted 18 January 2013 - 11:45 AM

View PostCancR, on 18 January 2013 - 11:33 AM, said:

It's also a deep lack of skill. People are either not leading at all or leading way to far ahead and then blaming lag shield rather then themselves. As a light that is hitting other lights and killing them using nothing but lasers (6 MLs on a jenner f) it does take a bit of skill to learn where to aim the shots but you don't learn this by complaining, you learn by practice.


You should not have to lead at all with lasers. They should put damage on the section of the mech you are hitting.

That is not a question of skill.

I can lag shoot too, but it is HIGHLY in accurate. If a mech is without armor in his left torso, and I want to hit that exact part of the mech guess what, I dont shoot the left torso, I shoot barely in front of his right torso ( if he is facing me and running left to right). If i want to shoot his right torso I have to shoot even farther in front of him. But it is in accurate. VERY INACCURATE.





5 user(s) are reading this topic

0 members, 5 guests, 0 anonymous users