Author Topic: Star Citizen Breaking News  (Read 281387 times)

dsmart

  • Supreme Cmdr
  • Administrator
  • Hero Member
  • *****
  • Posts: 4915
    • Smart Speak Blog
Re: Star Citizen - Breaking News
« Reply #120 on: December 12, 2017, 06:44:17 AM »
Did Clive Johnson just said "We have the wrong engine for SC" ? Am I right?

It depends on the interpretation; but that's how I read it. Claiming that the engine is being pushed to do something it wasn't designed to do, is the closest he came to admitting that.
Star Citizen isn't a game. It's a TV show about a bunch of characters making a game. It's basically "This is Spinal Tap" - except people think the band is real.

N0mad

  • Hero Member
  • *****
  • Posts: 597
Re: Star Citizen - Breaking News
« Reply #121 on: December 12, 2017, 10:43:19 AM »
Regarding Clive Johnson's post about the netcode, I thought I'd throw my opinion out there. I know very little about networking so pay more attention to Derek's excellent blog on this than me.

That said, Clive makes some very contradictory statements. Firstly saying the Netcode doesn't impact FPS and performace, but then saying that fresh servers and offline games have better performance because they have fewer players.

I think it all depends on how you define Netcode, which as Derek says, is a very non specific term. Clive may be referring solely to the mechanism by which data gets serialized and transmitted between the client and the server, perhaps also relating to how that data gets dealt with in the core code at either end. What this doesn't address, of course, is HOW much data is being transmitted, which as we know is an excessive amount and base Cryengine is extremely poorly optimized in this regard. Clive's sole responsibility may be the code which deals with the data flow, and, as he points out, this isn't the problem. When he goes on to talk about fundamentally changing systems in the engine in his last paragraph, he may well be referring to the entire multiplayer system in Cryengine and how every single asset in the game handles multiplayer, including physics updates etc. Which kind of suggests that they haven't even been looking at the multiplayer code in Cryengine (it may be outside his remit anyway).

Anyway, this was my take on the post. I'd love to hear back from any network programmers out there !

Scruffpuff

  • Jr. Member
  • **
  • Posts: 67
Re: Star Citizen - Breaking News
« Reply #122 on: December 12, 2017, 01:32:11 PM »
Regarding Clive Johnson's post about the netcode, I thought I'd throw my opinion out there. I know very little about networking so pay more attention to Derek's excellent blog on this than me.

That said, Clive makes some very contradictory statements. Firstly saying the Netcode doesn't impact FPS and performace, but then saying that fresh servers and offline games have better performance because they have fewer players.

I think it all depends on how you define Netcode, which as Derek says, is a very non specific term. Clive may be referring solely to the mechanism by which data gets serialized and transmitted between the client and the server, perhaps also relating to how that data gets dealt with in the core code at either end. What this doesn't address, of course, is HOW much data is being transmitted, which as we know is an excessive amount and base Cryengine is extremely poorly optimized in this regard. Clive's sole responsibility may be the code which deals with the data flow, and, as he points out, this isn't the problem. When he goes on to talk about fundamentally changing systems in the engine in his last paragraph, he may well be referring to the entire multiplayer system in Cryengine and how every single asset in the game handles multiplayer, including physics updates etc. Which kind of suggests that they haven't even been looking at the multiplayer code in Cryengine (it may be outside his remit anyway).

Anyway, this was my take on the post. I'd love to hear back from any network programmers out there !

Although the technology can be complicated to implement and requires experience and expertise, the basic concept of optimization with relation to network communication is in fact quite simple, and easily alluded to through the following shorthand:  What is the least information I can provide that will still get across the important data?

Consider a game of Battleship.  You have a carrier, which takes up 5 slots in a row, in positions E2, E3, E4, E5, and E6.  What is the least amount of information to allow someone who does not see your board to replicate your ship's position?  You could give the ship type along with 5 coordinates.  You could cut that down by more than half by just sending the ship type, E2, and E6, because that confers enough data to fill in the blanks.  You could just give E2 and E6 and skip the ship type, since there is only one ship that takes up 5 spaces, slimming down the transmitted data even more.

Network communication in gaming is the same kind of thing.  You have to plan for it from the very beginning to make significant gains in efficiency.  A good example of that is my reference above - since both clients know the carrier is the only ship that takes up 5 spaces, then it becomes valid communication to simply transmit the two endpoint coordinates.  If the clients don't understand this rule, you would have to send the extra data (the ship type.)

This is where Chris Roberts obsession with his rendition of "fidelity" dooms him.  Take World of Warcraft for an example.  If two players are playing together and kill a crocolisk, the only data sent to each client is that there is a dead crocolisk there.  That's it - a number in a database representing a dead crocolisk.  (Let's keep the "lootable" aspect out for this example.)  Now it's the client's job to decide which of several dead crocolisk art pieces to render on the screen.  Maybe it's a red one dead on its stomach.  Maybe it's an albino one flipped onto its back.  Maybe one's facing north and the other east - decided on by the client.  It doesn't matter, it's not important, just send the minimum data to get the point across.

This is not good enough for Chris.  In his mind, The Matrix was not a movie but a documentary, and everything seen by one client must be seen, identically, by every client at exactly the same time.  Each raindrop must be individually rendered, not just a "rain" event, and each raindrop's position sent over the network in real time.  Each laser blast, each eye blink of your commando, each ripple of your many jackets as you stride into a malfunctioning airlock, it ALL must be broadcast back and forth in huge networking datasets so that everyone sees the same thing.

That approach would fail in a single-player client-server game.  Attempting to do it in an MMO?  Insanity.

nightfire

  • Full Member
  • ***
  • Posts: 212
Re: Star Citizen - Breaking News
« Reply #123 on: December 12, 2017, 02:49:18 PM »
Each raindrop must be individually rendered, not just a "rain" event, and each raindrop's position sent over the network in real time.

See? He's doing everything right like you proposed in your Battleship example – he's just sending the raindrop's position.

Someone who didn't understand game development would send the size, 3d volume geometry (influenced by air resistance), and dust particle contamination (which determines reflective and translucent properties) of that particular raindrop over the network as well. Chris doesn't do that; he just sends the position of each drop. How efficient is that!

That's because Chris and his team understand Serialized Raindrops, Raindrop Culling and Water 2.0 to make this miracle of a game work with rainfall fidelity never seen in a game before.

Scruffpuff

  • Jr. Member
  • **
  • Posts: 67
Re: Star Citizen - Breaking News
« Reply #124 on: December 12, 2017, 07:14:37 PM »
Each raindrop must be individually rendered, not just a "rain" event, and each raindrop's position sent over the network in real time.

See? He's doing everything right like you proposed in your Battleship example – he's just sending the raindrop's position.

Someone who didn't understand game development would send the size, 3d volume geometry (influenced by air resistance), and dust particle contamination (which determines reflective and translucent properties) of that particular raindrop over the network as well. Chris doesn't do that; he just sends the position of each drop. How efficient is that!

That's because Chris and his team understand Serialized Raindrops, Raindrop Culling and Water 2.0 to make this miracle of a game work with rainfall fidelity never seen in a game before.

You're right!  And once he leverages the power in the machines, there's no limit to what else he can serialize.

StanTheMan

  • Hero Member
  • *****
  • Posts: 676
Re: Star Citizen - Breaking News
« Reply #125 on: December 12, 2017, 08:20:46 PM »
Quote
there's no limit to what else he can serialize.

We will be getting a Star Citizen promotion on the back of Kellogg's boxes next.
« Last Edit: December 13, 2017, 05:05:40 PM by StanTheMan »

dsmart

  • Supreme Cmdr
  • Administrator
  • Hero Member
  • *****
  • Posts: 4915
    • Smart Speak Blog
Re: Star Citizen - Breaking News
« Reply #126 on: December 13, 2017, 06:21:26 AM »
Each raindrop must be individually rendered, not just a "rain" event, and each raindrop's position sent over the network in real time.

See? He's doing everything right like you proposed in your Battleship example – he's just sending the raindrop's position.

That Battleship example is an excellent one. The hilarious part of that is when you compare a game like to one like Star Citizen the problem increases exponentially.

The other part of this is that, as I wrote in my latest article, for the longest time, they have been using these buzz terms in order to make excuses for the shitty networking layer. I wrote - many times - that those sort of things are part of most (if not all) networking engines but that at the end of the day, it's not going to make much difference - and that they're never going to get an MMO anyway.

Designing a networking layer for an MMO takes a LOT of work, and even when you use third-party engines (I have used ReplicaNet and RakNet in my multiplayer games) to do all the heavy lifting (much like how you would use middleware for audio, graphics, AI, UI etc), your game design and code still needs to be robust enough because at the end of the day, you have to massage your data to make the most efficient use of the networking middleware. If like Star Citizen you're sending in excess of 4K data packets from each client, with the server doing upwards of 170K, no middleware is going to help you.

So in their case, network culling, serialized variables etc, aren't going to solve the problem to any large extent. In fact, from what I have seen of their transmission packets, I would bet that they're probably going to get 1 maybe 2 additional clients worth of performance. Yes, it's that bad. 

In case you missed it, this is what I wrote back in May

As I've written before, our networking engine for Line Of Defense which uses RakNet, took us the better part of 2 years to do. And it was designed from the ground up to support either a standard session based game, or an MMO. Why? Because I wasn't sure that by the time the game was finished, that MMOs were still going to be a trend or worth the on-going costs and hassles. So I planned ahead. Especially due to the fact that we also needed support for consoles for which doing an MMO was never in the plan. Basically, what they are trying to do at the point in the game's development, is the worse thing possible. And that's why they are never going to get a reasonable experience out of it.
Star Citizen isn't a game. It's a TV show about a bunch of characters making a game. It's basically "This is Spinal Tap" - except people think the band is real.

dsmart

  • Supreme Cmdr
  • Administrator
  • Hero Member
  • *****
  • Posts: 4915
    • Smart Speak Blog
Re: Star Citizen - Breaking News
« Reply #127 on: December 13, 2017, 07:15:41 AM »
BREAKING DEVELOPMENT NEWS STAY TUNED
Star Citizen isn't a game. It's a TV show about a bunch of characters making a game. It's basically "This is Spinal Tap" - except people think the band is real.

the_wolfmann

  • Jr. Member
  • **
  • Posts: 55
Re: Star Citizen - Breaking News
« Reply #128 on: December 13, 2017, 07:22:00 AM »
BREAKING DEVELOPMENT NEWS STAY TUNED

:f5:

justme

  • Newbie
  • *
  • Posts: 30
Re: Star Citizen - Breaking News
« Reply #129 on: December 13, 2017, 07:33:04 AM »
just my thoughts.

the latest shows, relating data optimization always showed us ways, to reduce the information for clients,
but only by cutting of the parts, far away.
of course i don't need to know, if somebody takes out landing gear, during my shopping tour at levski.

but what about these things, important for a mmo? i mean, that thing should be a mmo, not a single player.
i want to play with friends, flying with multi crew ships and doing stuff. so what about optimization of information
from my party members. no words, yet.


and if we are looking at the possibilities with all theses ships, very specialized, there will not be THAT ship you need,
you always would need some more ships as a fleet.
take the polaris as example. at the moment it needs a crew of 14 players. but you need some support ships. so at the
end, you will need about 20 players minimum in a very short range. all these players won't just fly from a to b. they wanna
do a mission, with even more players or more npc. and they will use turrets, missles and torpedos. so many many more information
for all these players in a small area.

i did not heard anything, how they wanna handle this.

but what i hear are visions of hugh fleets with hundreds of ships.

this is just a technical thing. i really think, with the right people you can handle that. i don't know, i hope.
but technical problems will always be solved one day.


but there are more concerns, nobody really speaks about.
everything should be hugh and massive. 90% npc, gigantic systems, whatever.


just believe one moment, that they build the perfect game,
like every backer always wanted.
what about the player base?

the conversative mmo players stagnate. even the big games have to handle with
less and less players. these mobas needed to be hidden in these speculations...
why should there be such a hugh amount of players for a space sim mmo?

i played some mmo, all the famous ones. even there, with many willing players,
it was hard to get 16people motivated to join the game at a special time for a specific task.
raids with more players needed, was a hugh amount of work to get them coordinated. it took
weeks to find 50+ players.
and the most are just not willing to bind there freetime for other players.
these days from vanilla wow are over.

yes there are organizations in the game with thousands of players, all around the world.
but do they really think, that everybody from europe will take part in the US peaks?
and even the players from one country combined to one single organization will have problems,
to finde enough willing and skilled players to do that specific missions, maybe offered one day.


so it is totally wrong to take that as reference not to build a game,
but you should ask yourself, what the consequenses would be and how
to counteract this in a smart way.

at the end i'm lucky if we can reach 4million players around the globe, playing a finished mmo.
but this would mean about 500-1000 players in a timezone. and now devide that number to all
the systems we expect. so the verse should be pretty empty.



but hey. they first need to publish a stable build. just one worked fine for me,
all the other... let us talk about another topic.



« Last Edit: December 13, 2017, 07:42:15 AM by justme »

Bubba

  • Jr. Member
  • **
  • Posts: 90
Re: Star Citizen - Breaking News
« Reply #130 on: December 13, 2017, 07:33:40 AM »
Derek, your latest rant tried to do three things at once, and it might have been better separated.
In particular, by descending into Clive's post in detail (task two), the major point got a little muffled: the fact that, in fact, Clive's post is truthful. This isn't where the problem lies, nor are they going to get earth-shattering improvements here. Netcode works as planned and isn't holding anything up at the moment (sure, you can call it crappy and lousy and all you want, and you may be right, but even if it were the best in the world, it wouldn't matter). Rather, the problem is that, for so long, RSI has been making netcode the bad guy rather than incorporating multiplayer constraints as something to be incorporated into development and budgeted for  (task one).
So the gamer thinks "it's buggy netcode", whereas we look at it and say, "you can't just drop something that looks really good as a single-player simulation and expect it to run across a thousand partially-synchronized clients. Hell, you can't even expect it to run with 16 players."
Take the Battleship example and throw it out. The design question isn't "I have this game -- how do I transmit the information most efficiently?" but rather, "I have these constraints: client performance, network performance, talent, capital. How do I build the BDSSE?"
In the case at hand, the main question seems to have been: "How do I shape development to generate the most revenue?"
The Australian article is spot on. Meta is the best this game will ever get.

Oh, your third point was to say to the developers suffering in a situation worse than deserved: "This is how you should redesign your game, which you'll never get to do, because you're all going under quickly." That's going all Tokyo Rose on the poor bastards. The only way you coulda made it worse was by asking who was watching their spouses while they wasted their most productive and healthy years crunching away at the office.

Kyrt

  • Full Member
  • ***
  • Posts: 176
Re: Star Citizen - Breaking News
« Reply #131 on: December 13, 2017, 07:54:24 AM »
In particular, by descending into Clive's post in detail (task two), the major point got a little muffled: the fact that, in fact, Clive's post is truthful. This isn't where the problem lies, nor are they going to get earth-shattering improvements here. Netcode works as planned and isn't holding anything up at the moment (sure, you can call it crappy and lousy and all you want, and you may be right, but even if it were the best in the world, it wouldn't matter).

Well....as I understand Clive Johnsons point, he is saying that the netcode isn't at fault, but the reason the game has performance issues is because the client is being overloaded with data from 39 other players.

One question that would pop into my head at this point is why each client is dealing with the data from 39 other players.

This, I think, is why so many people were blaming the netcode. It is easy to point to an aspect such as culling and say performance is suffering because the client is getting too much needless data...that there is no prioritisation going on

And...to a degree...this would be correct. But I doubt the answers to the performance issues are that simple. Data culling would remove needless data from the system but it could still cause problems when you get 40 players in the same area.

dsmart

  • Supreme Cmdr
  • Administrator
  • Hero Member
  • *****
  • Posts: 4915
    • Smart Speak Blog
Re: Star Citizen - Breaking News
« Reply #132 on: December 13, 2017, 08:00:59 AM »
BREAKING DEVELOPMENT NEWS STAY TUNED

https://twitter.com/dsmart/status/940959144219152384

Quote
Star Citizen backers now have a choice to make. 1) Dig deep & continue to fund a failed venture 2) ask questions about the money or 3) pull out and get a refund.

What choice will you make?

It's here. It's bad. And it's only going to get worse.

Stay tuned.
Star Citizen isn't a game. It's a TV show about a bunch of characters making a game. It's basically "This is Spinal Tap" - except people think the band is real.

dsmart

  • Supreme Cmdr
  • Administrator
  • Hero Member
  • *****
  • Posts: 4915
    • Smart Speak Blog
Re: Star Citizen - Breaking News
« Reply #133 on: December 13, 2017, 08:04:55 AM »
In particular, by descending into Clive's post in detail (task two), the major point got a little muffled: the fact that, in fact, Clive's post is truthful. This isn't where the problem lies, nor are they going to get earth-shattering improvements here. Netcode works as planned and isn't holding anything up at the moment (sure, you can call it crappy and lousy and all you want, and you may be right, but even if it were the best in the world, it wouldn't matter).

Well....as I understand Clive Johnsons point, he is saying that the netcode isn't at fault, but the reason the game has performance issues is because the client is being overloaded with data from 39 other players.

One question that would pop into my head at this point is why each client is dealing with the data from 39 other players.

This, I think, is why so many people were blaming the netcode. It is easy to point to an aspect such as culling and say performance is suffering because the client is getting too much needless data...that there is no prioritisation going on

And...to a degree...this would be correct. But I doubt the answers to the performance issues are that simple. Data culling would remove needless data from the system but it could still cause problems when you get 40 players in the same area.

Precisely. And so far, the majority of people are saying that this was my most clear and understandable blog article yet.

I separated Clive's comments into parts, and broke it down into layman terms for easy consumption because most gamers (esp the Star Citizen backer target) won't understand any of it otherwise.
Star Citizen isn't a game. It's a TV show about a bunch of characters making a game. It's basically "This is Spinal Tap" - except people think the band is real.

Motto

  • Hero Member
  • *****
  • Posts: 1023
Re: Star Citizen - Breaking News
« Reply #134 on: December 13, 2017, 08:05:20 AM »
Dear whales, despite your 170 million funding so far, we forgot to pay some licenses. Kindly and quickly fork up more money to pay for those licenses, and then some more so we can continue to scale Star Citizen further down to a "one player per server" model in order to achieve the promised highest level of fidelity.

 

SMF spam blocked by CleanTalk