The Star Citizen Thread v5

Status
Thread Closed: Not open for further replies.
Sounds like what Dual Universe is doing. Another CR special, see someone else's idea/technology and try to rip it off.

It's certainly interesting tech. Wether it's going to work in a real-time FPS game or not is going to be interesting to find out - especially when some of those players are going to do absolutely everything they can possibly think of benefit themselves, and mess everyone else up.
 
I just watched the latest reverse the verse-special edition and it was kind of interesting to watch CR, Erin Roberts, Tony Zurovec, Brian Chambers, and John Erskine talking about the current and future state of the SC....to cut the chase around 20 min in they start to talk about the networking and how many players we could expect per instance in the "near"future so let me quote CR:

"So it’s really more just about the fact that there will be a limit to how many players we can simulate on one server. Right now it’s 24 in Arena Commander and it’s about 40 when you’re running around Arc Corp. Really, that’s actually 8 instances running on one server. So, if you just times it by 8 and figure it out we could have 200 players if we could scale the game linearly that way but right now we can’t because it’s not – parts of it are built for multithreading and distributing to all of the cores but some aren’t. That’s what we’re doing a lot of refactoring on, so an individual server instance could perhaps run 200 players which then is obviously a lot denser than what we have right now and you would still seamlessly go between locations. But, on top of that if you can mesh those servers together and each server is authoritative over a group of players and generally those players would be based on colocation and each server tells the other servers what it has done with the players it’s responsible for. You can sort of, on a peer-to-peer basis on the server side, you can have thousands of players being simulated all at once but each server is only doing the work for its little portion. That’s the new model, we’re not the only people that are working on this or trying to work on this. It is the new model of using the big – using the cloud to give you a lot more game experience and that’s where we’re going. At some point, you know, we may even – that may allow us to have 2,000 people in the same area or whatever but what if there was maybe 10,000? At that point we probably would still have to instance."

https://www.youtube.com/watch?v=Qyymbmis7xA
20:00 min start

So Wow....Is this really going to work???Well if this work.....then we certainly can expect kind of revolution in the near future gaming...right???Or CR just talking to talk???Or maybe it is the truth that will happens in +few years...or beyond the 2020??

Oh that's good. So If I've got £24 then times it by 8 then figure it out I can have £200. Never realised it was so easy. Key words like maybe, perhaps, may, sort of, trying to, are very apt.
 
I'm confused. CIG have never been balked before at showing a buggy mess, on the grounds that it's an alpha: And things like the floating not-a-Mako, artificially moronic not-Tusken-raiders, and the repeating not-Shai-Hulud suggest they are still cool with that. So just far down the toilet was the S42 demo that they decided not to show it?
 
So just far down the toilet was the S42 demo that they decided not to show it?

The technical problems will take some time to iron out - but that is not the problem. The mix of builds, systems, assets and lack of directional control are what are causing the problems.

X Days to Citizencon - we are nearly ready! We need more fidelity in these bits or it just looks like another rubbish FPS game.

X-1 Days to Citizencon - I want a Sandwormie in there! And I don't just want it wobbling around like this low-fidelity concept you gave me! I want real Sandwormies!
 
The technical problems will take some time to iron out - but that is not the problem. The mix of builds, systems, assets and lack of directional control are what are causing the problems.

I'm willing to bet that this was/is a huge issue. They had a contractor company working for them for two years before they figured out that the two builds each side was working on wouldn't work. What if the same thing happened with SC and SQ42? Is this why all the models are off in height, and the other issues of textures "meshing" together? With 4 studios in all different time zones how do they handle the engine builds? Who's in charge of compiling all that code that is changed or added on a hourly base?

I'd be willing to bet that all these modules were kept in each studio's "house" running off what ever build they had when the marching orders came in. While the studio in charge of changing the engine happily ticked away while the others fell build after build behind. I feel really sorry for the artists who complete their work only to be screw when they try to integrate it in.

Probably right now each artist downloads the most current build first thing in the morning and checks yesterdays work to make sure it's still running...
 
Quite possible. I have no idea what change management they use - but it's quite clear from that video that change is not managed.

One guy quite clearly states that they fixed one particular problem, and in the newest build it's reappeared - and Genuine Roberts is heard complaining about the same issue with a different problem.

It's almost like they are making things up as they go along, and expecting everyone else to know their new development path. Emailing other teams to let them know about important events? Sure - nothing wrong with that. But isn't the entire point of having inter-continental studio development with people available 24/7 (in theory) - that you can just Skype your important meetings so that everyone knows immediately what is going on?
 
It's the " It was more about the awe I felt and the indescribable freedom and looseness that the gameplay seemed to be presenting." guy I feel sorry for.

He may be better off saving his money by ignoring expensive ship jpegs and going for hello.jpg instead :D
 
Oh that's good. So If I've got £24 then times it by 8 then figure it out I can have £200. Never realised it was so easy. Key words like maybe, perhaps, may, sort of, trying to, are very apt.

No it more like this you got 24 little cargo couriers on a 24 lane high way and big part one single huge truck.
Xeon 22 core can handle 44 threads so if you got 44 lanes on the high way and have 44 little courier cargo halers.
You get more done. But the big truck is holding it back.

So the key to efficient and good scaling multithreading is get the seqiential stuf to nice zero for best scaling.
If the truck is 50% and could reduce that to under 10% better 1% then it scales much better with more cores.

In cases where the compute job is pure independed and can be split off in larg amout of small computation . Then scaling is extreem. But in games often this is not the case. There are dependancies. Partly can be reduced with refactoring or better designed multithreading engine. But you will keep some.
 
Well, I always find it funny that people attack CIG for "nice furniture". Please tell me how many programmers would love to work 80+hr weeks in a miserable workplace. The answer is none.

There is, I think, a grey area between "miserable workplace" and "designer furniture with giant artwork murals". Thinking back to my coding days, a comfy chair and a kettle (alternating between coffee and pot-noodles) was enough to pull an all-nighter...
 
There is, I think, a grey area between "miserable workplace" and "designer furniture with giant artwork murals". Thinking back to my coding days, a comfy chair and a kettle (alternating between coffee and pot-noodles) was enough to pull an all-nighter...

You programmers have it so easy :)

Comfy chair and a kettle? Luxury!

Try being stuck in a hardware clean room for a working day - with the nasty lights, constant hvac noise, bootie covers, grounding wires, no chance to go outside on a break, and you have to troubleshoot some pointless PCB against the wrong test data :D
 
People talking, people talking, people talking, same CitizenCon demo(ish but nicer)

I really don't care about the script anymore. I think it's been well established we can't trust what they say - was there a point in that video where we can see the Sq42 demo that I missed? I mean it's been quite a while since the con-show now and they've had time to deal with those few issues that caused them to *just* miss it right?
I understand if someone just has lost 10000 dollars in this and now needs talking heads babbling excuses for 20 minutes. That's all what I get from this material.

I don't have any money in Star Citizen, I don't care at all for those excuses. CIG/RSI/whoever, start delivering something tangible and keep those cult videos inside your circle. Nobody really cares about how a bunch of guys fails at creating a video game for five years.

How about showing something noteworthy, which actually works on a customer PC and is not a blatant ripoff of something else?
 
Last edited:
I just watched the latest reverse the verse-special edition and it was kind of interesting to watch CR, Erin Roberts, Tony Zurovec, Brian Chambers, and John Erskine talking about the current and future state of the SC....to cut the chase around 20 min in they start to talk about the networking and how many players we could expect per instance in the "near"future so let me quote CR:

"So it’s really more just about the fact that there will be a limit to how many players we can simulate on one server. Right now it’s 24 in Arena Commander and it’s about 40 when you’re running around Arc Corp. Really, that’s actually 8 instances running on one server. So, if you just times it by 8 and figure it out we could have 200 players if we could scale the game linearly that way but right now we can’t because it’s not – parts of it are built for multithreading and distributing to all of the cores but some aren’t. That’s what we’re doing a lot of refactoring on, so an individual server instance could perhaps run 200 players which then is obviously a lot denser than what we have right now and you would still seamlessly go between locations. But, on top of that if you can mesh those servers together and each server is authoritative over a group of players and generally those players would be based on colocation and each server tells the other servers what it has done with the players it’s responsible for. You can sort of, on a peer-to-peer basis on the server side, you can have thousands of players being simulated all at once but each server is only doing the work for its little portion. That’s the new model, we’re not the only people that are working on this or trying to work on this. It is the new model of using the big – using the cloud to give you a lot more game experience and that’s where we’re going. At some point, you know, we may even – that may allow us to have 2,000 people in the same area or whatever but what if there was maybe 10,000? At that point we probably would still have to instance."

https://www.youtube.com/watch?v=Qyymbmis7xA
20:00 min start

So Wow....Is this really going to work???Well if this work.....then we certainly can expect kind of revolution in the near future gaming...right???Or CR just talking to talk???Or maybe it is the truth that will happens in +few years...or beyond the 2020??

So basically he's talking about what everyone calls "mega servers" and have been doing for years. Substituting "cloud" servers for their own servers doesn't really change the fact.

It's alright pipedreaming about 2000 people or 10,000 people in one area but it would still need massive amounts of instancing or culling, look at other games using their own server farms like World of Warcraft, on tried and proven tech and far less demanding than SC, they don't put 2000 people in one place, or EVE, again tried and proven and their methods for handling so many players in one place, or GW2 or TSW etc
 
[wacky]
No it more like this you got 24 little cargo couriers on a 24 lane high way and big part one single huge truck.
Xeon 22 core can handle 44 threads so if you got 44 lanes on the high way and have 44 little courier cargo halers.
You get more done. But the big truck is holding it back.

So the key to efficient and good scaling multithreading is get the seqiential stuf to nice zero for best scaling.
If the truck is 50% and could reduce that to under 10% better 1% then it scales much better with more cores.

In cases where the compute job is pure independed and can be split off in larg amout of small computation . Then scaling is extreem. But in games often this is not the case. There are dependancies. Partly can be reduced with refactoring or better designed multithreading engine. But you will keep some.

Well why didn't he say that?[wacky]
 
Last edited:
You programmers have it so easy :)

Comfy chair and a kettle? Luxury!

Try being stuck in a hardware clean room for a working day - with the nasty lights, constant hvac noise, bootie covers, grounding wires, no chance to go outside on a break, and you have to troubleshoot some pointless PCB against the wrong test data :D

Is "no, thanks" an option? :D
 
I take it back. What was quoted does not match up with what Dual Universe is doing. DU is doing something incredibly advanced. The quote with the "times it by 8" I just can't figure out exactly what is being said there. There is no basis for why 24 or 40 suddenly became 8 instances and then magically was multiplied by 8. I'm only going based on the words there, I was having fun in other games that work when I read that but either something was missed in the transcribing or something was completely ignored in the mind of the person that said that text.
 
Status
Thread Closed: Not open for further replies.
Back
Top Bottom