Vapourware Scam Citizen - Only people with too much money can become StarCitizens! WOULD YOU LIKE TO KNOW MORE?

[Perkel]

probably means Vulcan is ready to be deployed in 3.23

A Lard-Ass has been been saying the same sort of shit since 2021, I'm sure that as long as you keep at it, one of these years you'll get it right.

Just read some info from one of devs. Vulcan won't bring CPU improvements for now. Currently the engine is split into a game thread and a rendering thread. They already "fixed" the rendering thread as it is fully multicore and it is a half of game thread in miliseconds. They are currently working to make the game thread fully multicore which should come after Vulcan implementation.

Vulcan also is not something that will come in 3.23. Apparently GI, DLSS etc. don't need Vulcan as it is part of their GEN12 renderer that works regardless of it being vulcan or not.

GI DLSS FSR etc. are bound for 3.23 but not vulcan.

 

[Perkel]

They are finally testing end goal for server meshing aka multiple servers per system

The current limit is around 200 players per server per system (normally they go for 100 people which gives you about 7fps server tick). It is low amount mostly because server has to have in memory whole system and calculate stuff like pathfinding for whole system at the same time. Not only does it mean it has to be specced very high with gigatons of ram (which is costly for CIG) but also that the more sprawled players are in system rather than in one place the harder is for server to calculate stuff.

It is an exponential curve. The worst situation for server is when all players are sparsely populating server rather than clumping together.

Which means that if they implement finally server meshing and each location will get their own server it means that each server will be able to hold much more people than 200. Considering exponential curve i wouldn't be surprised if very small location like city could hold 1000s of players on single server.

 

[Baron Tahn]

Exponential curve? Buh?

Surely if more players are on a server it means more latency, regardless.

If its going to act faster in a city that still makes sense, because you arent actually tracking what the game is meant to be doing - ie keeping tabs on where all the ships are and space combat all the time.

It needs to be fast for as many players as possible in the combat, not the shop hub that games seem so fond of trying to make us stay in these days.

Given that the game 'will' release with dozens of star systems it makes sense that players would still be spread out so might work, but Im still not seeing how 100's of crew members in an idris + fleet of constellations or whatever is going to work.

 

[Myobi]

i wouldn't be surprised if very small location like city could hold 1000s of players on single server.

Your PC already struggles with an empty city.

 

[Hag]

They are finally testing end goal for server meshing aka multiple servers per system

The current limit is around 200 players per server per system (normally they go for 100 people which gives you about 7fps server tick). It is low amount mostly because server has to have in memory whole system and calculate stuff like pathfinding for whole system at the same time. Not only does it mean it has to be specced very high with gigatons of ram (which is costly for CIG) but also that the more sprawled players are in system rather than in one place the harder is for server to calculate stuff.

It is an exponential curve. The worst situation for server is when all players are sparsely populating server rather than clumping together.

Which means that if they implement finally server meshing and each location will get their own server it means that each server will be able to hold much more people than 200. Considering exponential curve i wouldn't be surprised if very small location like city could hold 1000s of players on single server.

That's just a big bag of "if", maybe" and "hopefully". The community really must be desperate to cheer such technical talk instead of actual results.

 

[Virgil Brummond]

Which means that if they implement finally server meshing and each location will get their own server it means that each server will be able to hold much more people than 200. Considering exponential curve i wouldn't be surprised if very small location like city could hold 1000s of players on single server.

This is some hardcore cope. Although points for the "exponential curve" BS; this is not part of the standard starter pack for server meshing shilling.

 

[Perkel]

That's just a big bag of "if", maybe" and "hopefully". The community really must be desperate to cheer such technical talk instead of actual results.

I mean they are finally reaching the goal that was years in production.

Moreover it is not just StarCitizen goal but kind of big deal for MMOs in general. If their technology will work the barrier between single player design and multiplayer design will be effectively removed.

The issue with MMOs is that there are hard limits to design every developer has to deal with that come from networking side which is why most of MMOs look like retarded brothers of their single player brothers. It's not like developers don't want to make good games, they just can't as there are hard limits to what they can do.

This is why almost every kickstarter MMO fails. Because people who have ton of ambition don't know steel wall exist which limits their ability to deliver those ambitions. Literally something CIG themselves realized. But unlike other developers they had Chris Roberts dude who can't say no to scope and ton of money with 0 strings attached to try punch through it.

This is some hardcore cope. Although points for the "exponential curve" BS; this is not part of the standard starter pack for server meshing shilling.

It is not BS. It is far easier to calculate 1000 npcs standing in same room that use shared pathfinding area vs each of them standing in complete separate parts of universe with unique area each. In first case you can reuse data and you have 1 area in memory vs 1000 calculations AND 1000 places in memory which could be used for something else instead. Same 1000 npcs and the effect on final computation is diametral different on final computation with second one being not 1000 times harder but more than that as co-computation compounds and you don't just count single thing per npc.

 

[Justicar]

2 MORE WEEKS.

 

[Virgil Brummond]

It is not BS. It is far easier to calculate 1000 npcs standing in same room that use shared pathfinding area vs each of them standing in complete separate parts of universe with unique area each. In first case you can reuse data and you have 1 area in memory vs 1000 calculations AND 1000 places in memory which could be used for something else instead. Same 1000 npcs and the effect on final computation is diametral different on final computation with second one being not 1000 times harder but more than that as co-computation compounds and you don't just count single thing per npc.

Pathfinding is primarily CPU bound and it cannot be done all in one go (due to collisions, conflicts and what have you), that's why your "exponential curve" theory is BS.

In star citizen the AI basically stops working if you have more than 10-20 players, having a smaller area is not going to let you have a thousand NPCs.

 

[Perkel]

It is not BS. It is far easier to calculate 1000 npcs standing in same room that use shared pathfinding area vs each of them standing in complete separate parts of universe with unique area each. In first case you can reuse data and you have 1 area in memory vs 1000 calculations AND 1000 places in memory which could be used for something else instead. Same 1000 npcs and the effect on final computation is diametral different on final computation with second one being not 1000 times harder but more than that as co-computation compounds and you don't just count single thing per npc.

Pathfinding is primarily CPU bound and it cannot be done all in one go (due to collisions, conflicts and what have you), that's why your "exponential curve" theory is BS.

In one go ? No, but it can be optimized thanks to it. And you didn't address memory problem at all. And we are not talking here only about pathfinding.

 

[Virgil Brummond]

OK, memory use will go down because a physical server only services say 1 planet and some moons, so what?

Even when you have ~50 players, the server can't complete all calculations with an atrocious server tick rate of ~5 Hz (i.e. in ~200 ms), that's why the AI stops working.

How is more free memory going to change this? The server ram sticks are going to turn into magical single-thread performance accelerators?

 

[Perkel]

OK, memory use will go down because a physical server only services say 1 planet and some moons, so what?
Even when you have ~50 players, the server can't complete all calculations with an atrocious server tick rate of ~5 Hz (i.e. in ~200 ms), that's why the AI stops working.

How is more free memory going to change this? The server ram sticks are going to turn into magical single-thread performance accelerators?

Why is server rate tick rate so bad ? You are trying to put cart before the horse.

 

[Virgil Brummond]

Star citizen is focused on selling 3D assets and JPEGs in their cash shop. The game is little more than a sophisticated "asset viewer" for cash shop items. They don't really give a shit about releasing a working MMO because they know backers will keep buying JPEGs; Chris Roberts knows his marks.

The server tick rate is bad because star citizen cannot support more than ~20 players at reasonable tick rate (i.e. approaching 20-30), it's not even close, it's not unusual to get mid-single digit server ticks when the server is approaching its capacity (not even full).

Server mashing alone isn't going to change this. It doesn't matter if a single physical server is assigned to a planet, if the server has more than ~20 players (who are actually playing the game, not just standing AFK), tick rates will start approaching ~10 or less and NPCs/AI will stop working. You'll be back to square one.

This is what non-scam FPS games from 5+ years ago target for tick rates (store citizen does claim to have FPS gameplay, right?):

 

[Perkel]

SC server latency is around 41ms for me (just checked it) so according to your chart SC has better networking latency than most of those games. SC does use a lot of broadband which frequently sits at around 1-1,5Mbit which is a lot for online games just shuffling data.

The server tick rate is bad because star citizen cannot support more than ~20 players at reasonable tick rate (i.e. approaching 20-30), it's not even close, it's not unusual to get mid-single digit server ticks when the server is approaching its capacity (not even full).

Server mashing alone isn't going to change this. It doesn't matter if a single physical server is assigned to a planet, if the server has more than ~20 players (who are actually playing the game, not just standing AFK), tick rates will start approaching ~10 or less and NPCs/AI will stop working. You'll be back to square one.

Not really. Currently each fresh server has to handle around 7milion entities refresh + pathfinding for npcs + players + 1000s of other things. All of that at server tick rate.

This is why i talked about sparse vs clumped distribution of playerbase. If there are 3 players at same location, calculation for 5 npcs with them is just 5. But if all of those 3 will move to separate cities and each have 5 npcs that gives you 15 npcs to calculate not 5. 3 times more. Now let each of those 3 players do bunker missions on each planet where you have around 20 enemy npcs fighting with 5 allied npc and suddenly you have ~80npcs to calculate. And this is just npcs ai and pathfinding, there is bullet trajectory calculation, keeping track of entities, etc.

As you can see the more sparse distribution of players the harder it hits server and size of world on server creates exponential curve as players spread out. This is why when you get into fresh server it chugs at 30fps because everything is clumped together but then as players move to different location it starts to grind to a halt. This is the reason too why server with smaller area to care can handle higher player counts.

Their end goal is dynamic server meshing where servers will dynamically spool up additional servers according to load in certain location and subdivide that location down to single room if needed.

Pathfinding on 10x10m square is just fundamentally much easier than doing it with 100x100m square as calculation complexity is squared. But SC is also 3D game so we are taking here about pathfinding in 3D space which means not square but cube. Calculating 10x10x10 box is 1000 times easier than calculating 100x100x100 box.

Anyone who ever played Dwarf Fortress knows it in their bones. If you run 3x3 map in it you can have 250 fps+ with tons of dwarves but run 6x6 which is just twice higher in each direction and you won't even get to 150 dwarves before your fort will reach FPS death.

 

[Virgil Brummond]

SC server latency is around 41ms for me (just checked it) so according to your chart SC has better networking latency than most of those games. SC does use a lot of broadband which frequently sits at around 1-1,5Mbit which is a lot for online games just shuffling data.

1,000 / 41 = 24 ticks per second, so your alleged tick rate is only higher that COD: BO4 Blackout with more than 60 players. Totally "most of those games"!

How many players are on your server?

And at any rate, you full well know that a much more typical tick rate for SC is in mid to high single digits. Even committed star citizen fanboys admit this.

There is no "exponential curve" and the server works closer to 30 ticks because there are under ten players on a fresh server. As soon as you have more than 20-30 active players, server ticks rate tanks and starts approaching single digits.

Pathfinding on 10x10m square is just fundamentally much easier than doing it with 100x100m square.
Anyone who ever played Dwarf Fortress knows it in their bones. If you run 3x3 map in it you can have 250 fps+ with tons of dwarves but run 6x6 which is just twice higher in each direction and you won't even get to 150 dwarves before your fort will reach FPS death.

IN DF, the whole map is relevant. In SC NPCs just walk from seat to seat on a "local" map and don't do anything, SC NPCs don't have schedules or actions outside of sitting in their limited local map.

And if your "game" server can't even handle 20 players being active, that's not going to help with your AI

The UI gurus at CIG are working on a new inventory system:

 

[Perkel]

1,000 / 41 = 24 ticks per second, so your alleged tick rate is only higher that COD: BO4 Blackout with more than 60 players.
How many players are on your server?

100 which is current limit.

I think you are confusing latency with server tick rate. Those two are separate things. Latency means how delayed is information from server and server tick rate is how fast state of server is updated. You can have 120fps server tick rate and 3 second delay or 2fps server tickrate and 5ms latency.

There is no "exponential curve" and the server works closer to 30 ticks because there are under ten players on a fresh server. As soon as you have more than 20-30 active players, server ticks rate tanks and starts approaching single digits.

But i proved you there is. If you have 3 bunker missions on 3 separate planets that's 80 npcs to calculate rather than 25ish. If all 3 players are in single bunker mission then you have 25ish npc to calculate. If you would have server only for that bunker then server could handle a lot more than 100players inside of it. Just updating state of all entities in ram takes time let alone some calculations on npcs.

Currently SC singular server literally has to handle millions of km3 and 100 players. Handling 100 player on small map is easy, doing so over huge map is a lot harder. That's the prime reason why SC servers are so shit.

 

[Virgil Brummond]

100 which is current limit.

I think you are confusing latency with server tick rate. Those two are separate things. Latency means how delayed is information from server and server tick rate is how fast state of server is updated. You can have 120fps server tick rate and 3 second delay or 2fps server tickrate and 5ms latency.

No I assumed you misunderstood the difference between latency and tick rate. What is your server tick rate?

 

[Perkel]

100 which is current limit.

I think you are confusing latency with server tick rate. Those two are separate things. Latency means how delayed is information from server and server tick rate is how fast state of server is updated. You can have 120fps server tick rate and 3 second delay or 2fps server tickrate and 5ms latency.

No I assumed you misunderstood the difference between latency and tick rate. What is your server tick rate?

Depends entirely on state of the server you get.

Fresh servers go usually 25ish fps with 100 players. Everything works.
Most of the time you get 7-10ish server. This is where usually AI breaks
Old servers before they crash or get switched off go 2-4fps. This is usually when stuff start to go horribly wrong like elevators stop working etc.

latency to either of those is around 40ms.

 

[Baron Tahn]

Reading this argument is hilarious. Especially with the name Virgil - it's like Perkel is Dante being led through hell. A parody of a conversation with Dante trying to hold on to his sanity and explain away things he can't comprehend, desperate to hold on to an illusory worldview built on lies and sunken cost fallacies; while Virgil patiently destroys everything he thinks he knows.

Great stuff, bravo.

 

[Socrates]

I'm sure SC will sell FPS friendly ships for the low-low price of 999.99 USD.

 

[Myobi]

https://massivelyop.com/2024/03/19/...n-about-investors-and-their-expected-returns/

Star Citizen’s 2022 financials offer additional information about investors and their expected returns
​

When Star Citizen finally elected to post its 2022 financial report to the UK government (and in related reporting since), you might have seen some people in the comments shout, “CALDERS CALDERS CALDERS!” like they’re trying to summon forth the Candyman. Those calls aren’t by accident, as additional reading and untying of the report has found important details about what CIG’s investors board looks like – and what those investors expect in return.

The Calders in question are Clive and Keith Calder, who invested in CIG in 2018 and 2020; there are obviously other “minority investors” as well, and Chris Roberts maintains majority stake in the company and its investor board, but the Calders represent around 85% of the minority investors’ interest.

As fellow blogger The Nosy Gamer points out, the 2022 report provides further context about the value of the put option – aka the option to sell “a specified amount of an underlying security at a predetermined price within a specified time frame” – the investor group can expect. In the case of the Calders, they have until March 2025 to exercise the option, or all investors could wait until 2028 to cash out, which would see an ROI of about $130M at the stated six to seven per cent interest value of their initial purchase price.

What this ultimately means for CIG is explained in the report, which admits the company does not have the cash to pay off investors at the moment:

“Another reason for not recognizing is due to the nature of the put terms, as if exercised the put proceeds would be payable out of available cash in excess of that required to effectively operate the business and thus that liability would remain undischarged until such available cash became available.”

In other words, assuming pre-orders of Squadron 42 are announced during this year’s CitizenCon, and assuming that the game is priced at $70 a copy, CIG would have to sell roughly four million copies in order to successfully pay off the investors.

 

[Crispy]

In other words, assuming pre-orders of Squadron 42 are announced during this year’s CitizenCon, and assuming that the game is priced at $70 a copy, CIG would have to sell roughly four million copies in order to successfully pay off the investors.

 

[Gargaune]

It's worth noting that the FUDsters in the refunds subreddit discovered some interesting new info in CIG's latest annual Companyhouse report for the UK subsidiary.

It seems that Calders (billionaire family that invested in CIG in 2018) have an exit clause that allows them to get their initial investment back and 6% compounded annually or another formula based on JPEG sales (whichever ones is bigger).

They have an option to demand this money in 2025Q1 and a final option in 2028Q1.

So it seems that Roberts has a deadline to "release" SQ404 and Scam Citizen.

[...] The year is 2026 +/- 2 and macrotransaction yields have diminished to the point they no longer justify the hassle. Roberts puts the screws to his devs to wrap up the bare minimum shippable product, just enough to provide plausible deniability before the backer hordes, and releases Star Citizen. [...]

 

[Baron Tahn]

Ouch. What are the odds though? Clearly they invested in order to make money from a full release at some ROI?

Setting the dates for an exit clause a full decade after investment sort of indicates they knew full well this was a scam anyway. Or did they invest in CIG itself? Curious...

 

[Perkel]

That's exit clause. I doubt they invested into CIG to quit early. If CIG would show signs of folding this is where that clause would probably be used so they could quit early as well. CIG is growing year on year since 2012 crowdfunding start.