Server Meshing Is Closer Than Ever...

[@tenpoundfortytwo]

Feeling more or less confident after this test?

[@blamsmith]

More confident in the DGS and RMQ and less in the hybrid/replication layer 😂

[@shh532]

I'm staggered it's coming along this quickly. I really think if we can get ahead of the crashes, get it even faster, and we can get a 6:1000 shard running consistently, 4.0 is absolutly possible EOY.

[@Tsudico]

It did appear to be a good improvement over the previous test. As an aside, BwIn (Bandwidth in) will always go to 0.00 Mbps and stay there before a 30k so that is how I always determine if one is coming. It appears it is much more of an indicator than looking for Server FPS freezing.

[@otockian]

more

[@Irritated_Considerably]

Cautiously positive. It may be a good idea for CIG to unlock LEO Stations as *Homes*, as well as the Space-Ports to reduce the strain a little, at least in the short term. The game itself can't handle massive log-ins: Not enough beds or hangers. Thanks @tenpoundfortytwo

[@mcdowntrend]

I really appreciate your very straightforward and clear style.

[@tenpoundfortytwo]

Thanks!

[@chiron03]

The 4:500 I played on stayed up for hours without any crashing. From what I saw in the test. It was a MASSIVE improvement. There was a Tweet that went out from Roger Godfrey talking about how excited the CIG devs were during the test because their backend improvements were working massively. I think if the hybrid service wasn't failing, the 6:1000 test would have gone very smoothly. It felt smoother than the 4:500 setup which is weird to say. Interactions where damn near instant. I personally see this as a massive win, it truly felt like they made some big progress. I think once they solve the hybrid crash and put that 6:1000 setup back up. Its going to stay up for 24 hours with minimal issues.

[@echo5827]

No invisible players and minimal teleporting or interaction delay. Huge win, just have to workout those crashes but we weren't even getting those before so they'll be fixed quick

[@tenpoundfortytwo]

@chiron03 great info thanks! Should be an exciting next few months!

[@QuotidianStupidity]

What is the hybrid crash?

[@alexanderdooley5833]

I belive a hybrid crash is multiple causes contributing to single crash.

[@QuotidianStupidity]

@@alexanderdooley5833 sounds logical, thanks. 👍🏼

[@DankestJedi]

Glad it seems like meshing is actually becoming a thing. cant wait to see the verse just filled with people!

[@tenpoundfortytwo]

Yeah it's going to feel pretty different! Plenty of the game is going to need to change to make it work though- elevators are just one example of something that will need reworking!

[@DankestJedi]

​@@tenpoundfortytwoI've noticed that from the streams and videos I've seen of the test. The transit system is definitely in need of that rework!

[@ephalanx1]

@@tenpoundfortytwo Yes and no. With the newer log on methods, everyone will not be spawning in the habs. But in hangars, bases, and habs.. Global chat and more will though. This is really high in max edge case with everyone spawning into the same general area. The game wont really be that way in the future.

[@ephalanx1]

@@ChristoffRevan I understand and agree with much of that. I’m just pointing out that login/logoff and locations will be different in the future and not an attempt to undercut tenpounds assertion. There are changes they will need to make. Atc, chat, landing zones hangars, transit, etc. But this is an edge case and meant to stress test. Even with 1000 people per system most likely they will be spread out in different systems on different planets, space stations,, hangars, etc. This will be more out of the gate week one release style spawn volumes at landing zones. Either way yes they will need to change some things. To be honest every MMO has issues with condensed player volumes even if they support massive numbers of players. Clients will still have issues at times with rendering, etc. I think there should be some realization that SC is not magic and will still have bottle necks here and there. Ie Everyone piling into a single towns square, battle zone etc. Hopefully they do a lot better but there are physics realities that still govern things. To be honest I’m more interested in how they leverage more so traffic at locations where nice this is in place more so players. That way they can put in AI ship traffic so things are more alive and ‘act accordingly’ per the lore and world. Ie ships hovering upside down or players moving erratically. Yes a player will stand out but if there is more general AI traffic that would feel more immersive. We all know many players are always considering looking immersive. lol.

[@dtrjones]

Very positive news. As a software engineer myself, the crashes arn't really a problem from the point of achieving network performance as they'll continue to look at resource and load issues. It's really about achiving and sustaining a level of network performance load which can meet user needs and it looks like they are almost meeting that expectation.

[@johnokean8216]

It’s crazy how much of this game is server dependent but it also makes sense given its scale. Really is a unique gaming experience.

[@johnokean8216]

Server decay is another phenomenon that blows my mind. It’s just weird to imagine lines of code start to fumble.

[@SETHthegodofchaos]

it makes sense to be this server dependent to (eventually) prevent cheating. Server decay isnt about code getting fumbly :D It's about items (game objects) being spawned when people move around in the game world, load locations and spawn items and ships themselves. There is just too much stuff going on, so there is too much loaded and simulated for the server that it starts taking longer than the 33ms that it is supposed to take for each game simulation update.

[@johnokean8216]

@@SETHthegodofchaos isn’t that essentially the same thing that happens to a football player when he fumbles the ball? Too much going on in front of the runner, lots of obstacles, eventually folding under the pressure and detrimental mistakes occur. I like this analogy a lot.

[@SETHthegodofchaos]

@@johnokean8216 I guess :D I personally prefer the analogy of a cook of a restaurant. He is given ingredients and tasked with recipes. The more people come to the restaurant, the more tasks are given to him. If there are too many people he will become overwhelmed with too much too cook. The customers will have to wait a longer time for their food. It might even be undercooked at times.

[@amineabdz]

@@johnokean8216 I'm not a game dev, but a web developer and I think for games, there 2 challenges that are extremely hard to balance for large scale : 1- Deadlocks : there are definitely some logic flows in the code where it leads to the server entering an infinite loop, recursing until OutOfMemory or missing to handle a crashed service that is essential for other components. 2- Request Queing : This is somewhat more relevant for web development, it's basically action timeouts, where each request has basically a "time to live", if the server is busy or is not responsive that request will expire and be dropped and ignored. From my experience CIG must have either a very high Timeout or a queuing system for actions between clients and servers, that is why when sometimes if the server is on it's knees, your action that you performed minutes ago gets processed, This can be optimized in many ways, one of which was the introduction of RMQ which is a more efficient processing messaging queue, another way is to optimize the processing power requires to process these actions by the server. It is also possible to have a "shorter" queue size or a shorter ttl for requests and that would basically prevent the server from locking up too much, but at the expanse of ignoring client requests, This is a tradeoff that can be used for most websites, where if the requests just times out it's not a big deal, usually just an image or a video not loading, and a refresh would most likely solve the issue but for real time stuff like games, not so much.

[@cyvan1750]

Yeah I logged on for the 4:500 and I was amazed how well it ran. I even went into the bunkers and promptly got slaughtered. The AI had faster reaction times than I did. They'd come around the corner alert and start firing the minute they saw me. It was intense. I'm not concerned about the crashes, that's something new; what's important is that what was broken before seems fixed. The crash could be something as simply as someone misplaced a semicolon in some code that didn't get executed until u reached a threshold. They'll figure it out and fix it :) Great work CIG!

[@XiXGoldeyXiX]

Really appreciate these comparisons between the tests. I haven't been able to play myself, so it's nice to see how things are progressing!

[@reamoinmcdonachadh9519]

Better to have consistent incremental improvement as each test is done. The thing is to have SOMETHING reliable when all the testing is done AND a set of known variables you can knock on the head and eliminate when problem solving any issue and thus leaving you with NEW things you can check for!!

[@EliasWindrider]

Thanks for keeping us in the loop.

[@MersageSW]

Great insights! Thanks for making this video

[@tenpoundfortytwo]

My pleasure!

[@MackyMus]

I couldn't believe that the tram at A18 could have so many players with little desync, I have the screenshots before the 30k. It was the most epic thing I've seen in my 40 years of gaming.

[@dzengrinder]

We're gonna make it boys, it's happening

[@snowmind]

Banger of a video, very well edited too. Also to add, that as I was also in that shuttle myself, we seemed to suffer from the same +50-100ms (at a min) extra latency due to our location, which didn't help things. All in all though, progress is steadily being made across all fronts (evo .2 also made huge strides ahead in the patch released at the same time as the tech preview one).

[@AbdiYohan]

Great summary, thank you!

[@tenpoundfortytwo]

Thanks!

[@a.j.9722]

Thanks! Great vid with substantial proof and not just “I feel like it was X and Y”. Very appreciated. I’d be also interested in server degradation now with meshes.

[@karma_cardinal]

comprehensive and understandable report. thank you very much!

[@AngelicStreak]

And people also forget that these Tech Preview do not simulate ANY of the hundreds of ongoing missions that the Live servers do. I wonder what the server FPS would be if they did.

[@saleseng]

That’s exactly right none of the server meshing stuff they’re doing now means anything until they apply it to Pyro and other multiple systems not to mention all of the backlog of large capital ships like the Banu merchant man javelin, etc. if all runs smooth, then server meshing can say it was success, but as of right now it’s really not anything

[@adamchambers1393]

Pyro makes zero difference in itself. That will be similar to Stanton when it's split. Yeah ships to a point but even watching landing of 3 890 jumps and the hammerheads racing about etc. Was pretty damn solid for that even now, let alone how it will progress in time still.

[@saleseng]

@@adamchambers1393 Yeah, but I was merely repeating what CIG and the player base have been saying for years now, the whole excuse for not having the big capital ships in the game or other systems they said has always been they have to wait on server meshing. Even that guy Benoia , the tech officer said that there was too many entities in the game for everything to function well without server meshing plus tens of thousands of players on the servers

[@adamchambers1393]

Indeed but this I think is already showing that those things will be possible. But yes It does seem like dynamic meshing needs to be fully working for you to get to point of large fleet battles though and that is a way aways yet.

[@BuildandPlay_]

I did manage to map out the entire system in the 4:500 setup, my shard was stable (and running significantly better then Live) for at least 4 hours. I also tested switching from DGS to DGS with another player, I'll post all my results in your Discord.

[@tenpoundfortytwo]

Thanks!

[@Anonnymouse53]

I managed to spend 30 minutes on the 500:4 configuration after that had been running for about 4-5 hours. I had no problems - but I didn't do much, just flew around a little bit and ran a bunker. I'd say it was about the best experience I had since running the "offline mode" hack. My client side FPS was solid 80-110 in landing zones without stutter or hitching. Server FPS was 18-25. Desync was barely noticeable with other players. It really did feel like a whole different game. I'm still pretty cautious about becoming optimistic, but for sure it was a noticeable progress from the live we've been dealing with for years.

[@DustinHarms]

Generally jaded and negative long-time backer here: I actually felt pretty positive about this test in comparison to the previous ones. Yes, there's a hybrid crash, and there's a chance that could implicate some problems with the hybrid service, which would be pretty bad - but that isn't guaranteed to be the situation. It could just be a crash that they can fairly easily fix. But the performance while the servers were up? Vastly improved. Considering the difference is only like 1.5 weeks between tests, I actually have to say this is great to see. Again, I just hope they can keep it up. CIG still pisses me off, on the whole, especially in some particular areas, but I can't bring myself to doom and gloom over this test because there was a fast-repro crash occurring. I'm a credit-where-credit's-due guy, and there's some credit due here. I hope folks understand that this is still positive momentum and that crashes aren't usually a harbinger of failure.

[@AdolfoWWolf]

That is incredible, seeing that even interactions wouldn't work well, with other players being stuck in place, this really is impressive, i'm more curious on what they will be able to pull off, and prefer to not try any guesses now, but that was definitely very cool to see.

[@OPossum76]

Happy to see someone looking at the important stuff for once. Nice video.

[@aguy446]

I can't imagine why 4.0 would need more than 4:500. Sure, ultimately they'll need to go further than that but for now it looks like they've got it in the bag

[@Billy-bc8pk]

People tend to forget that Pyro is multiple times bigger than Stanton. 200 players in Stanton will make it feel okay, but 300 players in Stanton will definitely make it feel a lot more alive. 700 players in Pyro will give it decent bit of player interactivity and breadth, but even then it will still likely seem kind of sparse. Then again, who knows? Maybe they will go 1,000 for Stanton and a 1,000 for Pyro. But 500 for both will definitely give it a good feeling of population.

[@tenpoundfortytwo]

I guess they want to push it as far as they can in these tests and then work out what to go live with in 4.0. I found it really interesting that they were aiming to go above 1000 in this test-will be fascinating to see what they end up pushing.

[@melandor0]

Pyro is bigger but that doesn't matter since people aren't in deep space. It's the planets and moons and stations that actually count towards how often you see other players.​@@Billy-bc8pk

[@SC-monzo-nt2lo]

The 4:500 test felt just as smooth as LIVE, I was on a shard that didn't crash for a long time, it was amazing. The 6:1000, like you said, had some desync and a lot of 30k, but aside from that, it was great. The progress is amazing, I wouldn't be surprised to see a successful 2000 players test in a few weeks at this point.

[@blamsmith]

Hey man, did you ever revisit the vram usage in vulkan in 3.24? Seems smoother and could be good to see if there's a difference in usage now. Cheers!

[@tenpoundfortytwo]

Hey! I was about to take a look but then 3.24.2 went to evo, so I think I'll wait until that goes live. Seeing as it's based on the 4.0 code branch it should have the latest Vulkan changes in!

[@amineabdz]

​@@tenpoundfortytwoDoes that mean they finally merged their fix for the smudgy animations ? AFAIK it had to do with TSR and their new AA techniques

[@tenpoundfortytwo]

@@amineabdz definitely something I want to check out- not had a proper look at the Evo patch

[@Silvan278]

@@tenpoundfortytwo yep it has lots of changes and the renderthread is a job now as well. Parallelizing the renderwork is in progress right now:) still a lot to do but the renderer is shaping up really well!

[@tenpoundfortytwo]

@@Silvan278 Great stuff, thanks! Will you be at CitCon? Would be nice to say hi!

[@TheMinistry-SC]

thanks for the update!

[@echo5827]

Big win, didnt think they'd get 1000 players working before 4.0 by how much they teleported or went invisible like we were in early 3.0 again Now I feel like 1000 players for pyro is easily achievable. 500 is already good. Hope they get 4.0 to a wider PTU soon since they won't get much meshing testing done with so few evos

[@Billy-bc8pk]

It's crazy to me how the 500 player tech preview runs so much better than live. The fact that CIG seems to be aiming for a stable 1,000 players for 4.0 is pretty crazy too, given that it would be a ten times jump up from where we are right now with player counts, and it will run better than the current live build.

[@fleetreadygaming]

If we can manage 6dgs/2k pop cap the verse would feel so much more alive, at least with Stanton, each server would have over 300 players in it 🎉❤ Let’s Go CIG! I’m ready to continue testing this till we get it down!

[@adamchambers1393]

I think they will have to try and push more like a 10:2000 as you are then not getting enough performance per server. As numbers go up and you're more condensed you can shrink the volume of entities the server picks up so you might split moons into individual groups. At moment as example you've got - space - arccorp - hurston - microtech - crusader - moon cluster a - moon cluster b As example. This can be split into each moon being a server and then even splitting planets in half across their equator as next steps but ramp up numbers of player so you can increase density per server as it smaller total entity count in example we moving from 100 people a server to 200 people a server in that 10:2000 example then. Then you could maybe go 14:3000 and split planets into quarters meaning your density steps up again to around 215 players a server. Each increase per server also reduces costs for CIG as they need less servers if they can increase density. So kinda becomes a win win. They just need to work away at that and also make sure they performant enough when an event happens and suddenly 500 people or whatever show up. That will likely cause a server to fall over till dynamic meshing happens.

[@UPLYNXED]

Couldn't get out of Orison on the 4:500 test. Transit completely broken (except for the Siege Of Orison shuttle that we can't get to anyways). Was getting crashes to desktop simultaneously with other players. Not a great experience.

[@tree0311]

Everyone should be extremely optimistic following this test, and I'll try and explain why (this is why CIG was so stoked about the results). Other than the Hybrid crash that CIG identified early on in the 1k player test, it was a massive improvement for that scale, and the 4:500 test I was on for quite a while and it was utterly flawless. Stable ping no different than live for me, no interaction delay, stable server FPS 25-30, and no desync whatsoever. I think what everyone is confused about is thinking that CIG is shooting for 1k players at 4.0 launch. Testing at 1k players, and maybe more next test after they address the hybrid crash and include some more optimizations, is just to see at what points does the tech begin to strain and even break. It isn't an indicator of what CIG are shooting for with a Static Server Meshing configuration by any stretch. If they launch with 4.0 with 6 DGS and up to 1k players in a shard, it means that when concurrency isn't at peak hours they'll have more servers tied up being under utilized (and costing $), with much more risk of poor performance and crashes when concurrency is high, which would threaten their goal of doing more to preserve the live experience. 4.0 and Static Server Meshing isn't about high player counts. It's about proving out the tech and giving the data needed in the live environment to continue development towards Dynamic Server Meshing, which allows for the higher player counts but also the flexibility to be efficient with server resources. We will get Tech Preview tests following 4.0 once they are testing Dynamic Server Meshing and then they'll really push the shard player counts, but they'll be testing at what points to scale up both the number of servers but also the number of individual replication layers (replicants). Another thing of note is that the Replication Layer model isn't new, ESO uses it as well (in a static server mesh configuration) and they have an upper limit to how many connections theirs can handle. I'm brain farting but one of their network engineers told me it was either 600 or 900, I forget which. So CIG is testing to see the limits of their Replication Layer in terms of connected clients and servers, because those limits are where they need to make sure they are spinning up another Replicant (individual Replication Layer) and Dedicated Game Servers well in advance of to avoid over taxing the system and causing a crash or other disruptions to the game experience, but this is for Dynamic Server Meshing, not Static. So, in summary, if CIG gets 6:1000 working reasonably well, but they launch 4.0 at say 4:500-600, when they move on to testing Dynamic Server Meshing they can spin up another Replicant when the connected client and server count to a Replicant reaches a certain point so that once its ready they can further subdivide the workload before reaching the point where the system starts to really strain or reach the point of failure, and vice versa when the inverse occurs.

[@cox870]

Thanks for the insight ! I wonder how long it will take for CIG to get dynamic SM in place. My best guess is they'll want to have it near completion/in testing by the time SQ42 comes out (2026) to attract more potential players. Maybe that's too optimistic, so I guess we'll have to wait until 4.0 to see how static SM performs

[@tree0311]

@@cox870 long before that, Benoit thinks a matter of months. But they'll be iterating and improving on it for a while.

[@Funkm4ster96]

the 1000 test, besides the 30k, was really good. no delays and it played like butter for me. There was fairly big progress and I'm more confident about what they can deliver. the 4:500 test played like a smoother live build. I think its very likely we will see the 4:500 player environment for Pyro or at least that could be their goal

[@Dano_Hobo]

Consistent, reproducible crashes are great. Easily tracked and fixed. Fixing the delays are huge

[@marcuswynn89]

I’ll believe it when missions work

[@tenpoundfortytwo]

Yeah going to be interesting to see when they merge back into the 4.0 branch and turn on the missions!

[@Skyborn]

thanks !! this was a good watch o7

[@CitizenScott]

Why, tenpoundfortytwo? Why must you make me hyped?? Why???

[@tenpoundfortytwo]

It's been a long time coming! I think a bit of hype is ok when CIG do well!

[@Demonic-i9j]

We love our hopium

[@angepano8591]

Dude, how are you getting any of this to work for you? I never got past the hab or hospital elevator panels no matter where I tried to start off. There were crowds of players stuck with me.

[@tenpoundfortytwo]

Hmm I wonder why that would be happening. I tried on a laptop as well as my main system and that was working fine too.

[@1scrub2]

@@tenpoundfortytwo There's that 'streamer bias' lol🤣

[@velez910]

I see great progress. Far better then before. And constent testing real hard push with the public fast changes. Yes a he'll of a lot better

[@1jbrady]

I don't think any other person monitors the performance of star citizen as much as you. So you if say there is a significant improvement, I'll take that as fact . Thanks for your work .

[@cmdr_thrudd]

Encouraging improvements :) o7

[@brandonswitzer3907]

yea i jumped in and was impressed how much progress in a month.. don't get me wrong its not where they want it... as some one who works with large code bases, i Love the attuited of CIG right now

[@joebaxter6895]

I hope you don't Pidgeon hole yourself into just tech stuff. You have a great communications style and hope to see you expand into other arenas of Star Citizen coverage!

[@JRM-VSR]

I finally got into one of these tests, and my experience matches yours pretty closely. The 1000 player test crashed very frequently. So frequently I doubt anyone got into space. (I did manage to spawn a ship once, but had a 30K before I got in to it). But between the crashes it was surprisingly smooth with players moving around properly, and the transit was surprisingly un-deadly. I've absolutely seen worse in Live. An easily reproduced crash like that should be easy to identify and reasonably easy to fix -- chances are it's already fixed internally. Expect another test next week. Will 1000 players per shard work next time? Odds are good that it will. Will they try to go past 1000 players in the next test? Almost certainly. Will they find more problems? Also yes. But that's what testing is all about, right? With that crash bug fixed, a 6:1000 config that performs like that would be a valid launch config for Stanton and Pyro -- three servers to cover each system. It will also give server headroom to improve AI compared to live, and fix some other server performance related issues. This says that a 4.0 with Server Meshing in PTU is a reasonable prospect this year - at this point it's back to being dependent on the other things in 4.0, not Server Meshing itself. What other factors? The mission system doesn't work at all in a multi-server config, and hard requires a rework. While the transit system was surprisingly un-deadly in the test, it certainly wasn't 100% functional -- there were places it hard didn't work, like ... all the Lagrange point stations. So Transit Refactor is going to be required too. There were obvious comms issues (constantly being kicked from channels), and issues with ATC as well, both of which will need some level of work (if not a complete refactor) before this can be considered for live. And that's before we find issues with Pyro itself, Engineering, or anything else. All things considered, my predictions from about a month ago of a 3.24.2 to go live before CitizenCon (with as many of last years remaining features as they can scrape up), a 3.24.3 to live for IAE, primarily driven by ships, with a featureset to support what's being launched, and 4.0 going through PTU waves to get to open PTU likely in the week before Christmas still looks pretty solid. I just think the shards are going to be configured larger now than I did then. And with perhaps a thousand players in Stanton, it's going to feel like the universe is much less dead after this change.

[@JoelBondoux]

It's all guesswork for us players, but as long as CIG are seeing incremental advancements then I'm happy. The real question I have is does this tech need polishing and completing before 4.0 releases? If so I wonder how many iterations it'll take...I guess we'll find out soon enough.

[@The_Opinion_of_Matt]

Meanwhile on reddit... Could be a meme. Server meshing testing is the real important news.

[@Citizen-Nurseman]

Definitely need to see how this survives PES in a non-fresh server, its impressive, but will it be impressive with 3 trillion cuz bottles on the floor. and is client performance going to be preserved in busy hubs from all the ships, players, objects being rendered.

[@tenpoundfortytwo]

Yep a very good point- still lots of things for CIG to work out. But I guess they can't get to that until they've got the basics working!

[@echo5827]

3 trillion bottles really doesn't hurt performance they've said this a bunch. Once they're static they stop updating after a few seconds The main thing are ships and anything players are interacting with like cargo

[@Crittek]

Can you do a video on virtualized geometry? They announced that and nobody has seemed to of blinked an eye. This is what nanite is, and that’s why they said they’ll be able to render trees into the horizon. This is huge news and was on my wishlist of features.

[@tenpoundfortytwo]

I must have missed the announcement, when was that?

[@Crittek]

@@tenpoundfortytwo Challenging my memory is a tough request lol but here is what I found: Monthly Report June 2024: “GPU terrain generation was implemented as a stepping stone to the team’s plans for fully GPU-generated virtualized terrain, which will be utilized in Planet Tech v5.” I know it was at least CitizenCon but I’m sure it was a more recent SCL or ISC (perhaps “Q&A Tech Talk”?) where I specifically remember somebody mentioning this and “this will allow us render trees up to the horizon and eliminate pop in” This isn’t exactly “geometry” as I stated but initially Nanite was only compatible with static meshes and not landscapes but now it does and can be applied up to animated actors as of very recently.

[@tenpoundfortytwo]

@@Crittek very interesting! Well I'm sure we won't have too long to wait for more info- they should cover planet tech v5 at citcon!

[@Crittek]

@@tenpoundfortytwo Ya V5 is pretty hype. Unfortunately it’s hard enough to wrap my friends minds around Nanite let alone describing it as “virtualized geometry” lol If or more like when, this affects more than just landscapes we will get much more stable frame rates with high player counts. I would love to see someone ask this question on an SCL or at CitCon because I don’t think they’ll mention it willingly.

[@Heppetronics]

Rapid iteration doesn’t mean iterating on something to rapidly reach a solution… it actually means the length of time BETWEEN each iteration is rapid, although the end goal might actually be quite a fair way off. Just for clarity and to limit expectations with CIG wanting to do “rapid iteration”. It’s a dev strategy, not a problem solution.

[@antoniovinciguerra8982]

I do feel more confident but 90% there in terms of stability and server frames and delay won‘t cut it - we need 100%… now it‘s down to whether they can hammer away at those last few percent

[@jimashby43]

I think its still going to take a long time to figure this out.

[@tenpoundfortytwo]

Yep, don't think it will be on live servers anytime soon! But some progress is better than none!

[@RonPaulBot1234]

I think it's a rabbit hole but im downplaying the amount of engineering and foresight of CIG. But if they can run 250 players it's already positive. Having players teleport is a bad sign as it shows they dont bake latency into the gameplay but the fact that you already can see stability at this level and I know there is a lot of room for improvement, this makes at least part of the vision more attainable.

[@Sparky_007]

It's progress, but it's obviously not really playable. So let's see if they can get it to a hardened and reliable play-state.

[@widt94]

Think they are doing really well progression wise :)

[@kon-lun-os-kuh-pee571]

What is dgs and hybrid crash?

[@BGIANAKy]

Progress on a fresh server. Please tell me what happens after 2-3 hours. 6 hours. 12 hours.

[@everythingpony]

So is this where all the players are going? Anytime I play there's less than 20 people on any server?

[@lights4bre890]

sure.

[@beeleaveme]

Did you see the Unreal Engine presentation on "Server Meshing" they created

[@Starlette-bh4qn]

how many players were actually on your server?

[@StarCitizenSpeakeasy]

do you remember if the servers were full?

[@Bronwyn031]

A tad bit optimistic I'd say. The issue with CIG servers is the DEGREDATION over time. It's a start but I wouldn't start celebrating just yet. If the whole thing cannot support play sessions across hours, then most mission types will not be playable. And I'm still firmly in the camp of "The current game design cannot support thousands of players across the Landing Zones." Not enough habs, kiosks, vendors, elevators or hangars to be a pleasant experience.

[@ntsarb]

20 fps on the server, seriously? I can't get more than 8 fps, most of the times it's below 5 fps! It's good that server meshing is being tested. It's a complex piece of tech, I hope it works because the impression I've got all these years is that CIG don't have the required know-how. I hope they develope it in a timely manner.

[@Tainted-Soul]

I think all they need to do to help the players is limit the number of people that can spawn in each place. spread players about using all the cities and Orb stations and even the Mic l1 for example and once in then you go you own place 100 players each place 10 places but what do I know LOL they may even trying to test the mass player start up fingers crossed they sort it

[@aventus2140]

nice keep going CIG its going the right way

[@reamoinmcdonachadh9519]

My big worry is when all this is done and everything is working as intended and as it should, it might require players to upgrade their rigs to be able to play the game. I hope not, I hope to be able to play on my current rig, and while I will (if I can afford to) upgrade over time I still want something I can play during that process.

[@1scrub2]

Honestly, my FPS has been going up each patch. Vulkan improvements and server-side optimizations that cause hangs will only improve this... BUT then ray-tracing/universe capable AI/ larger player and ship counts....

[@otockian]

As a developer, this is 100% good progress. Anyone saying otherwise has absolutely no clue how development works, and their opinion matters very little.

[@corwyncorey3703]

...which rather distinctly describes a vast number of people who complain about "700 million and can't make an even vaguely stable game". Heck most of them think a decade in development for a studio built from scratch is "scam level long term" instead of par for the course. They forget how long it takes even established AAA studios to work on titles. Let alone crowdfunded ones. If only an actual "ignore" button in game chat existed...

[@littlefoot2869]

on live it busted just like ever patch not say this bad but still think it going be years still before it works how it should and think they need 4 for 300 then maybe it crash out every 3 days long runs

[@DaringDan]

Really stoked on this. What's 5-8 years over timeframe prognostications by CIG between friends?

[@tenpoundfortytwo]

😆 let's not talk about the dates- too depressing!

[@1scrub2]

Damn elevators kill players and servers with equal abandon.

[@SpaceDad42]

It won’t change much. Certainly not what the majority of the player base thinks it will change or fix.

[@pigozz]

Read thoroughly about the meshing and its pretty obvious the replication layer is getting massively bottlenecked as server nodes are trying to write their data they have authorities over. Server meshing wont solve shit

[@hirmor666]

Dream on 😂😂😂😂

[@Thor_Asgard_]

Closer lmfao ^^ years away again as always.

[@The_Fallen_1]

I was pretty disappointed by last night's test. Not because it performed poorly (I had less than a second of interaction delay at 1k players and players actually moved around in real time), but because of the 30ks. Even then it's not because they were disruptive, but because it was clear the devs where pumped to push past the 1k player limit for the first time and had to stop because of the constant crashing. It felt like it was going to be a really good test that was killed off early due to a single issue.

[@tenpoundfortytwo]

Yep was a shame that we couldn't see what they were planning past 1000. Hopefully that'll come in one of the next tests!

[@dawnphantom2439]

That's just how testing is. Solve one provem at a time.

[@HappyChap8]

Why have 1000 player servers when only 30 can fit on a train? I swear this game is being made backwards😂

[@TheWandererTiles]

This will not work at all. We can already see it fails in a tiny indoor scene with almost no interactions. Total fail. Imagine how badly this will go with more players with huge amounts of objects and interactions in an open space. In other words, it fails with relatively little network traffic, it will not work with more.

[@GTOOtt]

WOW server meshing will fix everything! Will it stop CIG treating it's backers like $h!t?

[@dipperspace]

The game has just become garbage, there's nothing to do even for one player, let alone a hundred. no complete mechanics. and now it's not a space simulator, it's a simulator of swimming in a swamp... or should I say swimming in shit.

[@Lisa-Azra_Broad]

Yeah! server messing, pointless waste of time in the game's current state, sure 10+ solar systems go for it but now why? when we still fall through the planet, cargo lifts don't work

[@borgthepig]

Why? Delays on actions becomes far less. Physics calculations work far better. NPCs respond the way they should. All of these improvements are literally requirements for the game to become playable, and your saying that server meshing is pointless??

[@Lisa-Azra_Broad]

@@borgthepig current improvements seen with the npc Ai are nothing to do with server meshing, same as changes to the flight physics, currently on the live there is no issue with response rate and there is no server meshing on live. There is F'd up cargo elevators making that gameplay on the whole non viable, still falling through planets, how many years has that bug existed

[@Luciphell]

That's nice, but the physical technology that would allow CIG to do with the PU what they claim to be shooting for DOES NOT EXIST. They haven't created a way for servers to magically do more than they can. Server meshing is not that. Not only that, CIG continually builds their tower higher on shaky, unstable foundations and they act like they can just fix that shit down the road. Which means 15 years from now when they release 1.0, it will still be just as much a buggy, janky mess as it is now, only then people will be buying the "just wait for further patches" bullshit.

[@1scrub2]

You realize CIG has already done what the nay-sayers prognosticated 'couldn't be done' right after their successful kickstarter, right? And they will keep finding new and inventive ways to push barriers. Its what trailblazers do. Its what I paid for. If you don't like it, be happy with another game.

[@taurel2013]

@@1scrub2 For the moment, strictly nothing is 100% functional, in any field. We are far from victory

[@1scrub2]

@taurel2013 I never said victory. I said they already succeeded beyond the naysayers and will continue to do so. Only those without vision cannot fathom new things under the Stanton sun. The scale, with the granularity of sub meter positioning with persistence was said to be an outrageous scam, yet here we are. The ability to stream such volumes was said to be impossible. Yet here we are.

[@cmdrls212]

The problem is that their best still unplayable. And we don't even have capital ships and these huge battles they sold everyone on. They should consider Releasing the game with less player counts and admit the game doesn't work for large scale multiplayer. capital ship owners will get screwed but they can still keep their ships as hangar queens.

[@Sieztt]

yes let's get the server people to fix the bugs that are currently plaguing the game when server side the game is the most stable it's been holy shit it's like they're doing their jobs!

[@PhantomRider-78]

60 players and 940 bots…just saying

[@jakeramp1]

No. unless you're referring to unreal engine, then sure.