Game Discussions Star Citizen Discussion Thread v12

It seems like desync wasn't totally killer on the first day of the event. More just a case of missions failing / not starting etc.

Back to PTU levels now though...


Not to worry tho. IAE will fix it, with the power of ships!
So much for the "We didn't see this in PTU, it had to go Live before we saw it!" crowd. Though that will be the line being spouted in a couple of days ...

Edit: Actually - a lot faster than that :)
1731579338216.png
 
Last edited:
If current content is a guide, 50% of it will involve:

Discussing why the latest ships are flawed / superfluous / OP. In minute detail. (While also slavering over their greebles. And listing the ways that older ships need to be updated ;))

A remaining 25% will be spent trying to fit X into Y...
 
Last edited:
:sneaky: don't tell anyone but seem to remember owning privateer or maybe privateer2 but never really got into it.

I think my problem with WC specifically is it's like the old Barbarian game in the late 80's. You really wanted it to be good, you even thought it was good for a while, but then you realised it was all superficial and the same jerky sprites did the same jerky things every time you played.

6nmy4C5.gif
I bought Privateer because the concept was so cool, and discovered that the economy in the early game basically obliterated the fun of the concept utterly, and shelved it forever. Thankfully, I only spent $6 Canadian for it.
 
I bought Privateer because the concept was so cool, and discovered that the economy in the early game basically obliterated the fun of the concept utterly, and shelved it forever. Thankfully, I only spent $6 Canadian for it.
Yeah, the only fun thing in terms of generic missions was taking that fighter and doing several patrols at the same time in the same system, for a quick buck. I never looked at trade at all.

What was cool is that you could continue doing that without even triggering the main plot of the game.
 
Last edited:
(Thor's Day) We currently have 2 goals today 1) We are working on a new 3.24.3 build for release onto the PTU later today for all backers. 2) We have a new 4.0 build for release to Evocati later today on EPTU with a 2:2:380 mesh config. This is a potential Wave 1 candidate if build proves worthy after a few hours with Evos!

Wave 1 means they are allowed to PvP in the contested zones. At 5 fps ...
 
Good news everyone...

Source: https://youtu.be/Mgbgp4pRSJ4?si=fCgf59I1ZnywrcWM

Once they clean up dirty syncs and messy entities Benoit expects 10x the performance of the PU...

[Lots of evo player footage]

The Head of IT -> CIG Meshing Producer guy: Way more data transfer than expected. UK team built RMQ over few months. Clive, Dave Kaplan etc. Should scale.

2m45s: Jordan Wood: student hire -> Network Programmer 3: RMQ is a more stable foundation...

Expected to work better than old system at scale. It didn't.

Benoit says it did ;). But it revealed other issues.

They were still seeing interaction delay. High latency. Packet loss. Msgs taking long time to go thru the hybrid service etc

[Fan footage of dysfunction...]

Traffic jam on one thread.

[Repeat: with RMQ really thought about how not to end up here again...]

6m: The prev NMQ was tech debt

Not just handling player data, but NPCs, doors, elevators etc

Issue now = lack of parallization when updating dirty messages / variables. IE info or variable attached to an object not yet communicated to all other members of the mesh.

'Sync dirty' taking up as much networking bandwidth as NMQ was...

Now addressing... (optimize + parallelization etc)

Clients in area receiving same update need to be checked to confirm reception. Synchronized etc. Can be 'read only' / all updated at same time in theory.

Parallization showed some gains. Pushed up to 1k players. Hit issues with 'binds'. IE client being bound to server's version of entities [?].

Avg Stanton game session = 1m entities... Only gets bigger. Starts at 720k entities.

After a week approx 3m entities per shard [for Stanton alone?]

Want to parallelize binds now...

Benoit expects 10x the performance of current PU.....

Message from Chris: Go as big as we can, as quick as we can...

[Excited content creator babble]

Student hire excited about ripping out old things from last 10yrs and making best thing they possibly can...

Lando: Binds have made great progress, revealing the layers/issues underneath...

Just needs Janitor Tech IMO...
 
Last edited:
Yeah, the only fun thing in terms of generic missions was taking that fighter and doing several patrols at the same time in the same system, for a quick buck. I never looked at trade at all.

What was cool is that you could continue doing that without even triggering the main plot of the game.
I couldn't afford the fighter and didn't feel like doing battle with encounter RNG to get a shipment of potatoes from point A to Point B enough times to earn the money at a profit of $0.12 per run.
 
Good news everyone...

....Avg Stanton game session = 1m entities... Only gets bigger. Starts at 720k entities.... After a week approx 3m entities per shard [for Stanton alone?]

....Want to parallelize binds now...

....Student hire excited about ripping out old things from last 10yrs and making best thing they possibly can...
If you are trying to track, record and replicate too much data, the simple fact is that every/any/all technologies, their underpinnings and networks will at some point reach capacity, particularly if you exponentially exacerbate issues with multiplying stateful nodes.

Providing parallel/multiple routes for information can help.

BUT....

At some point, a single entity has to see/control the whole picture (where a finite state is necessary), no matter the frame's size, and the/any underlying point of constriction can often be extremely obtuse, and is very often not where people tend to look/fascinate. Particularly not excited student hires.

But don't take my word for it. I've only had one career in real-time distributed messaging systems.

Oh, and as for losing packets..... (broadly speaking) no.

For balance, some of what they say here is quite right, although it is concerning that some of it seems to be 'news' to them, and not ground out at the design stage.

TL/DR: maybe hold off on buying that Idris for a while yet.
 
Last edited:
If you are trying to track, record and replicate too much data, the simple fact is that every/any/all technologies, their underpinnings and networks will at some point reach capacity, particularly if you exponentially exacerbate issues with multiplying stateful nodes.

Providing parallel/multiple routes for information can help.

BUT....

At some point, a single entity has to see/control the whole picture (where a finite state is necessary), no matter the frame's size, and the/any underlying point of constriction can often be extremely obtuse, and is very often not where people tend to look/fascinate. Particularly not excited student hires.

But don't take my word for it. I've only had one career in real-time distributed messaging systems.

Oh, and as for losing packets..... (broadly speaking) no.

For balance, some of what they say here is quite right, although it is concerning that some of it seems to be 'news' to them, and not ground out at the design stage.

TL/DR: maybe hold off on buying that Idris for a while yet.

Yeah I feel like we've seen this scenario so many times now. An excited student hire having fun sniffing around old blockers, ripping them out, and extemporising solutions. (While almost certainly making next year's blocker in the process ;))

You can already imagine next year's videos...

'We realised that packet loss was affecting the gameplay experience...'

'We've created a new service called THE ONE TRUE RING...'

'We're proud to finally announce! Janitor Tech! Also coffee cups are no longer interactable.'
 
Berks found a wrinkle. Or possibly a seam...


(He Alt-F4D-ed and came back to this)


Still unable to exit his seat or control much once he brute forced his way out of the tunnel. Looks like classic authority limbo...


Had to reset his account.
 
Last edited:
Back
Top Bottom