5000 Players Over 4 SQM Demonstration

Potato Basket

Here's the live stream from 17 Dec, 2019 for those who asked for it on Discord

I wonder if it's actually a good indication on their confidence for 250 vs 250 battles?

Is it more than only a technical demo of UE4?

It's interesting to see how far they've come since then, maybe PI testers can share anything?

Marscelllo

48:54 - time stamp for the 5000 people in 4sqm

Potato Basket

Marscelllo wrote: »

48:54 - time stamp for the 5000 people in 4sqm

Thanks, for some reason the embedded players ignores the timestamp

Should be fixed now

Niraada

Steven said that they're extremely confident in their servers handling 250vs250, but that battles at larger scales (I think he said 500vs500) would prove more challenging.

Lazyactor

also tests like that are flawed because they dont account for the random latency differences from different users. Example being 1 player has 63ms, another has 89, another has 24, etc etc. Wouldnt seem like a big deal but the more people pack into one area the more stress each latency difference makes on the server. A lot goes into synchronization . That being said it does serve as a decent stress test i suppose.

mcknightrider

That's pretty cool. But this is also just people moving around and no attacks or spells. Still a positive sign though.

Monorojo

Theory is nice but need to see this in action, 5000 people running around aimlessly and doing nothing else is one thing but this won't have any practical use in game.

Potato Basket

Monorojo wrote: »

Theory is nice but need to see this in action, 5000 people running around aimlessly and doing nothing else is one thing but this won't have any practical use in game.

Theory? I wouldn't reduce to that extent, it does demonstrate at least some capabilities of the engine and it's porential

But yeah, it's not indicative of real world scenario with the netcode in play

StevenSharif

A couple things of note;

We randomize latency by hosting batches of 200 of these simulated players on separate servers from different aws hubs that then connect into the game server.

Movement updates are the most intensive aspect of network traffic for most multiplayer games (aside from heavy projectile/hitscan based shooters) but most mmos, the majority of the legwork on the network traffic is updating movement and position to all net relevant clients in the area.

We’ve made some significant progress and will continue to test this progress and make even more as we enter widespread testing phases. The road ahead will be perilous and bumpy. But will yield good progress for Ashes

Niraada

@StevenSharif particularly for things like castle and node sieges, can we expect real time (or regularly polled) network health monitors of participating members for raid/party panels? I'm sure we've all seen people get caught in a pvp slideshow, not always due to issues on the server's end, and knowing which teams have more stable connections can (and likely will) dictate strategy.

Having the tools to identify 'transit' issues like this would be very useful for group leaders (and members), and I think falls largely outside of the scope of things like combat trackers.