News:

One Minute Game Review by The Happy Friar: https://ugetube.com/@OneMinteGameReviews
Also on Rumble: https://rumble.com/c/c-1115371

idTech 4 (aka Doom 3 tech) Discord Server! https://discord.gg/9wtCGHa

Main Menu

(MCS) idtech4 netcode rewrite progress updates

Started by oneofthe8devilz, February 06, 2015, 08:20:01 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

oneofthe8devilz

Mars City Security: Taking Doom³ PVE (Coop) and PVP (Deathmatch) multiplayer combat to the next level and beyond


-==February 2015 Update==-

Hey to all MCS fans,

today we would like to showcase some of the new network code rewriting progress that we have been making with MCS.

The first video is showcasing MCS Version 1.1.8.5 Internet Gameplay recorded @ 1080p @ 60FPS, with a distance of ~2000 Kilometers between the dedicated server and client.



And for all the tech-geeks among us, we also recorded a second Client / Spectator video session, that pushes our new netcode to the limits with an average ~800ms ping delay between the clients and the dedicated server.



With this second hardcore latency scenario we want to make sure that all network states (like Player/AI entity positions, rotations and animations and all multiplayer relevant events such as Player/AI: weapon discharges, deaths, projectiles traveling and impacting and all concomitant effects are properly and smoothly replicated across all clients throughout the network regardless of the distance and location between dedicated server and clients on our planet).

There is still plenty of fine tuning to be done regarding the netcode and so hopefully the small crowdfunding campaign scheduled for Spring 2015 will help speed up that process during crunch time.
I got six little friends and they all run faster than you ;)


Check out our mods at
moddb or the SPS Homepage

wintch

Have you seen http://ra.is/unlagged/? Is that similar approach?
Cant wait to try it, as i am still stuck with native netcode :(

oneofthe8devilz

#2
Quote from: wintch on June 27, 2015, 10:55:12 PM
Have you seen http://ra.is/unlagged/? Is that similar approach?
Cant wait to try it, as i am still stuck with native netcode :(

We do not use "unlagged" or any other "backward reconcile" netcode methods... The MCS netcode has all client side prediction issues rewritten that were causing visual consistency and performance problems at high latency due to extrapolation and rollback re-prediction...
I got six little friends and they all run faster than you ;)


Check out our mods at
moddb or the SPS Homepage

BielBdeLuna


motorsep

Quote from: wintch on June 27, 2015, 10:55:12 PM
Have you seen http://ra.is/unlagged/? Is that similar approach?
Cant wait to try it, as i am still stuck with native netcode :(

Doom 3 BFG netcode has something similar for hitscan weapons.

VGames

This looks absolutely amazing. I have 2 gripes. The chaingun should be moved to the right a bit when in first person view and the push from the plasma rifle needs to be reduced so that baddies don't got flying when killed by it. Since those are really easy to fix I say u got a masterpiece on your hands.
Get the latest on Perfected Doom 3 here - http://www.moddb.com/mods/perfected-doom-3-version-500

wintch

#6
Are you crowdfunding something for a programmer's job?
Plans to release MCS sourcecode someday? Just asking :)

oneofthe8devilz

Quote from: VGames on June 28, 2015, 03:52:10 PM
This looks absolutely amazing.

Thx :)

Quote from: wintch on June 30, 2015, 02:18:36 PM
Are you crowdfunding something for a programmer's job?
Plans to release MCS sourcecode someday?

Yes and Yes.
I got six little friends and they all run faster than you ;)


Check out our mods at
moddb or the SPS Homepage

oneofthe8devilz

#8
Finally got around to create a direct and representative comparison video of the benefits of the MCS netcode over the default Doom3 one.

At peak moments in this video the MCS netcode client performs up to 725% faster compared to the default Doom3 netcode client  :o

(best watched in fullscreen @ 1080p @ 60 fps)

QuoteMCS Netcode Comparison @ 750ms Latency

This footage is a direct comparison video between "the default Doom3 netcode" and "the MCS netcode" demonstrating the clientside optimizations of "the MCS Netcode Rewrite Project" resulting in a massive raw client performance boost and a significant improvement and smoothing of the client gameworld replication at higher latencies.

Aside from the obvious increase in performance the replication of gameplay-events now gets properly and smoothly propagated clientside across the network at any latency and at any client performance most noticeable at the following points:

- AI now smoothly animates, moves, attacks and dies with proper and smooth ragdoll behavior across the network at any latency and at any client performance

- Projectiles ( AI & Players) properly render, travel and impact with all according impact, lighting and sound effects across the network at any latency and at any client performance

- Local and remote player actions and animations such as attacking, dying, platform-interacting, etc now properly and smoothly replicate across the network at any latency and at any client performance

- Physics simulations involving any interaction between players and/or AI now smoothly replicate across the network at any latency and at any client performance
I got six little friends and they all run faster than you ;)


Check out our mods at
moddb or the SPS Homepage

motorsep

Why not to make MCS vs Doom 3 BFG comparison ?  >:D

oneofthe8devilz

Quote from: motorsep on November 15, 2015, 06:27:44 PM
Why not to make MCS vs Doom 3 BFG comparison ?  >:D

Well if you'll rewrite the BFG client to be able to connect to a D3 1.3.1 dedicated server so I can run both clients side by side....

Anytime !
I got six little friends and they all run faster than you ;)


Check out our mods at
moddb or the SPS Homepage

motorsep

Why would you need that? Run BFG server, connect BFG client and do your recording. Then do the same with MCS. Same map, same assets, same ping. Should be pretty demonstrative. There is no need to connect D3BFG client to D3 1.3.1 server. Makes zero sense.

oneofthe8devilz

#12
Quote from: motorsep on November 15, 2015, 08:26:06 PM
Why would you need that? Run BFG server, connect BFG client and do your recording. Then do the same with MCS. Same map, same assets, same ping. Should be pretty demonstrative. There is no need to connect D3BFG client to D3 1.3.1 server. Makes zero sense.

That's complete nonsense...

First of all, stock BFG is lacking so many components (from the lack of AI synchronisation to the completely different way of shading and much more) to achieve "compare-ability" but even more shocking is that you seem to be oblivious to the basic principle of compare-ability...

Even if I would use two completely identical builds of MCS, recording two demos with identical settings and identical assets the results can and will be completely different due to the nature of the game's simulation and the micro differences that occur during the recording (input & simulation).

This is why the netcode comparison video uses 2 clients spectating a third one synced across the network to absolutely eliminate these factors and produce true compare-ability...

Ask one of your programmers and let him explain you the basic principles...  ::)
I got six little friends and they all run faster than you ;)


Check out our mods at
moddb or the SPS Homepage

The Happy Friar

Quote from: oneofthe8devilz on November 15, 2015, 10:00:51 PM
Even if I would use two completely identical builds of MCS, recording two demos with identical settings and identical assets the results can and will be completely different due to the nature of the game's simulation and the micro differences that occur during the recording (input & simulation).

I think that's what he wants to see.  How the differences effect the latency.  But you're right, no AI/coop capability in BFG means you can't compare the biggest changes.

motorsep

I am well aware that BFG doesn't have what MCS has. However, I was thinking about core networking in Team DM, 16 to 32 players on the server, comparing lag. To see whether BFG's lag / server load handing is any worse / better than MCS's.

Note that by recording a demo and the recording video from demo playback you are showing misleading results. I can run the game at 10 fps, but demo will always run at 60 fps (or 120 fps uncapped). I bet same goes for networking.