Ten Reasons why Nuclear Dawn is a Failure

[Vertex]

1. Server Performance: 4.0ghz (or higher) on dedicated haswell is required for 32 slots at 100% performance. This is becuase game servers only use one thread. When the game was first released, there was no such thing as haswell. This creates a condiction which hardly no one can host a server with high enough performance standards.

"Reducing the number of packets scheduled by the server helps." -UltimateX

2. Team Balance: Proper balance hasn't been possible for the last two years. The top skilled players in the game are approximately 16-18 times the level 80 cap. To test this, someone reset their level and reached 80 within three days. This can be solved with something like this. (prestige system 1-5[www.desmos.com]) More information here.

3. Server Instability: At this moment in time you cannot launch a game server on windows and expect it not to crash every few hours. (last game patch) Furthermore, there are memory leaks currently such as OnTakeDamage() and InitSpeeds() which limit server uptime to aproximately 72 hours. Lastly, sourcemod for developing plugins isn't as stable as it needs to be, the FakeNativeRouter() issue is one thing.

4. Folded Developers: The this link explains it all. It just got merged into this thread by a moderator. It was fully approved becuase steam filters were used to censor it.

5. Game Balance: This might have been good back afew years, but once skilled players found all the ins and outs things got ugly. The entire game is going from start to finish from someones base. What happens in the first five minutes decides who wins.

6. Game Bugs: There are afew issues which will crash your game client. Thirty miscellaneous bugs have been reported by Vertex currently in the issue reports section. Any more reports are uselessly time consuming at this point. (no sign of fixes)

You can learn more here -- how to clean afew gigabytes of dumps from game crashing.
"I run this when my Windows PC Logs in, you got to do what you have to on a SSD."

7. Game Client Performance: 3ghz on an intel quad or 4ghz on an amd quad is required to play this game under 60fps standards. There's a definate lack in optimization and not enough computing power is availible for the engine. A simple low effect detail default would help alot of clients tremendously. But there's still not enough mult-threading support.

8. Seperation of Skill Levels: New players are blended into a melting pot of competitive players and only afew end up surviving and reaching this exponential skill level. This is a side effect of number 1 to 6 which limits the amount of players on the game.

9. Locations: Currently this is limited to mostly to European and East Coast clients. There is no way possible to seed anywhere else currently. The further you get away the server, the less likely it becomes you'll be able to receive proper routing to play without lagg. Until it gets to the point where there's only afew clients left. Side effect of 1 to 6.

"I've tried updaterate 33 and 66 again yesterday on an overseas server and it was still equaly laggy regardless." -Post by UltimateX

10. Bad Community: This was more of issue in the past, which took alot of time and effort to resolve. It still isn't perfect to this date, but atleast things like harrasment are gone. 2012: *Rookie enters the commander chair* Veteran: "Get the ♥♥♥♥ out." 2015: *Rookie enters the command chair* Veteran: "Lets vote to demote guys" Veteran 2: "No one else stepped up, unless you want to command, lets give em a shot."

[kozec]

#3 is weird, I can't imagine anyone running game server on Windows for something more than close groups of friends.

And #7 is definitely nonsense, I've been playing ND when it was new thing with *Wine* on Dualcore i3 and GeForce 9800 GTX with, I think, >100 FPS.

[Vertex]

Originally posted by kozec:

And #7 is definitely nonsense, I've been playing ND when it was new thing with *Wine* on Dualcore i3 and GeForce 9800 GTX with, I think, >100 FPS.

Yeah, I wouldn't be suprised if you have 100FPS on an emtpy game. However, a full server bleeds a hell of alot more than a light game. (net_graph 4 shows triple)

Also, setting your effect detail to low is a way to get around #7. But it's defaulted to high which no one can really handle. Only a handfull of people end up changing it.

Originally posted by kozec:

#3 is weird, I can't imagine anyone running game server on Windows for something more than close groups of friends..

I can't imagine the flexibility that running a gameserver on Windows would be provide. When you enter a Linux only situation, that's 50% of the servers gone/unstable right away.

Alot of hosting providers will not support Linux and server operators will have no clue how to use it. Redstone would be on 4.0ghz dedicated haswell right now if Windows was stable.

[amarkula]

Originally posted by kozec:

And #7 is definitely nonsense, I've been playing ND when it was new thing with *Wine* on Dualcore i3 and GeForce 9800 GTX with, I think, >100 FPS.

Ummm, I'll be swapping my gtx970 to that or similar if you get 100. :P Sometimes even with this it's like walking on tar pits. :D

but well, like to try it on linux, never had the time... it's good right / runs well?

btw, Vertex how's #1 with virtualisation, like wrap it with vmware player (or better) with 2 - 4 windows/linux slots in it? server is not using graphics card, right?

[Vertex]

Originally posted by amarkula:

btw, Vertex how's #1 with virtualisation, like wrap it with vmware player (or better) with 2 - 4 windows/linux slots in it? server is not using graphics card, right?

2-4 slots with not bots can be run on much pretty any processor.

The server doesn't use the graphics card no.

20 slots at 3.6ghz haswell xen is what i'm benchmarking currently for maximum performance in the higher player count ranges. Dependent on how many entities. (buildings)

[RlyDontKnow]

Personally I'd add a reason #0 which made ND a great game from my pov but also made it fail so much *imho*:
ND is more of an RTS than an FPS. I.e. a lot more emphasis is put on the commander than it is on the rest of the team which barely matters unless the commanders are on an even skill level.

On one hand this was a vast amount of fun in a competitive setting with carefully picked teams back in TAW (and against the ND rus guys) where everyone knew the game inside out and everything boiled down to coordinating your foot soldiers ensuring your commanders plan can go through while slowing down the opponent's to a crawl (or quickly adopting in case things go against the plan).

On the other hand this was also the reason things went completely fubar in public matches due to a lack of good enough commanders (or an excess, depending how you look at it). In any case it was quite rare to find a game that was even close to what you'd experience in the competitive scene. Most public games went as pretty much one of
1) one commander finishing it early on with a quick and hard push
2) the whole thing turning into bunkering down and failure to do a final push

tl;dr: the public matches were nothing even remotely resembling what is/was fun about the game and usually ended up being boring or pointless for 90% of the team making the income of new players to actually get into the game way too small to seed a suitable player base

for your actual points:

#1: can't say anything about it as I didn't run the server, there were issues at times, but it didn't feel like a major game stopper tbh. also: 32 slots seems overkill, everything beyond 20 slots usually turned into chaotic meh games.

#2: I personally think it's a consequence of #0 rather than actual balance issues

#3: same as #1

#4: why don't you have a #4? that makes it only 9 reasons, doesn't it?

#5: true if you have coms that blatantly abuse obvious bugs (building outside the map, in buildings and in the sky comes to my mind there) - beyond that I can't really recall any major issues

#6: yeah there were some quirks, but nothing overly disruptive tbh (put aside the map issues hinted in #5)

#7: never had issues with it nor had anyone else I played with (ran a Phenom II x4 950, HD 5770 and 8GB RAM back then)

#8: I think this is again #0

#9: lack of players, see #0 :P

#10: this is a prime example of #0 - the reason noobs got kicked out was because everyone knew how important the com is. however how should they get better if they keep getting kicked out? who volunteers if you're always on the edge of being kicked? the few public matches I had where the enemy com actually slowed down a whole lot and tried to guide the opponent after recognizing he's new to commanding were my personal gems (that's what the game would've needed a heapload of!)


tl;dr: a great game idea dashed against the reality of the contemporary PC player community

[Vertex]

Originally posted by RlyDontKnow:

Personally I'd add a reason #0 which made ND a great game from my pov but also made it fail so much *imho*:
ND is more of an RTS than an FPS. I.e. a lot more emphasis is put on the commander than it is on the rest of the team which barely matters unless the commanders are on an even skill level.

Natural Selection 2 is the same type of game and is successful.

If you load a server up with only rookies, it doesn't even matter how good the commander is, so long as forward spawns are built.

This problem exists when you mix rookies with competitive players.

If the server has only competitive players, pretty much any commander will do and the team will completely decide the game.

Originally posted by RlyDontKnow:

#1: can't say anything about it as I didn't run the server, there were issues at times, but it didn't feel like a major game stopper tbh. also: 32 slots seems overkill, everything beyond 20 slots usually turned into chaotic meh games.

How many slots you legitimately need depends on the map and skill levels. For instance, in a metro senerio when the average person is level 80 twice over, 24 is the maximum you can do without a grind. Whereas, clocktower with all new players requires 30. There's also your typical 24 corner or 32 oil field based on map sizes.

Originally posted by RlyDontKnow:

#7: never had issues with it nor had anyone else I played with (ran a Phenom II x4 950, HD 5770 and 8GB RAM back then)

I had 90 fps on my 4850, what matters is the cpu. quad core @3.2 ghz amd should hold up near 60fps perfectly fine with your favourited 20 player count. But the client performance issue is mainly triggered from running high effect detail instead of low. Few processors can handle it and it's defauled to high for virtually everyone.

[kozec]

Originally posted by ᵡᴳḽᴰ::Vertex:

Originally posted by kozec:

#3 is weird, I can't imagine anyone running game server on Windows for something more than close groups of friends..

I can't imagine the flexibility that running a gameserver on Windows would be provide. When you enter a Linux only situation, that's 50% of the servers gone/unstable right away.

Alot of hosting providers will not support Linux and server operators will have no clue how to use it. Redstone would be on 4.0ghz dedicated haswell right now if Windows was stable.

Yeah, me neither :) But in all seriousness, I never saw anyone providing any kind of game hosting platform using Windows. How that would even work...

@amarkula Well, it was under Wine, so I'd guess some FPS-heavy effect was not supported. I can test it with native version when I come home.
And no, ᵡᴳḽᴰ::Vertex, it was not on empty server. Those were rare back in those good times :(

[RlyDontKnow]

Originally posted by ᵡᴳḽᴰ::Vertex:

Natural Selection 2 is the same type of game and is successful.

After having tried it for quite a while (with the same guys I played ND with as we moved collectively to NS2) I personally don't think that's really true. It doesn't share much with ND besides being a RTS-FPS mix.

NS2 has a very different pace from ND, is pretty much completely focused on CQC, highly assymetric, individual performance of FPS soldiers counts a whole lot more than it does in ND and they also have more ways to do things without the help of their commander. I personally didn't like it and still perfer ND over NS2, but I do see why it'd be more appealing to a wider audience. (PS: from my experience that community has pretty much the same attitude to commanders that ND had back then - know what you're doing or get lost)

[McFlurry Butts]

The way people all went in the same direction consistantly rather than trying to flank was annoying and made the game boring and undynamic.

[Arsonist]

Community
The community is why I quit playing this game. Commanders always take the blame even when its the ground troops that suck. You can supply your team with ammo stations and turrets guarding them, but turrets dont last forever and if your team gets steamrolled trying to counter attack, its always the commander that gets blamed for it when their team has bad aim. Commanders are treated as nothing more than a scapegoat when their team losing.

If your team is too broke, you can tell your team to capture points but most of the team doesn't listen. But then when they start ♥♥♥♥♥♥♥♥ they need a new forward spawn, supplies, more turrets, or something expensive got wrecked and needs rebuilding and if you can't afford to do so quick enough, commander is blamed as if team not capturing points had anything to do with the failure.

Commanders who are actually bad everyone wants to keep in command such as no supplies, turrents, no forward spawns, no advanced weapons and structure research even though your team is sitting on about 50 G's of resources.

Making new content for the game, the community tends to be at its rudest. If your unfished map isn't eye candy yet or has a few balance flaws, you can get some very disrespectful comments. Feedback is a good thing but not when its in that form.

I only play TF2 now. Don't really have near as much hassle with team co-operation on it and feedback people give when you make maps is rarely ever snarky/rude.

Server Performance
There is a server cvar for all source games that enables some form of parallel processing which is disabled by default. Theres also a cvar to change or force a particular threading mode.

[Vertex]

Originally posted by ₪ UltimateX ₪:

Server Performance
There is a server cvar for all source games that enables some form of parallel processing which is disabled by default. Theres also a cvar to change or force a particular threading mode.

Do you happen to know what that cvar is? And wether or not it's valid on Nuclear Dawn?

[Arsonist]

They all should be valid. These are engine related settings. Every game using the source engine would and should have these in common. Try them and find out. Be aware if you get an unknown command response in server console, it doesn't necessarily mean it doesn't exsit. Sometimes in source games some cvars are hidden and you need a mod to unlock them. Usually involves changing the cvar flag value. Sometimes you may need a memory hack mod to break unreasonable restrictions on things. IMO control freak develpoers need to lay off and give server owners all the freedom they want to change things.

Threading
"host_thread_mode"
- Run the host in threaded mode, (0 == off, 1 == if multicore, 2 == force)
0 appears to be the default in SourceDS

sv_parallel_sendsnapshot
sv_parallel_packentities

Found about about these in the NFO hosting forums. Keep in mind the engine's parallel processing is limited and imperfect so don't expect load distribution between cores to be completely even.

Rates
Also lowering sv_maxupdaterate could possibly lower some stress on the cpu by not having to prepare as many packets. Doing packets every tick is redundant and interpolation (the client side predection) does a pretty accurate job guessing between updates from server. I play with updaterate set to 33 and notice no difference in smoothness from 66 considering TF2 is more faster paced than both CS and ND. For ND updaterate 66 is most likely overkill. Updaterate of 33 uses 50% less bandwidth too.

Packet compression
// Packet compression can save some bandwidth but at the cost of more time on the CPU
sv_compressstringtablebaselines_threshold
net_compresspackets_minsize //If packet is larger than this amount, compess it. Default is 1024 bytes.

net_compressvoice //These ones take a boolean value
net_compresspackets

Other
sv_alternateticks //If on, world is only simulated every other server frame. Off by default. Use only if CPU usage is a serious issue.

[Vertex]

Originally posted by ₪ UltimateX ₪:

Rates
Also lowering sv_maxupdaterate could possibly lower some stress on the cpu by not having to prepare as many packets.

Rates should only be lowered for clients facing a threat to higher bandwidth amounts. Packets for these clients can often travel long distances on congested carriers. Lag compensation is not a good thing specially with higher latencies.

Originally posted by ₪ UltimateX ₪:

processing is limited and imperfect so don't expect load distribution between cores to be completely even.

Processing will never be split evenly between cores. There will always be that main thread doing the majority of the processing, maxing out in the worse case scenerio.

Originally posted by ₪ UltimateX ₪:

Packet compression
// Packet compression can save some bandwidth but at the cost of more time on the CPU

Packet compression has to stay high due to the threat of congested carriers for overseas clients.

[Arsonist]

If you already know about all this stuff then why did you ask me to post the cvars?

Rates should only be lowered for clients facing a threat to higher bandwidth amounts. Packets for these clients can often travel long distances on congested carriers. Lag compensation is not a good thing specially with higher latencies.

You shouldn't really cater your server for players overseas anyway. Setting of 33 or 66, its still going to lag like hell if your ping is too high to begine with. From my experience playing on most overseas servers, I felt no drawbacks using a lower setting, no movement delays and no goofed up hit registration. I only mentioned this as an option as it takes CPU to process packets and lowering updaterate doesnt really hurt anything. Forcing lower rates on clients gloabally can be necessary if server resources are getting tied up hence why theres a server version of the cvar for it to clamp rates which holds priority over client defined settings.

Processing will never be split evenly between cores. There will always be that main thread doing the majority of the processing, maxing out in the worse case scenerio.

Yeah thats what I said.

Packet compression has to stay high due to the threat of congested carriers for overseas clients.

I've never said anything about lowering or raising the values. Again just tossing options out there as potential CPU load reductions as compression adds more time on CPU for packet handling. But let me just say, whats more important when server CPU is getting maxed? Lagging overseas players or lagging all players? In such case, foreigner satisfaction should be the least thing to care about. Put your local player base 1st

[Vertex]

Originally posted by ₪ UltimateX ₪:

In such case, foreigner satisfaction should be the least thing to care about. Put your local player base 1st

Originally posted by ₪ UltimateX ₪:

Forcing lower rates on clients gloabally can be necessary if server resources are getting tied up hence why theres a server version of the cvar for it to clamp rates which holds priority over client defined settings.

There's better ways to make performance optimizations without shutting out certain types of players. Forcing the highest possible rates is always the best for performance.

Something small overloads the main thread with the last four slots, afew other optimizations should do it. ie. Looking into the threading mode currently.