Why single thread is en of spring for me

[klapmongool]

In that thread it says about your replay:

no, this did not "go a lot better in 94.1" (you can reproduce it as easily there as in 95.0), and no, MT would not have mattered either.

So if you want to compare 94.1 performance to 95.0 it is a bad example.

That refers to the bug that happend at the 35 mins mark. I posted this replay because of the low fps before and after that. 94.1 MT did handle that better.

My point was not to judge the quality of games.
But the type of game must be considered when discussing performance.
I could try to play a speedmetal game with 50k units in 95.0 and then complain about the low fps.
Hope you agree that such game would not even work in 94.1?
And with the 8v8 example I feel same: I do not believe such game was ever played in 94.1 without players lagging out.

Ok, glad to hear you don't want to judge the quality. Of course the type of game affects performance. That is what this thread is about: Large games run worse on 95 than they did on 94.1 MT. Insanely large games with 50k units? No idea.

Consider this:
With modoption maxunits =1000 and 16 players, there can at maxium be 16 * 1000 units = 16.000 units.
Nothing is really stopping players to build so many units so it is in theory possible.
Now consider if a game with 16.000 units would be playable, 94.1 or 95?
If you think yes, then please record a video of 16.000 peewees fighting each other because I am sure that would look epic!

Actually 1000 units is not enough for me. I regularly hit these hard unit limits and I end up suiciding or self-ding units that I deem less useful at that time. Then I switch to different tactics that use less units because otherwise I can't play anymore. The unit limit thus forces me to change the way I play. I don't think that is a good way to shape a game nor do I think it is fitting for a large scale RTS like BA.

Thing is in small games (player wise) you might actually controll those 200-500 bombers. In a 8v8, never. 2500 units divided to 16 players is just 155 units per players. Now from those 155 units substract eco, substract turrets, how big will your army be? Maybe 50 units...

I actually do control hundreds of units at the same time, and so do many others of the better players. I don't know why you deny that. It is easy to see, for example in the replay I provided. Stop theorizing about games that actually are being played and watch them.

I want to add that I don't mean to say that 95 is bad or that I think it won't get better. It is a great engine and most stuff it does is awesome. The thing 94.1 MT did better was performance in massive games. I hope 96/97/98 will catch up to this. Also I think Spring should be easily usable for all game devs, not only *A games.

[dansan]

Replays of diff. eng. versions cannot be compared well. Is there a way to create almost the same match with different engine versions?

Something like a list of cheat-commands to create a set scenario (10 player bases, and 10x100 units) at start and then give the units move/attack orders... actually just like chickens does? ... would need a backwards-compatible game (mod)... most are?

Ofc the matches would differ slightly, but it'd be the best I can imagine to compare performance of different engines. (Maybe the devs already have that for regression tests?) Would be nice to have, so that we could move from "feels different" to "scenario X ran in 0.84 with 10/20/30 fps (min/avg/max) and in 98.0 with 20/30/100 fps", or maybe even with graphs like knorkes, to see what the boundaries are etc.
(I know one day I'll have to learn lua :)

[zwzsg]

Yes.

[smoth]

Replays of diff. eng. versions cannot be compared well. Is there a way to create almost the same match with different engine versions?

Something like a list of cheat-commands to create a set scenario (10 player bases, and 10x100 units) at start and then give the units move/attack orders... actually just like chickens does? ... would need a backwards-compatible game (mod)... most are?

Ofc the matches would differ slightly, but it'd be the best I can imagine to compare performance of different engines. (Maybe the devs already have that for regression tests?) Would be nice to have, so that we could move from "feels different" to "scenario X ran in 0.84 with 10/20/30 fps (min/avg/max) and in 98.0 with 20/30/100 fps", or maybe even with graphs like knorkes, to see what the boundaries are etc.
(I know one day I'll have to learn lua :)

Unpossible because mods are frequently updating and changing. Even if they don't want to, the slightest IMPROVEMENT to anything results in people crying over the lost of old behaviors, which they used to exploit etc.

*rantaboutplayers"feel'ofagame*
People talk a lot about feel and any time I read that I immediately discount them. We have proven within our modding community that you can tell players COMPLETE lies about stat changes and their perception will completely allow them to believe it works as such. Fang and I did thought experiments for years about it in our balance. Don't get me started on all the years of ignorant filth coming out of these so called pro players who don't even know what the numbers under the hood are like(big ups on the current BA team for fixing a lot of that shit~!)!
*/rant*

It would be nice if there was a repeat testable version of a project ! That project would have to be played and if it is played, that project needs to be maintained to keep the players happy otherwise they will leave it for a more actively patched project. At which point you lose all your player base to generate test cases etc.

it is kinda a catch 22 and a lose lose. Zwzsg himself once said that if a project stops being worked it is dead.

[dansan]

Unpossible because mods are frequently updating and changing.

Ofc old results could not be compared with new ones, but if a mod would be backwards compatible, then a new mod release (for a new engine release) could be rerun on old engine versions to create comparable results. Only results of the same mod-version are comparable.
The tests must be a little synthetic, because widgets of new mod versions don't work in old engines. So the test-mod must be a (as little as possible) stripped-down version to make it backwards compatible.
As an example I'd take a current BA/0k and remove all widgets&gadets that don't work with 0.88 and run that BA-mod on all engine versions for comparison.

[albator]

It is funny how much denial there is on this subject. 94.1 MT did give better performance with thousands of units and actions than 95 does.

Yeah thus 1).

Assuming that you meant to write ´end´ in the topic title: Good bye Alba. I'm sure you'll be back though. There is still no RTS alike out there.

Nah, I meant end of fun . But I barely play anymore cause of that, and also I would like to believe it will be fixed but I have doubts because of some denial of the community thus this thread. As a player I just wanted to impulse a direction where I would like it to go to, since I looks like barely any players hang out in the forum, thus my 2)

1) Why not bringing multi-thread back ?

Why no read forums? It's been asked a lot of times.

I did:

What happened to spring-multithreaded.exe though? I didnt come with the installer.

zerver (the only maintainer) refused to fix the problems it had.
.

etc...

Then I read:

thats wrong! he can! he just has to send a pull request to github with fixes / changes.

Not really. You banned me and then just hours later ripped out large amounts of code essential to MT, and this was accompanied by lots of evil commit comments about me. I am not contributing to any projects unless there is a certain degree of trust, and your actions didn't exactly help!

So I thought abma was being sarcastic implying devs dont care about mutithreading , other possibility is one of the 2 lies.

By then I read:

http://springrts.com/phpbb/viewtopic.php?f=1&t=31148
And I was involved as well.

So my opinion started to bent. Should I thrust more FPS or theory ?


Anyway, I am just glad if I made appear the FPS performance more of an issue than what it appears it was supposed not to be

[smoth]

Or maybe you should stop hunting for conspiracies when there are none.

The current multithreading is incomplete but will get better over time. Dude you didn't even KNOW it is. When people said mt was gone they meant zerver's code. Whatever.

[Forboding Angel]

I like how people act like Zerver's method of MT is the only method.

[knorke]

Knorke what are you using to graph those details?

widget+gnuplot: http://springrts.com/phpbb/viewtopic.ph ... 11#p551511
Widget is easy to use, so everybody can in theory log fps during games/replay and show difference between 94 and 95, or whatever.

Is there a way to create almost the same match with different engine versions?

For start just start spring.exe and type:
/cheat
/give 100 armpw 0
/give 100 armpw 1

Actually 1000 units is not enough for me. I regularly hit these hard unit limits and I end up suiciding or self-ding units that I deem less useful at that time. Then I switch to different tactics that use less units because otherwise I can't play anymore. The unit limit thus forces me to change the way I play. I don't think that is a good way to shape a game nor do I think it is fitting for a large scale RTS like BA.

So how else should units be limited? Not limits at all? Then lag becomes the deciding factor, that would be no fun. Well maybe one player with supercomputer has no lag, but the other 10 players have.
I think it is better game design if players maybe have to reclaim some solars to stay within a unit limit, instead of forcing them to play with lag...

[Beherith]

Hehe, unit cap should be a tradeable resource like co2 emission quotas. Where I can trade 50 cap for 5000 metal :).

[Jools]

Hehe, unit cap should be a tradeable resource like co2 emission quotas. Where I can trade 50 cap for 5000 metal :).

Except that CO2 trading doesnt work at all.

[klapmongool]

Actually 1000 units is not enough for me. I regularly hit these hard unit limits and I end up suiciding or self-ding units that I deem less useful at that time. Then I switch to different tactics that use less units because otherwise I can't play anymore. The unit limit thus forces me to change the way I play. I don't think that is a good way to shape a game nor do I think it is fitting for a large scale RTS like BA.

So how else should units be limited? Not limits at all? Then lag becomes the deciding factor, that would be no fun. Well maybe one player with supercomputer has no lag, but the other 10 players have.
I think it is better game design if players maybe have to reclaim some solars to stay within a unit limit, instead of forcing them to play with lag...

Well, if I spend a lot of resources on units that don't help me win there is a good chance the enemy can win with the useful units he made.

Of course there has to be a limit somewhere but imho that should be pushing the hardware limits. You can't expect to run late game on a dual core 1.3 laptop(fictional stats).

Often there is often only 1 really big player on a team and perhaps a 2nd player who is maybe half the size (in the games I play at least) in these massive games. With a unit limit of 1000 I don´t expect a 16000 unit game as you suggested earlier. More like a 2500-3000 units total. This is because most players already died when the game reaches this stage. Also, a lot of players simply aren't able to keep on growing so they never end up that big.

In conclusion:

No, the game should not be decided by lag. The game should be decided by the players' abilities to use the game mechanics to win. I don't think that in BA a unit limit should be taken into account by the player in strategical and tactical decisions. If it should it should probably be made into a resource (Spawn more overlords!).

[Forboding Angel]

Imo, BA should implement the population cap gadget that evo uses, and just change intrinsic supply to like 200 or so, and each mobile unit (not counting cons or comms, etc) to 1 supply, so in effect, each player can have 200 attacker units, but an unlimited (other than engine unit cap) number of buildings.

You could always raise or lower the total as needed, but with an 8v8, that is 3200 total attacker units on the field, which is a shitload, but at least there would be a decent limit.

I submit that a 100 attacker unit limit should be enough, because, lesbihonest... If you have 100 attacker units and are not attacking, you're doing something very very wrong.
[edited] dont be an asshole, -Beherith

[muckl]

"but an unlimited (other than engine unit cap) number of buildings."

will result in porce HLT/Annihilator/bertha spam

[klapmongool]

Imo, BA should implement the population cap gadget that evo uses, and just change intrinsic supply to like 200 or so, and each mobile unit (not counting cons or comms, etc) to 1 supply, so in effect, each player can have 200 attacker units, but an unlimited (other than engine unit cap) number of buildings.

You could always raise or lower the total as needed, but with an 8v8, that is 3200 total attacker units on the field, which is a shitload, but at least there would be a decent limit.

I submit that a 100 attacker unit limit should be enough, because, lesbihonest... If you have 100 attacker units and are not attacking, you're doing something very very wrong.
[edited] dont be an asshole, -Beherith

Yea, or not. 8v8 doesn't mean that 16 players reach their maxunits. It just means that a lot of players never get there and a few players get limited by the unit limit. And muckl is right about what happens then: building spam.

And I don´t think anyone needs to tell me how to win in BA tbh.

[AntiAllez]

Benchmark results with ... on ...

MT 96-20140112 vs Spring 96.0, BA EXTREME SPAM demo, SimFrame @5:00

i7 2600K, 4 cores HT, Windows
Spring: 308s
MT: 64s (380% faster)

FX 8350, 8 Cores, Windows
Spring: 362s
MT: 86s (320% faster)

i7 3630QM, 4 cores HT, Windows
Spring: 378s
MT: 89s (320% faster)

Core2Duo E6600, 2 cores, Windows
Spring: 639s
MT: 242s (165% faster)


btw: multi core is also a question for older and slower systems, it help to play endgames on ba and other mods with crappy cpu. gpu is not that question, but fps drop down by cpu if it is overwhelmed.

[klapmongool]

Impressive!

[Jools]

Very impressive indeed!

[malric]

@AntiAllez: could you please post links to sources/mod/scenario/whatever is needed to redo the tests? I would be curious to test also on my computers and see what I get...

[zer_ver]

Malric, you could try this:

<link to GPL violating software removed>

User was warned for this post.

I'm sorry, but you are well aware of the rules, and attempting to circumvent moderator action does you no favours - FLOZi