Spring Mark

[lurker]

Snipawolf: while you're right that one person's "lag" won't hurt anyone else, one person having their CPU overloaded will slow down the game. It's set to slow down so that no one drops out of game from a slow CPU being pushed past its limit. I slow down games all the time with my 1.5Ghz pentium 4...

That said, springmark sucks.

[SwiftSpear]

Lurker is correct

Spring runs at the speed of the slowest PC in the game. Even if the slowest PC is a spectator. There is definitely a valid reason to kick slow spectators, that being said, it's highly disrespectful to kick all spectators before they have even gotten into the game. Habitual spectators should know the limitations of their PC and not join games larger than they could handle playing because they will lag the players even if they themselfs are not playing.

That being said. Having a higher number of spectators does not make a difference at all. The amount of upload/download bandwidth spectators create is negligible, because spring's multiple simulation system of managing the game is extremely efficient for bandwidth uses, and the spectator doesn't submit and status changes to the game, only player do. Also, the spectator, having no units, does not increase the CPU usage of the other players at any greater amount than the pure text spectators send in their user name and ingame messages. If you can host an 8v8 game withouit spectators, you can host an 8v8 game with 40 spectators as long as every spectator in that game has the CPU speed necessary to keep up with the game simulation.

It's VERY VERY rare in spring to acctually hit the bandwidth limitation, and when it happens it's basically unheard of that it's spring's fault. The VAST majority of the time lag is created in game it is created by the CPU of one of the players or spectators.

[jj]

Springmark =

Period

[REVENGE]

Springmark =

Period

Wait, dammit that is weak, I have to find a better one.

|Searches for an hour|

Ok Found it.

[Masure]

I loved to read Snipawolf believing the game won't slow down cause of cpu limit.

I love people who are 100% violently sure and humiliated 2 posts later.


I played a big game last sunday. I didn't discrimate over cpu power and ran the game. After 20mn, two players were cpu limited (45% cpu usage) and the game was running at 0.6 speed. Many other players started to complain about the game speed and asked the slowers to be kicked. What I did. The game then ran fine.

To sum up, the 2 slow players wasted 30mn of their time in a game they couldn't play till the end. I've been insulted in PM cause they couldn't understand why they had been kicked, I couldn't explain because I was in game. All the people playing got a quite unbalanced game because of the 2 players kicked.

When I kick the slow players before the battle starts, I can explain why, the game run fine for all. Slow players can join a smaller battle that will run fine too.

[SwiftSpear]

I loved to read Snipawolf believing the game won't slow down cause of cpu limit.

I love people who are 100% violently sure and humiliated 2 posts later.


I played a big game last sunday. I didn't discrimate over cpu power and ran the game. After 20mn, two players were cpu limited (45% cpu usage) and the game was running at 0.6 speed. Many other players started to complain about the game speed and asked the slowers to be kicked. What I did. The game then ran fine.

To sum up, the 2 slow players wasted 30mn of their time in a game they couldn't play till the end. I've been insulted in PM cause they couldn't understand why they had been kicked, I couldn't explain because I was in game. All the people playing got a quite unbalanced game because of the 2 players kicked.

When I kick the slow players before the battle starts, I can explain why, the game run fine for all. Slow players can join a smaller battle that will run fine too.

Ok... that being said, springmark does a pretty poor job of providing useful information to relate to weather a player will preform poorly or not.
A: it's easily spoofable
B: it doesn't acctually benchmark spring, it just arbitrarily decides that good hardware = good performance, which is wrong for many many reasons
C: people running background shit can cause the game to slow, and springmark ignores that entirely
D: people with crappy PC's sometimes have a nack for setting them up in such a way where they get maximum performance where they need it.

Springmark isn't really a spring benchmark, it's an arbitrary hardware evaluation system, it doesn't really have any relation to spring at all aside from that it's for some very weird reason integrated into one of our third party lobbies.

[Masure]

A: it's easily spoofable

True. But I like to believe in humanity goodness. You ll be true if you say I shouldn't (see AF's posts always talking about spoofing)

You will always be able to spoof a system but it's totally fool when it is useful for the community. I use Satirik's client group highlighting to remember players I don't want to play with.

IMO spending 10x hours to develop a bulletproof hardware inspector is waste for 1% of spoofing fools.

Having the best hardware displayed is not the point, it only allows players to play homogenously. Players who can't understand that and spoof deserve to be blacklisted like I do.

B: it doesn't acctually benchmark spring, it just arbitrarily decides that good hardware = good performance, which is wrong for many many reasons

I don't care about the mark bulletproof accurary. I often don't use the mark (neither dthe autokick on score) but read the cpu brand & model instead. If I read P4 1.8, I ask the player to leave the big game I've planned to host because he will slow the entire game. If he doesn't want, I kick him.

I don't care about a gfx benchmark, neither a ping benchmark.

People can run at 15fps if they want. They can have a bad ping too if they don't massively lag in a 131425637673ms ping.

C: people running background shit can cause the game to slow, and springmark ignores that entirely

Once again, you can't control the whole comp, the whole internet issues, the whole electric issues, the whole big shot in the world.

Springmark doesn't aim to that at all, true. But IMO, going in this way is utopic, pointless and a big waste of time.

If people have a shitty comp, dl pr0n while playing, have an evil little brother pushing reset, a mother who unplug the modem at 11:00pm, you can't fight.

D: people with crappy PC's sometimes have a nack for setting them up in such a way where they get maximum performance where they need it. IMO

same as before

Springmark isn't really a spring benchmark, it's an arbitrary hardware evaluation system, it doesn't really have any relation to spring at all aside from that it's for some very weird reason integrated into one of our third party lobbies.

Springmark is not a real spring benchmark, just a cpu one. It is not perfect and should be improved to bench the cpu if you want.

It's integrated in satirik's lobby client cause it provides information we couldn't have with the Freq column in official client.

[SwiftSpear]

A: it's easily spoofable

True. But I like to believe in humanity goodness. You ll be true if you say I shouldn't (see AF's posts always talking about spoofing)

You will always be able to spoof a system but it's totally fool when it is useful for the community. I use Satirik's client group highlighting to remember players I don't want to play with.

IMO spending 10x hours to develop a bulletproof hardware inspector is waste for 1% of spoofing fools.

Having the best hardware displayed is not the point, it only allows players to play homogenously. Players who can't understand that and spoof deserve to be blacklisted like I do.

Goodness of humanity is not a reality. It goes both ways at the best of times. If springmark ever catches on large scale, there will be many players who spoof to avoid getting autokicked, and there will be many players who autokick and manually kick of very arbitrary and unfair qualifications. Honestly, I think it's far better not to entertain that in the first place, especially with springmark being as inaccurate as it is. If it were accurate and could reliably predict "this game is too big for this computer" then you'd have a case. Hell, I know my CPU isn't that great, but I probably tend to overestimate what it can handle if I really want to play a game. If the software suite were to accurately and reliably tell me "um, hey, no this game will be too much for you" then that would be much easier to accept, and I'd be much less pissed off at the game host who kicked be based on that advice. Springmark isn't even remotely close to that. Hosts kick too prematurely, and in a very disrespectful manner in spring, and players tend to want to play, no matter who has to have a slow, screwed up game because of it. Springmark is a contributor to that problem.

B: it doesn't acctually benchmark spring, it just arbitrarily decides that good hardware = good performance, which is wrong for many many reasons

I don't care about the mark bulletproof accurary. I often don't use the mark (neither dthe autokick on score) but read the cpu brand & model instead. If I read P4 1.8, I ask the player to leave the big game I've planned to host because he will slow the entire game. If he doesn't want, I kick him.

I don't care about a gfx benchmark, neither a ping benchmark.

People can run at 15fps if they want. They can have a bad ping too if they don't massively lag in a 131425637673ms ping.

The problem is that you tend to be the minority there, and you're also not the player in the position of wanting to play the game.

C: people running background shit can cause the game to slow, and springmark ignores that entirely

Once again, you can't control the whole comp, the whole internet issues, the whole electric issues, the whole big shot in the world.

Springmark doesn't aim to that at all, true. But IMO, going in this way is utopic, pointless and a big waste of time.

If people have a shitty comp, dl pr0n while playing, have an evil little brother pushing reset, a mother who unplug the modem at 11:00pm, you can't fight.

This is the whole point though. Spring already reports CPU speed, springmark isn't any improvement, it's scoring algorithm is crap.

Springmark isn't really a spring benchmark, it's an arbitrary hardware evaluation system, it doesn't really have any relation to spring at all aside from that it's for some very weird reason integrated into one of our third party lobbies.

Springmark is not a real spring benchmark, just a cpu one. It is not perfect and should be improved to bench the cpu if you want.

It's integrated in satirik's lobby client cause it provides information we couldn't have with the Freq column in official client.

Sure, that's valid. I don't have anything against springmark displaying more accurate CPU stats than spring client. The problem with springmark is everything else it does, or rather doesn't. It's being used as a comprehensive benchmark tool, when simply put, it's not even CLOSE to that good. Honestly, looking at springmark scores VS actual in game performance of the players with those scores, springmark might as well just be generating random numbers and autokicking based on that. It's scoring system is utter shit in relation to how spring runs.

This is the whole complaint. The tool isn't useful enough to be used the way it is being used. I'm getting handed a hammer and being told to screw in screws, and what I'm saying is, fuck that, develop a frigging screwdriver already.

[Satirik]

This is the whole point though. Spring already reports CPU speed, springmark isn't any improvement, it's scoring algorithm is crap.

you don't read (or you have a goldfish memory) or you are stupid or you don't understand anything to cpu core2duo 1.8 = good cpu, athlon xp 1800 = bad cpu !!!! but you will get 1.8 for both of them ... nobody is thinking on this topic or what ?

[Pxtl]

Alternately, modders could work on designing unit scripts that don't involve polling the whole unit-list once per frame for each unit just to fire a single proximity event.

[imbaczek]

Such scripts are possible today with LuaCOB.

[Masure]

I made some tests involving 1500 kbots to see when it was slowing down.

It's not surprising to see that pathfinding and shooting trajectory computing was the cpu hog.

With 1500 kbots without move order, the game ran perfectly smooth (1:1) but if I move them all I can slow down to (0.35:1). You can have 1500 kbots with only 50 moving without any slow down.

core2duo @3600

It was a quick test without true analysis just to see if pure high unit number could matter. Definitively not.

[Peet]

Springmark =

Period

So you're saying that lag is not possible when everyone has a good springmark?

[SwiftSpear]

This is the whole point though. Spring already reports CPU speed, springmark isn't any improvement, it's scoring algorithm is crap.

you don't read (or you have a goldfish memory) or you are stupid or you don't understand anything to cpu core2duo 1.8 = good cpu, athlon xp 1800 = bad cpu !!!! but you will get 1.8 for both of them ... nobody is thinking on this topic or what ?

STRAWMAN

I clarified that issue exactly one point below.

Sure, that's valid. I don't have anything against springmark displaying more accurate CPU stats than spring client. The problem... ...It's scoring system is utter shit in relation to how spring runs.

Searching for a weak point who's meaning you can twist is not arguing, it's just stupid.

[Neddie]

It's been like that most of this thread.

[AF]

you don't read (or you have a goldfish memory) or you are stupid or you don't understand anything to cpu core2duo 1.8 = good cpu, athlon xp 1800 = bad cpu !!!! but you will get 1.8 for both of them ... nobody is thinking on this topic or what ?

Youll get 1.81Ghz for the core 2 duo and 1.4 or 1.2Ghz for the athlon 1800. 1800 just means its equivilant to a 1.8Ghz intel pentium 4 cpu of that era.

In the mean time swiftspears has shown that someone here has tried to manipulate the situation. Isnt it about time this thread was locked?

[SwiftSpear]

you don't read (or you have a goldfish memory) or you are stupid or you don't understand anything to cpu core2duo 1.8 = good cpu, athlon xp 1800 = bad cpu !!!! but you will get 1.8 for both of them ... nobody is thinking on this topic or what ?

Youll get 1.81Ghz for the core 2 duo and 1.4 or 1.2Ghz for the athlon 1800. 1800 just means its equivilant to a 1.8Ghz intel pentium 4 cpu of that era.

In the mean time swiftspears has shown that someone here has tried to manipulate the situation. Isnt it about time this thread was locked?

Oh god, if only I could make argumentative fallacy a valid reason to lock a thread/ban a user. That would be heaven...

[Satirik]

you don't read (or you have a goldfish memory) or you are stupid or you don't understand anything to cpu core2duo 1.8 = good cpu, athlon xp 1800 = bad cpu !!!! but you will get 1.8 for both of them ... nobody is thinking on this topic or what ?

Youll get 1.81Ghz for the core 2 duo and 1.4 or 1.2Ghz for the athlon 1800. 1800 just means its equivilant to a 1.8Ghz intel pentium 4 cpu of that era.

In the mean time swiftspears has shown that someone here has tried to manipulate the situation. Isnt it about time this thread was locked?

... WRONNNNNGGGGGGGGGGGG you'll get 1.8 with an AMD Atlhon XP 1800 !

once again ... i say the truth ! IM SO GOOOD!

[REVENGE]

I think overall, Springmark is a great idea, but the current implementation is shiit. [No offense jj ]

When I first ran Springmark, I half expected it to be a benchmark that runs a demo with increasingly more units in Spring itself, then use a scoring algorithm based on the resulting performance to give a result.

That was not the case, but that doesn't make sense. Shouldn't real ingame performance be used as a benchmark for real ingame performance?

Ideally, there should be separate tests scoring three elements: CPU, GPU, and network performance. CPU and GPU should be Spring engine based, while network should request a speedtest from online providers.

The program should then give a detailed report with seperate scores per category, and finally an overall combined Springmark. Suggestions to improve performance would be ideal.

Instead of bitching about how current Springmark sucks, we should be discussing ways to make future Springmark better.

[Peet]

Instead of bitching about how current Springmark sucks, we should be discussing ways to make future Springmark better.

The problem with that is convincing them that it sucks