Spring Mark - Page 8

Spring Mark

Please use this forum to set up matches and discuss played games.

Moderator: Moderators

Locked
User avatar
Neddie
Community Lead
Posts: 9406
Joined: 10 Apr 2006, 05:05

Post by Neddie »

Actually, I have no idea what age you are. I assume you're around the same age I am - doesn't mean I can't be sick of the values instilled in my peers. The number of people who simply take the first solution and never do the backend work they should, making the world more difficult and complex in the end through laziness offends me, and frankly, I'm allowed to express this after beating my head against people like you for years.

The point is that the data I gave you is what SpringMark seems to look at. You won't know why it isn't performing well, and you won't be able to address it. This utility isn't completely useless, but it isn't good enough for the purpose it is designed to accommodate. If it isn't good enough, it must be revised - how blind can humanity be to every little issue like this?
Satirik
Lobby Developer
Posts: 1688
Joined: 16 Mar 2007, 18:27

Post by Satirik »

neddiedrow wrote:Actually, I have no idea what age you are. I assume you're around the same age I am - doesn't mean I can't be sick of the values instilled in my peers. The number of people who simply take the first solution and never do the backend work they should, making the world more difficult and complex in the end through laziness offends me, and frankly, I'm allowed to express this after beating my head against people like you for years.

The point is that the data I gave you is what SpringMark seems to look at. You won't know why it isn't performing well, and you won't be able to address it. This utility isn't completely useless, but it isn't good enough for the purpose it is designed to accommodate. If it isn't good enough, it must be revised - how blind can humanity be to every little issue like this?
why everytime you say something it is just shit ?
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

neddiedrow wrote:Actually, I have no idea what age you are. I assume you're around the same age I am - doesn't mean I can't be sick of the values instilled in my peers. The number of people who simply take the first solution and never do the backend work they should, making the world more difficult and complex in the end through laziness offends me, and frankly, I'm allowed to express this after beating my head against people like you for years.

The point is that the data I gave you is what SpringMark seems to look at. You won't know why it isn't performing well, and you won't be able to address it. This utility isn't completely useless, but it isn't good enough for the purpose it is designed to accommodate. If it isn't good enough, it must be revised - how blind can humanity be to every little issue like this?
So I think we only disagree on "how well does it evaluate the ability of a CPU to run a big game"

I'll make it short. When I see 1.8Ghz in cpu speed column, I cant say if it's a poor P4, Athlon XP or C2D E6300. With the spring mark column I can see that. As a C2D will perform twice better as the Athlon XP for the same displayed freq, springmark gives me useful data.

What can you say about my usage of springmark ?
Lippy
Posts: 327
Joined: 16 Jul 2006, 00:24

Post by Lippy »

lolz, my hardware's been banned from jobjol. :) No greenfields for me!
User avatar
AF
AI Developer
Posts: 20686
Joined: 14 Sep 2004, 11:32

Post by AF »

Satirik wrote:why everytime you say something it is just shit ?
A moderator needs to remove that post and give satirik a warning, this is a gross violation of forum rules.

Masure:

Its very important. Some cpus have reverse hyper threading running single threaded programs on 2 cores. Also there is no guarantee that spring will be ran on the first cpu core. It may even switch between cores during execution. A dualcore multithreaded benchmark is necessary.

On top of that, many users have horrendous practices and think its ok to have 120 background processes despite being able to get away with a meager 19 processes including browser and media player.

This tool just isn't good enough. It doesn't test network connections either or memory bandwidth, or HD usage. Argh with his supercomputer running at a snails pace because his HD chugging away would have a huge spring mark score.
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

You're not false AF but you harden the problem more than necessary.

You'll never be able to catch all possible issues. You can't monitor and predict the whole comp, the whole internet connection behaviour, the whole world disasters. And even if you could, going too far is not good.

We have a good proverb in french that describe what you re doing :

Le "mieux" est l'ennemi du "bien" > the "Better" is the ennemy of the "good"


You re trying to globalize all game issues when I say I don't want to play with poor cpus in big games.

You're wrong thinking someone with 500ms ping will screw a game. You can run a game fine with a 500ms average ping.

You're wrong thinking someone with 10fps will screw a game. It doesn't affect other people since the game can run 1:1 speed with 10fps (try it man)

You're wrong thinking a one-time lag wil lscrew a game. I can bear people who sometimes lag because they forgot a background application or have a 5s internet issue.

You're wrong thinking a super global benchmark (ping, lag, fps, cpu) will give the answer to the question :

"Can I really play with this guy on this big game or is it going to be a 0.4 speed game ?"



I think you've understood my point from the beginning and internally agree with it. But as you can't agree now with me by honor (don't tell me you haven't), you're trying to globalize it in a huge and big shit that will never be solved. You can't mesmerize me with this kind of args, i'm sorry.


Springmark answers my simple question.

You're only giving questions without any future possible answer.
User avatar
AF
AI Developer
Posts: 20686
Joined: 14 Sep 2004, 11:32

Post by AF »

No, springmarkis a global benchmark and as such hsa numerous flaws.

An alternative that provides much better benchmarking is available, average in game frame rate, average in game cpu usage, and it can be continuously updated and maintained to give a true reliable representation of the computers performance.
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

Ok you didn't read me

go to hell :wink:
User avatar
AF
AI Developer
Posts: 20686
Joined: 14 Sep 2004, 11:32

Post by AF »

I did read your post.

Springmark attempts to benchmark the cpu and arrive at a value about how good a players entire system is from that. 3D Mark and other more reliable benchmark program are better suited to this.

Instead it should benchmark by looking at the actualy measurable performance of spring itself.

What you're saying is that I'm making the problem a bigger deal by pointing out its flaws, aka "Itll never be good enough". Indeed this will never be good enough because its flawed conceptually, and there are other methods which are more appropriate, such as maitnaining an average fps and cpu usage rating based on data gathered directly from spring, ensuring consistent ratings that're far more accurate and are directly relevant to spring banchmarking.

CPU benchmarking is indirect benchmarking of spring, and as such is unreliable. Only direct benchmarking is reliable, thus only information gatherd from spring and about spring can be truly accurate.
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

I can point the lack in your benchmarking suggestion.

Averaging cpu usage (I'll take the fps appart if you don't mind cause it's another disagrement we have) should consider the game kind. you don't talk about that.

Player 1 has a uber comp and plays only 6v6 1 hour games.
Player 2 has a poor comp and plays 1v1 small divide 30mn games.

Player 1 will have a bigger avg cpu usage than Player 2. It's not accurate.

Ok lets say we can make the benchmarking better. Take into account the kind of the game.

How do you make the law between real spring cpu needs and the game usage you can average ?

Are you gonna trace the whole spring source ? Are you gonna make the correlation between cpu hogs causes (path finding ? shooting ? ...) and the cpu usage records ? Are you gonna make an adaptative mark that take into account the map general strategy, the known players strategy (micro ? spam ? porce ?) to give the final mark ?

I don't think so and saying so would be piece of shit, I'm really sorry.

That's why I can say "Le mieux est l'ennemi du bien" and you are "Le mieux"
User avatar
AF
AI Developer
Posts: 20686
Joined: 14 Sep 2004, 11:32

Post by AF »

Then how do you think task manager does that without modifying spring source at all? Or Process manager? Or the numerous other programs that display cpu usage. Fraps can count spring fps without modifying spring source code too.

Also the values generated would average out and the player would provide the context.

But then again your now shooting down a superior system, only your method is somewhat hypocritical.

Your reasoning is futile and full of holes. It does not address the issue and serves simply to distract and veer off into an unrelated direction. Manipulation at best.
Tobi
Spring Developer
Posts: 4598
Joined: 01 Jun 2005, 11:36

Post by Tobi »

Spring CPU usage is always 100% for other programs because it doesn't yield the CPU or sleep in the update loop (unless minimized, which could be exploited for exactly this purpose).

A spring based benchmark would work reasonably fine as long as it's a clearly defined startscript with e.g. 500 units doing all kinds of stuff for e.g. 5 minutes, and disallowing user interaction of any kind (run minimized?). The total percent of wallclock time spent in CGame::SimFrame() could then be a reasonably good measure of performance.
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

Tobi's suggestion is the good way to benchmark cause continuous averaging is, as I said before, not accurate. A "demo" script for benchmarking would be fine.
User avatar
AF
AI Developer
Posts: 20686
Joined: 14 Sep 2004, 11:32

Post by AF »

Indeed, in continuous averaging the test games with controlled stats would be weighted heavily so that they make up most of the marks.
User avatar
ginekolog
Posts: 837
Joined: 27 Feb 2006, 13:49

Post by ginekolog »

AF wrote:Indeed, in continuous averaging the test games with controlled stats would be weighted heavily so that they make up most of the marks.
+100
User avatar
LathanStanley
Posts: 1429
Joined: 20 Jun 2005, 05:16

Post by LathanStanley »

scripting a demo, a current GAME DEMO that was fek I dunno, encoded somehow in an archive, that was run, and logged the CPU useage of the player, the FPS of the player, and maybe even and upload bandwidth test, and then AVERAGED those numbers, would be an "OK" benchmark...

100% completely avoiding, and caring not what the CPU is, how fast it is, or how much ram it has.

If it works, it works, who gives a crap what the computer has inside.

hell, a top of the line computer with a slow harddrive can hurt games... blah...

I mean ell, it still states my CPU speeds and capability all wrong.

Spring is single threaded, and I make a more solid host than almost EVERY SINGLE OTHER PLAYER, just on the shear fact I have TWO! 4.4 GHz processing threads, with 4gb RAM.

that alone, is shear CPU power. Core2duo is badass, don't get me wrong, but until its Core2Extreme as standard, it can friggin lick my nuts.

but alas, the stupid program thinks I have a multi-core CPU running at 1.7 GHz a core, ignoring the fact its HYPERTHREADING, not MULTI-CORE, and the fact it ALSO ignores all of my overclock.

urgh, 1.7 (spring mark) vs. 4.4 (reality) ... thats not even 33% of my processor speed...

amusing. :roll:

still, springmark = bad coded fail.
User avatar
Snipawolf
Posts: 4357
Joined: 12 Dec 2005, 01:49

Post by Snipawolf »

Springmark does suck. It is fairly amusing that the only time I encountered major lag was recently.

NOTA with me vs. KAI..

>____>

I had .nocost on, and it killed me horribly *___*
User avatar
Peet
Malcontent
Posts: 4383
Joined: 27 Feb 2006, 22:04

Post by Peet »

LathanStanley wrote:Spring is single threaded, and I make a more solid host than almost EVERY SINGLE OTHER PLAYER, just on the shear fact I have TWO! 4.4 GHz processing threads, with 4gb RAM.

that alone, is shear CPU power. Core2duo is badass, don't get me wrong, but until its Core2Extreme as standard, it can friggin lick my nuts.
my c2d at 2.56 outperforms yours, on a per-core basis, btw.
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

LathanStanley wrote:...
hell, a top of the line computer with a slow harddrive can hurt games... blah...
Do you know how the hard drive is stressed while playing spring ? Can you give a law between avg access time + avg transfer rate and game speed issues ?

If so, just write it down with a side by side replay and hardrive I/O charts.

LathanStanley wrote:Spring is single threaded, and I make a more solid host than almost EVERY SINGLE OTHER PLAYER, just on the shear fact I have TWO! 4.4 GHz processing threads, with 4gb RAM.
I think you re going wrong. The point was not to know who has the fastest comp to be the best host but to have information about players before starting a game that will be a cpu killer.

I also really wonder whats your cpu that runs 4.4Ghz

LathanStanley wrote:Core2duo is badass, don't get me wrong, but until its Core2Extreme as standard, it can friggin lick my nuts.
Core processors have the best overall performances and overclocking abilities.
Masure
Posts: 581
Joined: 30 Jan 2007, 15:23

Post by Masure »

Snipawolf wrote:Springmark does suck. It is fairly amusing that the only time I encountered major lag was recently.

NOTA with me vs. KAI..

>____>

I had .nocost on, and it killed me horribly *___*
Where the link between springmark and your laggy game ?
Locked

Return to “Ingame Community”

cron