Questions for the devs.
Moderator: Moderators
Questions for the devs.
Ok, if I recall correctly spring has 3 areas that cause major slowdown:
Pathfinding
unit simulation
features
correct?
It is my understanding that if a SM3 is rendering >5 texture layers the game slows down?
Does Sm3 support features yet? If so, does the SM3 feature usage also cause slowdown?
Pathfinding
unit simulation
features
correct?
It is my understanding that if a SM3 is rendering >5 texture layers the game slows down?
Does Sm3 support features yet? If so, does the SM3 feature usage also cause slowdown?
I don't like this, I don't like this at all, I don't like the idea that people have questions like these, I think that we urgently need a place to write the current spring major slowdowns/bugs.
up until now I thought there were only a few but now there is a whole list, and no one is working on these things.
up until now I thought there were only a few but now there is a whole list, and no one is working on these things.
unimportant ? I am asking unimportant questions ? well you could have used Java for spring but you didn't.....
because I thought that spring would prefer to be in this priority order :
1. stability (bugs)
2. performance
3. features
correct ?
btw assuming tobi will assist me to get my spring SVN to compile, you will have alote more questions and they are going to be ^2 at least in detail and amount
because I thought that spring would prefer to be in this priority order :
1. stability (bugs)
2. performance
3. features
correct ?
btw assuming tobi will assist me to get my spring SVN to compile, you will have alote more questions and they are going to be ^2 at least in detail and amount
So I better don't help you j/k
Anyway, to get the facts straight, number 1 bottleneck in spring is CQuadField (as shown by 2 or 3 profiles I ran in the past).
For the rest it doesn't make sense speculating what is slowest etc.: compile spring in profile mode, play with it a bit (might even try online, not sure that'll work / be playable tho), and look at the top 5 slowest functions in the profile results... good chance CQuadField::GetUnitsExact or some other CQuadField function is at the top.
Anyway, to get the facts straight, number 1 bottleneck in spring is CQuadField (as shown by 2 or 3 profiles I ran in the past).
For the rest it doesn't make sense speculating what is slowest etc.: compile spring in profile mode, play with it a bit (might even try online, not sure that'll work / be playable tho), and look at the top 5 slowest functions in the profile results... good chance CQuadField::GetUnitsExact or some other CQuadField function is at the top.
Monohouse, have a bit of respect for the devs please. They do what they can, this project is pretty big and it does have it's share of issues. I have been around for a while and am aware of these things being listed throughout my time here. I am not even going to start on the Java comment. However, I may ask questions, at least I am doing something around here, maybe I need the answers so I can get some stuff done? You know some of us contribute... Hey, it is ok, we cannot all jump in other people's threads and derail them like you feel the need to.
LordMatt, I am senior student of computer science. I took assembly. Asm would mean mean system specific code which means no cross platform(portability) anything.
Tobi, that still leaves my question about the features unanswered. I recall one of your fellow devs (I believe JC) saying the features do in fact run slowly.
LordMatt, I am senior student of computer science. I took assembly. Asm would mean mean system specific code which means no cross platform(portability) anything.
Tobi, that still leaves my question about the features unanswered. I recall one of your fellow devs (I believe JC) saying the features do in fact run slowly.
derail is in reference to a train. I means that you are taking the thread somewhere that it is not supposed to go... so what I meant was that the questions I was asking were adressed to the DEVS for a reason. I need answers to them for a reason. However, you were bringing the subject off topic. That is what I meant by derail.
Ok did some more tests and it wasn't really CQuadField that's slowest in this case.
For a test 20 gametime minutes nanoblobs random enemies 2 at speed 10 on Nano Arena, most gfx max except water the profile is here (warning, 4.3 meg text file, no point downloading it unless you have brainzorz and l33t coding skillz) http://www.osrts.info/~tvo/spring/profi ... mies_2.txt
Another test 5 kroggies vs 2 doomsdays, some complosions, and 100 flash vs 100 flash fight, XTA v8.1 on SmallDivide (again 4.6 meg, so don't download unless you're sure it's useful for you).
http://www.osrts.info/~tvo/spring/profile_xta_test.txt
Because of this I realized the only proper results are after running the profile on a number of replays of real games and averaging the results. Maybe I'll do something like that sometime.
For a test 20 gametime minutes nanoblobs random enemies 2 at speed 10 on Nano Arena, most gfx max except water the profile is here (warning, 4.3 meg text file, no point downloading it unless you have brainzorz and l33t coding skillz) http://www.osrts.info/~tvo/spring/profi ... mies_2.txt
Another test 5 kroggies vs 2 doomsdays, some complosions, and 100 flash vs 100 flash fight, XTA v8.1 on SmallDivide (again 4.6 meg, so don't download unless you're sure it's useful for you).
http://www.osrts.info/~tvo/spring/profile_xta_test.txt
Because of this I realized the only proper results are after running the profile on a number of replays of real games and averaging the results. Maybe I'll do something like that sometime.
- PauloMorfeo
- Posts: 2004
- Joined: 15 Dec 2004, 20:53
Why are you no longer suporting the idea that Spring should have been writen in Assembly?monohouse wrote:unimportant ? I am asking unimportant questions ? well you could have used Java for spring but you didn't.....
because I thought that spring would prefer to be in this priority order :
1. stability (bugs)
2. performance
3. features
correct ?
...
I didn't say assembly, and I didn't say .NET or Java, what I did say in other words is "why .NET is used for osrts ?" because I thought that osrts was trying to be a better spring as in better performance and you can't do that with .NET.
and I thought that stability and performance are the primary issues, and that was a question mark there too...
im not here to do damage, im trying to improve spring AND osrts, but it appears that some "smart" people do not understand that.
and I thought that stability and performance are the primary issues, and that was a question mark there too...
im not here to do damage, im trying to improve spring AND osrts, but it appears that some "smart" people do not understand that.
1st: derailing again, twice in a thread with just 18 posts, quite an accomplishment!monohouse wrote:... osrts ... I thought ... it appears that some "smart" people do not understand ...
2nd: stop, thinking, really. Ur thinking wrong.
Now back on topic! Any replies in PM plz..
tobi, what seems to be the slowest process?