SEIV , CPU Usage and SEV
 

One thing i cannot stand about seiv is its cpu usage.

To do a 3 meg turn takes an hour on a 2+ gig machine at 99%

This really needs to be addressed for SEV.

Has anyone who talks to MM actually mentioned this problem to him ?

Re: SEIV , CPU Usage and SEV
 

The committee game takes only 60 seconds on my 1-gig, and its in the 1 meg range.

Re: SEIV , CPU Usage and SEV
 

Wha-wha-what? SE4 never uses that much of the cpu on my system. It gets slow, when processing the turns of many A.I.'s in the late game, but I hadn't seen that kind of CPU usage. But I've only checked a few times, to be honest.

Re: SEIV , CPU Usage and SEV
 

Whatever the root issue is-- it takes a long time for the AI players to process their turns-- and it should really be addressed for SE-IV

Re: SEIV , CPU Usage and SEV
 

Funny this never happened to me on my old computer and it was a big piece. The most it ever took on my computer was 10mins on turn 800 and all the AIs and I had hundreds of ships.

Re: SEIV , CPU Usage and SEV
 

Tesco should have provided some context... He is talking about a 255 system map with a 10 planet start. It is/was a PBW game. Such games get really, really huge... Too big for PBW to handle, in fact.

Re: SEIV , CPU Usage and SEV
 

The Ai has to make its turn too you know. http://forum.shrapnelgames.com/images/smilies/happy.gif If it takes us an hour to make our turn then why not the AI? http://forum.shrapnelgames.com/images/smilies/happy.gif

Re: SEIV , CPU Usage and SEV
 

The AI could play its turns in the background while you are playing yours. This would greatly reduce the wait time after you hit "end turn" because a lot of the work would be done already. http://forum.shrapnelgames.com/images/smilies/happy.gif

Re: SEIV , CPU Usage and SEV
 

That would be really good if that could be done. The shorter the wait between turns, the better it would be. A part of the reason I have only finished a couple of games ever is because the A.I. turns take so long that I get tired of waiting and so I start a new game.

Re: SEIV , CPU Usage and SEV
 

My experience is that the AI turn isn't the problem but the huge battles with hundreds of ships. If there are no battles the processing is quite quick.

Re: SEIV , CPU Usage and SEV
 

Another thing that I find causes problems a lot of times isn't even big battles, but lots and lots of little ones. A big battle with hundreds of ships might only take a couple minutes to process. But when you have an unarmed ship blockading an enemy planet sometimes it will run dozens of combats where nothing happens but the blockade ship running for the corner and hiding. Multiply that by several planets, multiply again if the blockade fleets have 6 ships instead of one hiding in the corner, pretty soon you have a turn with dozens of combats where nothing at all happens. If the game could do a little pre-combat analysis and decide that nothing is going to happen, it could save some time by not having to process all those combats.

Re: SEIV , CPU Usage and SEV
 

We could do some tests to see how much of the host processing time is taken up by giving AI orders and how much is simply processing the orders for all empires. We could run some simultaneous turn games and then run the same turns but go through and control the AI players and give them orders.

EDIT: Ok, I did a little digging. I think the particular game Tesco is so frustrated about is the Return to Hell 2 game. That's one that he's been having to run offline every turn because it takes so long to process that PBW times out on it. In that game there are 9 players and no AI. Of course occasionally someone will miss their turn and the AI will have to play for them, but for the most part they all do their turns and it still takes a tremendous amount of time to process. So at least in this case the AI taking it's turns isn't a factor. It's all processing of player commands and running through the combats that's taking so much time.

Re: SEIV , CPU Usage and SEV
 

I don't think Tesco is so much complaining that it uses 99% of the CPU. What I've found watching turns run is it will pretty much use whatever processor percentage is available to it. If I am not running any other applications it will use 99%, even for a quick turn. If you aren't seeing it use 99% you probbaly just have other stuff running in the background. The complaint he's got is that even when he's not running any other apps and Se4.exe has 99% of processor resources available to it, it still takes a long time to process the turn.

Re: SEIV , CPU Usage and SEV
 

Geo et al,

I'm a bit concerned about this for entirely selfish reasons. I have the Zeta 2 game starting, which is 250 or so systems, and will likely be rather large. It depends, of course, on the number of people who join, but does PBW really find such games, at the latter stages, to be too big for it to handle? Or is it just a special case with the Return to Hell 2 game? In sum, I guess I am asking, are there specific tech spec limitations on PBW games, and if so, what are they?

Thanks!

Alarik

Re: SEIV , CPU Usage and SEV
 

I wouldn't say it's a common occurance, but it's definetly not unheard of. There isn't a list of specifications, but anything that makes a game more complicated and take longer to process in the later turns can be a problem.

The issue is that PBW will timeout after 20 minutes of processing a turn. This is what the PBW code was specifically written to do, it's not a bug or anything. The assumption was that turns that take longer than that had frozen up and stopped processing. Early on there were some bugs that would cause that to happen. So the 20 minute timeout was written in to keep a game from tying up the server and making it unable to process any other games.

Most of the games now that hit that timeout aren't actually locked up, just so complicated they take more than that to run. You can download the turn files and .plr files and process those turns offline though. And then upload the resulting turn back to PBW and it will notify the players of the new turn. PBW still helps in these cases as a turn distribution and collection, it just doesn't actually process the turn so there is a little more work involved for the game owner.

Geoschmo

Re: SEIV , CPU Usage and SEV
 

fyron. the settings do not matter. what matters is that the game is a process hog and will hang a system until the turn is complete. This is an issue with the low level coding of the game. It needs to be addressed for SEV.

Re: SEIV , CPU Usage and SEV
 

Does anyone have a link to Ruatha's PHP script he made for his backup PBW site? I think if I had access to that I could use it to run the RTH2 game and hopefully keep the game going on auto...

Re: SEIV , CPU Usage and SEV
 

Actually settings matter very much Tesco. Just saying it's a process hog is not really correct, since many games process just fine in a matter of seconds, and isn't very effective in getting the issue addressed for SEV. Since some games run fast and others run slow it's important to discuss what the differences are between games so Aaron can determine what parts of the game cause the processing of the turns to bog down so much. Like combats, number of players, number of systems, etc. All those things are important to figuring out where the problem is so it can be resolved.

Re: SEIV , CPU Usage and SEV
 

I have this script and will post it at my humble web site this evening.

Re: SEIV , CPU Usage and SEV
 

Well the RTH2 game is a special circumstance. Every system had (I think) 12 huge planets, the systems were in clusters of about 13 systems each disconnected from every other cluster. It was a full tech start, 9 players, 10 planet start. I think it's just over 100 turns in, and I'm at 21 million points....and I'm only in 7th place.... you can see how huge the game is.