More on game delays and pacing
| Author |
Message |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
I haven't posted this to classictw.com yet, so the delays you're seeing there are outdated. I am testing these things on 21-6.com and I'll get it up on classictw.com soon.
I used the fig drop macro on classictw.com and it timed at about 27 ms. I did the same macro on 21-6.com and it timed at 7 ms.
Oh, I also verified that the only delay being called is that for the actual fig drop.
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Wed Nov 24, 2010 11:33 pm |
|
 |
|
Singularity
Veteran Op
Joined: Thu Jun 02, 2005 2:00 am Posts: 5558 Location: USA
|
 Re: More on game delays and pacing
K, whenever u get it up I'll re-test then.
_________________ May the unholy fires of corbomite ignite deep within the depths of your soul...
1. TWGS server @ twgs.navhaz.com 2. The NavHaz Junction - Tradewars 2002 Scripts, Resources and Downloads 3. Open IRC chat @ irc.freenode.net:6667 #twchan 4. Parrothead wrote: Jesus wouldn't Subspace Crawl.
*** SG memorial donations via paypal to: dpocky68@booinc.com
|
| Wed Nov 24, 2010 11:40 pm |
|
 |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
Up now. Sorry if I wasted any of your time.
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Thu Nov 25, 2010 12:23 am |
|
 |
|
Singularity
Veteran Op
Joined: Thu Jun 02, 2005 2:00 am Posts: 5558 Location: USA
|
 Re: More on game delays and pacing
John Pritchett wrote: Up now. Sorry if I wasted any of your time. No you're fine dude, I appreciate the chance to test this stuff to such detail. It's rare. Here's my new results: Macro: m20102^M m9183^M m9724^M m14699^M m18317^M Time: 1315 ms (263ms per sec) Macro: m20102^M f1^Mcd m9183^M f1^Mcd m9724^M f1^Mcd m14699^M f1^Mcd m18317^M f1^Mcd Time: 1452 ms (290ms per sec) Macro: m20102^M za99999^M fz1^Mzcdzr^M m9183^M za99999^M fz1^Mzcdzr^M m9724^M za99999^M fz1^Mzcdzr^M m14699^M za99999^M fz1^Mzcdzr^M m18317^M za99999^M fz1^Mzcdzr^M Time: 1457 ms (291ms per sec) So the 2nd and 3rd macro are basically identical in timing. But, there's a 27ms difference between the 1st two. That doesn't look entirely right... Macro: m20102^M fz1^Mzcdzr^M m9183^M fz1^Mzcdzr^M m9724^M fz1^Mzcdzr^M m14699^M fz1^Mzcdzr^M m18317^M fz1^Mzcdzr^M Time: 1440ms (288ms per sec) So the za is adding about 2ms per sec. The figlay then is adding about 25ms per sec. That should be only 5ms? Much better, but not quite there...
_________________ May the unholy fires of corbomite ignite deep within the depths of your soul...
1. TWGS server @ twgs.navhaz.com 2. The NavHaz Junction - Tradewars 2002 Scripts, Resources and Downloads 3. Open IRC chat @ irc.freenode.net:6667 #twchan 4. Parrothead wrote: Jesus wouldn't Subspace Crawl.
*** SG memorial donations via paypal to: dpocky68@booinc.com
|
| Thu Nov 25, 2010 12:29 am |
|
 |
|
Big D
Veteran Op
Joined: Tue Nov 28, 2006 4:04 pm Posts: 5025
|
 Re: More on game delays and pacing
John Pritchett wrote: The first step for scaling the delays will be a single scale multiplier that will be applied equally to all delays in the game. I'd like to find a solid set of baseline delays that people are comfortable with, and then allow gameops to scale those delays uniformly in order to decrease CPU load. I'd also like to provide the ability to customize these delays individually, but there are currently 162 of them (and I might have missed a few) and it would take a lot of work to add TEDIT support for that many new fields. There may be another way to handle that, outside of TEDIT, like an INI file or something. I think there's value in having something like that because as the game continues to evolve, I think gameops could better address balance issues if they could modify the pacing of individual actions like this. I just don't think I'll have that in as a feature immediately. A game-wide scale setting I could do right away, though.
I'm going to explore the handling of ship lists for xport because I don't think it necessarily should increase much in CPU for multiple ships. I think you are on to something here. Maybe have the delay varibles written to a config file that could be edited by sysops. Only the more experienced sysops would figure out how to edit them, but that probably would be a good thing. That might be complicated on your end depending on when and how these variables would be initially set into each individual game.
|
| Thu Nov 25, 2010 12:42 am |
|
 |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
Something that we have to factor in here is that whenever there is a break in the input stream, you could have an injection of additional delay. There are two possible sources of this delay. One is network latency. I get a ping of 20 ms to classictw.com. The other is an IO processing lag of 10 ms in TW itself. I noticed that this was showing up in my profiling when I was using a 100 ms lag for processing the input/output buffers, so I decreased this to 10 ms and improved the situation significantly. But it's very possible that we're seeing a combination of these factors.
I'm going to duplicate your tests here and see if I can see any additional delays being imposed, and if not, where that extra 25 ms per sector is going.
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Thu Nov 25, 2010 12:44 am |
|
 |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
BigD, I think as knowledgeable gameops tweaked these settings, we would see a consensus emerge about the best timing for the game (at least these fast-paced games), and then I could lock those settings in as default. So the game could evolve as the gameops evolve it. But also, in a situation where I'm away from the game and unable to keep up with needed changes, gameops won't be tied down and forced to wait for me.
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Thu Nov 25, 2010 12:47 am |
|
 |
|
Big D
Veteran Op
Joined: Tue Nov 28, 2006 4:04 pm Posts: 5025
|
 Re: More on game delays and pacing
John Pritchett wrote: Something that we have to factor in here is that whenever there is a break in the input stream, you could have an injection of additional delay. There are two possible sources of this delay. One is network latency. I get a ping of 20 ms to classictw.com. The other is an IO processing lag of 10 ms in TW itself. I noticed that this was showing up in my profiling when I was using a 100 ms lag for processing the input/output buffers, so I decreased this to 10 ms and improved the situation significantly. But it's very possible that we're seeing a combination of these factors.
I'm going to duplicate your tests here and see if I can see any additional delays being imposed, and if not, where that extra 25 ms per sector is going. I just wonder how these tests compare to the non beta version as far as differences between in timeing.
|
| Thu Nov 25, 2010 12:50 am |
|
 |
|
Big D
Veteran Op
Joined: Tue Nov 28, 2006 4:04 pm Posts: 5025
|
 Re: More on game delays and pacing
I can't believe you guys are working this hard on these problems 1 hour into "turkey day" laff. Everyone appreciates the hard work. You must be a night owl like some of the rest of us.
|
| Thu Nov 25, 2010 12:55 am |
|
 |
|
Singularity
Veteran Op
Joined: Thu Jun 02, 2005 2:00 am Posts: 5558 Location: USA
|
 Re: More on game delays and pacing
If it were a case of network latency wouldn't it affect both macros about equally?
I can see perhaps packets getting split or held up occasionally, but wouldn't that show up as variance over runs?
I just ran: m20102^M za99999^M fz1^Mzcdzr^M m9183^M za99999^M fz1^Mzcdzr^M m9724^M za99999^M fz1^Mzcdzr^M m14699^M za99999^M fz1^Mzcdzr^M m18317^M za99999^M fz1^Mzcdzr^M
10 times... here's what I get 1. 1452 2. 1472 3. 1452 4. 1452 5. 1463 6. 1452 7. 1492 8. 1482 9. 1459 10. 1472
So I can see a 40ms variance over the entire macro (avg 8ms per second, but it probably doesn't break down per second). So at most I should be seeing maybe 13ms per sector diff, rather than 27ms.
My in-game twx proxy ping is: S: ping : 170 ms S: ping : 180 ms S: ping : 151 ms S: ping : 230 ms S: ping : 160 ms S: ping : 161 ms S: ping : 170 ms S: ping : 170 ms S: ping : 161 ms S: ping : 150 ms Min: 150, Max: 230, Avg: 170.
So even with a max ping of 230, I'm not seeing that much difference here.
And D, I don't consider this work. I enjoy stuff like this.
_________________ May the unholy fires of corbomite ignite deep within the depths of your soul...
1. TWGS server @ twgs.navhaz.com 2. The NavHaz Junction - Tradewars 2002 Scripts, Resources and Downloads 3. Open IRC chat @ irc.freenode.net:6667 #twchan 4. Parrothead wrote: Jesus wouldn't Subspace Crawl.
*** SG memorial donations via paypal to: dpocky68@booinc.com
|
| Thu Nov 25, 2010 12:56 am |
|
 |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
I'll dissect the macro you're testing and hopefully see where the discrepancy is coming from. BigD, my wife is doing the cooking, so I don't have anything else to be doing  She'll be at it until 3 am easy. But yeah, I normally work until 2 to 4 am. I don't get to start work until 3 to 4 in the afternoon, so late night is my best time.
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Thu Nov 25, 2010 1:28 am |
|
 |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
Hmmm, I wonder if this is introducing some extra delay. Game S was set to 20 commands per second. I set that to 0 so it's not a factor. Fingers crossed that it helps.
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Thu Nov 25, 2010 1:33 am |
|
 |
|
Singularity
Veteran Op
Joined: Thu Jun 02, 2005 2:00 am Posts: 5558 Location: USA
|
 Re: More on game delays and pacing
Elapsed Time: 1451 ms Macro: m20102^M za99999^M fz1^Mzcdzr^M m9183^M za99999^M fz1^Mzcdzr^M m9724^M za99999^M fz1^Mzcdzr^M m14699^M za99999^M fz1^Mzcdzr^M m18317^M za99999^M fz1^Mzcdzr^M
Game time: 01:58:18 AM Fri Nov 25, 2022
_________________ May the unholy fires of corbomite ignite deep within the depths of your soul...
1. TWGS server @ twgs.navhaz.com 2. The NavHaz Junction - Tradewars 2002 Scripts, Resources and Downloads 3. Open IRC chat @ irc.freenode.net:6667 #twchan 4. Parrothead wrote: Jesus wouldn't Subspace Crawl.
*** SG memorial donations via paypal to: dpocky68@booinc.com
|
| Thu Nov 25, 2010 1:58 am |
|
 |
|
John Pritchett
Site Admin
Joined: Sun Dec 24, 2000 3:00 am Posts: 3151 Location: USA
|
 Re: More on game delays and pacing
Well, I'm seeing an extra 20 ms per sector on 21-6.com without the CPC, so it's definitely something else. My actual results were somewhat lower than yours, but basically the same ms per sector difference.
Ok, I got it. The 20 ms lead-in time for sector movement is being absorbed by the 250 ms move time when you're doing nothing but moves. But when you add in the fig drop between moves, the 20 ms lead-in time for sector movement is actually being charged, adding 20 ms per sector. This is an error in how I'm handling the pre-move synch time. Either the 20 ms should always be included, or it shouldn't be included at all. That way it'll be consistent whether you're doing moves only or doing other things between moves.
You believe that the 20 ms synch delay per move is necessary, correct? So that the actual move time is 20 ms synch delay plus 250 ms move delay?
_________________ John Pritchett EIS --- Help fund the TradeWars websites! If you open a hosting account with A2 Hosting, the service EIS uses for all of its sites, EIS will earn credits toward its hosting bill.
|
| Thu Nov 25, 2010 2:33 am |
|
 |
|
Singularity
Veteran Op
Joined: Thu Jun 02, 2005 2:00 am Posts: 5558 Location: USA
|
 Re: More on game delays and pacing
I wouldn't go less than 10ms on movements, given the new delays on pwarp and photon. But yes, I do think it's important. Later on those numbers might need adjusted for balance tho. I suppose if you want to reduce the photon delay to 5ms instead of 10ms, we could get away with something like a 5ms sync delay and it wouldn't be noticeable.
Also, question, do you have an exit and enter delay too? Whether 2ms, 5ms or 10ms, it would help with exit-enter floods.
_________________ May the unholy fires of corbomite ignite deep within the depths of your soul...
1. TWGS server @ twgs.navhaz.com 2. The NavHaz Junction - Tradewars 2002 Scripts, Resources and Downloads 3. Open IRC chat @ irc.freenode.net:6667 #twchan 4. Parrothead wrote: Jesus wouldn't Subspace Crawl.
*** SG memorial donations via paypal to: dpocky68@booinc.com
|
| Thu Nov 25, 2010 2:52 am |
|
 |
|
Who is online |
Users browsing this forum: No registered users and 15 guests |
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot post attachments in this forum
|
|