Author Topic: Some signalling oddities as always  (Read 2016 times)

0 Members and 1 Guest are viewing this topic.

Offline Junna

Some signalling oddities as always
« on: September 23, 2016, 08:12:59 AM »
There are 3 recurring problems on the network.

1. Trains at Whippoorwill South Railway station will at times not observe choose signals, and stop waiting as though it was a regular signal, even though platforms are clear for use. Opening schedule window will renew reservation which will fix it, but it keeps propping up at times. Occasionally trains will somehow end up on the departing fast-line out of the southern station, though they are scheduled for the approach.

2. At Tarton Junction, goods train from Saltingford sheep farm, which stops at the platforms between reversals in order to avoid it stopping too long to wait for clearance, will be given  a spurious green, resulting in a dead-lock, as it will not have platform clearance.

3. Similar to the above, trains at Parport will occasionally jam, and this should be reproducible eventually. I am not sure if these events happen while running or after a load, but they should repeat, for they have done so for in-game years. The Parport issue also involves a goods train being given spurious clearance though platforms are occupied. This issue also, but more rarely, will appear at Crosswych railway station; these 3 stations are all major through stations but with reversing local stopping services, which seems to have some to do with the problems.

The game also runs very sluggishly. It is not really possible to play with the speed unpaused, as mouse- and key lag is quite great, and selections often fail (when clicking a train or a tile, it will miss things, and be off, etc). I disabled city cars in an effort to improve performance but there was none.

http://www.mediafire.com/download/axlh8k9dmei13lh/Pak128.Britain-Ex-0.9.2J.rar

Pak

http://www.mediafire.com/download/clb4wl3a9x3g3b2/ders5611.sve

save.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #1 on: September 23, 2016, 08:54:40 PM »
Thank you for your reports. I will look into the signalling anomalies in due course, but wanted to focus on the performance for the time being. On my computer, running an optimised release build, I get decent performance on the map that you sent, albeit there are slight signs of it working hard (framerate drops for the most part). The UI is responsive.

May I ask what your system specifications are? May I also ask you to check the display settings menu, and give me the output from frame time, idle, FPS and simloops?

For context, my system is a Core i7 950 (the first generation i7), overclocked to just over 4Ghz, and 12Gb of RAM with reasonably decent memory timings.

My frame time fluctuates between 70-80ms, idle is mostly 0ms, FPS is usually 12 (at 4k resolution, but this makes little difference as the graphics portion is multi-threaded these days, as confirmed by my quad-core CPU showing 29% utilisation for the Simutrans executable, meaning that there is consistent load on more than one core) and simloops fluctuates a little, but is usually around 5.0.

I have wondered whether some of the signalling changes have put extra load on the CPU/memory bandwidth, and I may need to look into optimising this somewhat.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #2 on: September 23, 2016, 09:28:14 PM »
I have an i7 3790k, not overclocked, and 12 gb ram.

The FPS usually is 20-25, 10 when paused. Frame Time hovers around 40-50ms (paused and unpaused), idle is 100ms paused, otherwise hovering between 0 and 15, simloops goes between 5 and 6. The fast forward time factor hovers between T=2.20 to 3.60 (but does not really look to be much different). Sometimes it runs sluggishly for a while, but can be made to run smoother by fast forwarding and then getting it back to normal. The controller lag is really the most annoying issue, because it impedes construction. Clicking and dragging will take a couple of seconds to register, for example, and trying to place track will be difficult and delayed by a number of tiles when attempting to drag.

Edit: in the save game, there should be convoy 1385, approaching at around 4:40 Whippoorwill South, will not enter the station, but turn around and leave, as if it is heading towards the way point which should only be used by trains after stopping.
« Last Edit: September 23, 2016, 09:45:05 PM by Junna »

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #3 on: September 23, 2016, 10:52:22 PM »
Interesting. Your performance is actually a little better than mine. There is a little input lag; perhaps I am less sensitive to that than you. I will see of I can add more INT_CHECK calls to help with that. I shall have to have a go at profiling to see whether I can improve performance.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #4 on: September 25, 2016, 05:35:19 PM »
It seems the problem at Whippoorwill South and Longchester have to do with waypoints. Some times the trains will have the waypoint appear in schedule, and will not register passing it until it has completed a loop all the way around on the track segment in question.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #5 on: September 25, 2016, 11:50:30 PM »
Thank you for that additional information: I will look into that when I get a chance.

In the meantime, I have upgraded to Visual Studio Community Edition 2015, having previously been using Visual Studio Express 2012. The reason for this upgrade is because Visual Studio 2015 Community Edition has built-in profiling tools which Visual Studio 2012 Express lacked. Using these tools, I have carried out a little optimisation on the devel-new-2 branch. I have not added any INT_CHECK calls yet, but tried to make it run a little faster. Can you test with the latest code and let me know whether it is any more responsive now? Thank you.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #6 on: September 27, 2016, 10:02:42 PM »
I have now had a chance to do some more optimisation. Would you mind testing the latest build to see whether it is more responsive for you?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #7 on: October 04, 2016, 09:37:02 PM »
Looking at the reported anomalies, I am having some trouble reproducing issue no. 1. I am currently in November 1956 observing Whippoorwill South station, and have not found one instance of this behaviour. Can you point to a particular train and/or point in time in which this will be reproduced after loading the saved game?

Incidentally, this is a splendidly impressive saved game; it must have taken you a long time to work on this. May I ask, however, whether there is any particular reason that you set the power and force multipliers to 300%?

Edit: I eventually found a misbehaving train: convoy no. 1928. I think that I have fixed issue no. 1; are you able to re-test? I should be grateful.
« Last Edit: October 04, 2016, 09:58:30 PM by jamespetts »
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #8 on: October 05, 2016, 12:13:42 PM »
May I ask, however, whether there is any particular reason that you set the power and force multipliers to 300%?

No real reason other than avoiding having too many double- and triple-headed trains which take up too much platform space.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #9 on: October 05, 2016, 12:34:17 PM »
Interesting. Given that the performance at 100% is calibrated to roughly realistic levels, why did you need the double headers in the first place?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #10 on: October 05, 2016, 01:19:33 PM »
They have trouble reaching their top speeds with my rather lengthy trains, especially since the powers were recalculated down? I compulsively add carriages to everything, so to be able to reach their maximum speed they'd all need to have three, maybe four locomotives each, so I just upped the power factor.

As it were, the performance is somewhat improved in the newer build (T= improved by between 0.50 and 1).

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #11 on: October 05, 2016, 01:26:08 PM »
Interesting; the maximum speeds for steam engines are not a target: they are the highest speed that the engines can reach with a light load. A realistic speed with a heavier load will be less. More modern forms of transport will have official speed limits, but steam engines did not have these.

Is the performance improved enough for it to be responsive?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #12 on: October 05, 2016, 08:53:08 PM »
If they're slower, less people will find them usable and less traffic, won't they? The braking distances on the steam engines is such that they seem to slow down to 50km/h between any signal blocks on absolute block signalled sections anyway, even when block sections ahead are clear.

I can't really test the other fix as I can't compile it since the MSVS-2015 update, as I just get libbz2.lib errors now (unresolved external symbol ____iob_func).

Offline Ves

Re: Some signalling oddities as always
« Reply #13 on: October 05, 2016, 09:12:13 PM »
Quote
I can't really test the other fix as I can't compile it since the MSVS-2015 update, as I just get libbz2.lib errors now (unresolved external symbol ____iob_func).
A quick offtopic sidenote: Try DrSupergoods suggestion in this post: http://forum.simutrans.com/index.php?topic=14705.msg154718#msg154718. This helped me.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #14 on: October 05, 2016, 09:26:31 PM »
Obviously, the faster that convoys are, the more people that use them; but this would equally be a reason to make everything run at 1,000km/h, would it not? What matters, after all, is absolute speed, not speed relative to the vehicles' speed limit. Lest there were any doubt, I am not criticising you: you are of course free to make your own version of the pakset and play as you please, but I am always very interested in people's playing choices and the rationale behind them because it helps me to understand better how to make the game work well.

As to the compilation problems, have you been able to compile libbz2 using MSVC 2015? That is what you need to do to make it compile again. The steps for doing so are relatively straightforward.

Edit: As to problem no. 2, I am having some trouble finding the trains in question. Can you give me the line name or convoy number(s)?

Edit 2: Also, I have had it running for a fair while, and it has not jammed at Parport. How long does it need before this reliably occurs from the saved game that you supplied?
« Last Edit: October 06, 2016, 12:00:40 AM by jamespetts »
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #15 on: October 06, 2016, 02:03:27 AM »
I managed to get it to compile once I knew how to compile the bz2 as a .lib rather than dll file. The general controller lag appears to be significantly less so far, much more playable and generally being more responsive, with none of the skipped tiles as was seen before.

The Parport issue is quite rare. I am not sure if it occurs during regular running, if it has been fixed, or if it is something that happen only when a train is on a certain tile at a certain point upon loading the game and paths being recalculated; it has also happened at Tarton, where many services reverse, etc; I think it's probably related or a variant of one of the things Vladki has reported from the test map. It usually does not happen frequently, but happen once or twice during the course of an in-game year (or used to, anyway, but I have not, despite many efforts, been able to have it occur at a specific time (though I usually make a save every in-game month in case something happens.)

Edit:

http://www.mediafire.com/file/7wbl1601i274amq/ders571s.sve

It should jam at Tarton. The jams almost always involve the G Milk Saltchester - Greenwich, which reverses at a siding outside of Tarton before stopping at the platforms to get it through without waiting too long. It then ends up entering the platform before the J 61 Colfield - Tarton leaves, and thus ends up causing a deadlock, entering the occupied track. This should repeat eventually.
« Last Edit: October 06, 2016, 02:31:30 AM by Junna »

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #16 on: October 06, 2016, 08:38:00 PM »
Thank you for that. May I ask at what in-game time that it is reproduced after loading this saved game so that I can tell for sure whether the problem can in fact be reproduced?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Offline Junna

Re: Some signalling oddities as always
« Reply #17 on: October 06, 2016, 09:41:47 PM »
Should be before the middle of February at latest, but the issue might have occurred previously (and been related to the chose signal oddities which were just fixed), so if you cannot reproduce it, that might be why.

Offline jamespetts

  • Simitrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 15368
  • Total likes: 368
  • Helpful: 165
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Some signalling oddities as always
« Reply #18 on: February 18, 2017, 01:05:57 AM »
I am currently reviewing old bug reports and I notice that this one appears to have no reported resolution. May I ask whether this can still be reproduced?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.