Official v645 Thread

Collapse
This topic is closed.
X
X
 
  • Time
  • Show
Clear All
new posts
  • longiron
    1000 Post Veteran and Moderator

    • Jan 2007
    • 1453
    • Just south of NYC

    #76
    Originally posted by GoGoran
    I will turn it down to 0 now and also deactivate dynamic shadows, and update this post then.
    A profile of your system would also be helpful with the results of your next test.
    Thanks
    Chris
    "True rail fans have two favorite railroads. The B&O and one other."

    Comment

    • GoGoran
      Senior Member
      • Jun 2010
      • 880

      #77
      Originally posted by longiron
      A profile of your system would also be helpful with the results of your next test.
      Thanks
      OK, I just deactivated the sound (set to 0), but apart from the missing sound things did not change much - fps stayed almost the same - 11fps instead of 9 before.

      Then I reactivated the sound (set to 5), but without BinPatch Sound support: Sound process went down to 60%, with 30% waiting time (previously 100%, and 100% waiting time). Framerates remained unchanged at 11 fps on the medium-complicated route with the complex trainset I used for all of my tests.

      A word about my test train: 2 diesel locos with 20,000 polygons each, 20 waggons with 400 polys each plus 20 FreightAnims with 16,000 polygons each - so all in all a train of 368,000 polygons. I used this train for all of my tests in both v0.360 and v0.645.

      Finally I left the sound as above, but deactivated the dynamic shadows: Now the sound process was still at 60% process time and 30% waiting time, but framerates jumped from 11 to 35fps average, with a minimum of 30 fps and a maximum of 40 fps ! Overall process load was down to 160%, with 45% waiting time on average.

      My conclusions:

      1. Dynamic shadows require huge processing power, a double core CPU or better on v0.645. Deactivating them almost quadrupled my framrate !

      2. something has to be wrong with the sound implementation on v0.645 ! It worked just fine on v0.360 !


      PS: System profile:

      AMD Athlon 64 3500+ @ 2.2GHz
      2.5 GB DDR2 RAM
      Ati HD4350 graphics card with 512MB DDR2 RAM, PCIe 16x
      2TB HDD with 64MB cache, SATA2
      SoundBlaster AWE32 PCI sound card

      Windows Server 2003 R2 SP2 - 32-bit edition
      Hardware acceleration activated for both graphics card and sound card, both working correctly (verified)
      all other required software installed as described in the ORTS manual

      Comment:

      MSTS and v0.360 of ORTS work just wonderful on this machine !

      And with v0.645 the issues are only the dynamic shadows and the sound.
      Last edited by GoGoran; 03-09-2011, 17:55.
      sigpic

      Comment

      • BLR-IIC
        New Member
        • Dec 2009
        • 15

        #78
        The real reason you are having lousy frame rates is probably because dynamic shadows are enabled. Disable the dynamic shadows and everything should run OK.

        Comment

        • GoGoran
          Senior Member
          • Jun 2010
          • 880

          #79
          PS: I wonder why the DynamicShadows are so prohibitively CPU-hungry in ORTS. I hope it's just an early development stage and it will change soon.

          On MSTS, my framerates only took a mere 10 - 15% hit when I activated DynamicShadows, and I have been using it there ever since.

          Wasn't ORTS supposed to be less CPU-hungry, and able to outsource more work to the GPU ? Doesn't seem to have happened yet, though.
          sigpic

          Comment

          • 2slow4u
            New Member
            • Nov 2010
            • 14

            #80
            Originally posted by GoGoran
            2. something has to be wrong with the sound implementation on v0.645 ! It worked just fine on v0.360 !
            Agree...
            On 0.645, in my case, a lot of in cab sound effects are missed, 0.360 was fine.

            Comment

            • GoGoran
              Senior Member
              • Jun 2010
              • 880

              #81
              The car spawns on my test route (Biggetal version 2006 by GR (German Railroads, payware)) work just fine with v0.645.
              sigpic

              Comment

              • hminky
                1000 Post Veteran
                • May 2009
                • 1114

                #82
                I was running the London & Port Stanley to see how a densely populated route would do in OR. I got a 13 FPS with dynamic shadows and a 53 FPS without in the same area.

                The L&PS looked great in OR.

                Harold

                Comment

                • twpol
                  Senior Member
                  • Oct 2009
                  • 434
                  • United Kingdom

                  #83
                  Originally posted by GoGoran
                  PS: I wonder why the DynamicShadows are so prohibitively CPU-hungry in ORTS. I hope it's just an early development stage and it will change soon.

                  On MSTS, my framerates only took a mere 10 - 15% hit when I activated DynamicShadows, and I have been using it there ever since.

                  Wasn't ORTS supposed to be less CPU-hungry, and able to outsource more work to the GPU ? Doesn't seem to have happened yet, though.
                  Making sweeping remarks like this with no understanding of the issues is highly frustrating to the OR developers and not very productive either. Dynamic shadows, like the rest of the rendering, outsource huge quantities of work to the GPU but there is still work that must be completed on the CPU. To draw the dynamic shadows, every object that casts a shadow must be drawn at least one extra time and that obviously costs CPU and GPU time to do.

                  In general, this new release will draw slightly more objects in to the shadow map than v360 (it varies by time of day how much extra work it does) and does additional work to create the shadows without some of the artifacts seen in previous releases, which is why it expected to be more costly than in v360.
                  James Ross
                  Open Rails Management Team
                  sigpic

                  Comment

                  • twpol
                    Senior Member
                    • Oct 2009
                    • 434
                    • United Kingdom

                    #84
                    Originally posted by mikephfl
                    Any suggestions on what to look for/change? Also on the route I'm building I have no car spawners working on the entire route. Maybe I need to modify something?
                    Make sure there are no messages related to car spawners in the log and that "CARSPAWN" is printed during loading (it happens early, e.g. "Loading TRK TRP(default) TDB SIGCFG DAT ACT RDB CARSPAWN"). I'll have a look at Surfliner 1.2 when I have the time.
                    James Ross
                    Open Rails Management Team
                    sigpic

                    Comment

                    • GoGoran
                      Senior Member
                      • Jun 2010
                      • 880

                      #85
                      Originally posted by twpol
                      Making sweeping remarks like this with no understanding of the issues is highly frustrating to the OR developers and not very productive either. Dynamic shadows, like the rest of the rendering, outsource huge quantities of work to the GPU but there is still work that must be completed on the CPU. To draw the dynamic shadows, every object that casts a shadow must be drawn at least one extra time and that obviously costs CPU and GPU time to do.
                      I expect enough professionalism from the ORTS developpers not to misunderstand anything or take it personal. I really appreciate ORTS and everyone who's contributing to it. And when I post here, reporting any issues I have, I do this to aid you in the development, not to bash ORTS or its developpers. I have a vested interest in seeing ORTS succeed, since it is the only way to get my high-poly models running.

                      That being said, the fact remains that MSTS manages to draw the dynamic shadows with a mere 10 - 20% drop in fps, whereas the drop in ORTS v0.645 is currently at ca. 75% ! And this figure has already been confirmed by other users, too, it's not just this way on my system alone.

                      And I'm also not the only one experiencing issues with the sound implementation in v0.645 ...
                      Last edited by GoGoran; 03-09-2011, 19:03.
                      sigpic

                      Comment

                      • twpol
                        Senior Member
                        • Oct 2009
                        • 434
                        • United Kingdom

                        #86
                        Originally posted by GoGoran
                        I really appreciate ORTS and everyone who's contributing to it
                        Could you please show that appreciation by not making arbitrary guesses at things then?

                        I'm going to share some hidden settings in a new thread regarding the dynamic shadows and I'd like everyone with large FPS issues to read it and report back.
                        James Ross
                        Open Rails Management Team
                        sigpic

                        Comment

                        • muskokaandtahoe
                          3000 Post Veteran

                          • Apr 2003
                          • 3003
                          • Silicon Valley, CA, USA.
                          • ORTS

                          #87
                          Minor problem in the warning message... the captured/displayed text appears to be the Static_Flags(), not the UiD(). As a result there is no telling which entry is failing, making it much more difficult to figure out and fix.

                          runactivity.exe : Warning : 0 : Error in D:\Program Files (x86)\Microsoft Games\Train Simulator\ROUTES\WP3\WORLD\w-012915+014330.w

                          Missing texture filename in forest region. UID=4294967295
                          at ORTS.ORTraceListener.TraceEventInternal(TraceEvent Cache eventCache, String source, TraceEventType eventType, Int32 id, String format, Object[] args)
                          at ORTS.ORTraceListener.TraceEvent(TraceEventCache eventCache, String source, TraceEventType eventType, Int32 id, String message)
                          at System.Diagnostics.TraceInternal.TraceEvent(TraceE ventType eventType, Int32 id, String format, Object[] args)
                          at System.Diagnostics.Trace.TraceWarning(String message)
                          at MSTS.Tr_Worldfile..ctor(SBR block)
                          at MSTS.WFile..ctor(String filename)
                          at ORTS.WorldFile..ctor(Viewer3D viewer, Int32 tileX, Int32 tileZ)
                          at ORTS.SceneryDrawer.LoadAt(Int32 tileX, Int32 tileZ)
                          at ORTS.SceneryDrawer.Load(RenderProcess renderProcess)
                          at ORTS.Viewer3D.Load(RenderProcess renderProcess)
                          at ORTS.RenderProcess.Initialize()
                          at Microsoft.Xna.Framework.Game.Run()
                          at ORTS.Viewer3D.Run()
                          at ORTS.Program.Start(IEnumerable`1 options, String[] args)
                          at ORTS.Program.Main(String[] args)


                          EDIT: Sorry, typed the wrong thing.. .the values in the warning message are from the VDbId... as in VDbId ( 4294967295 ). Perhaps the message is using the last value parsed instead of the one that caused the error?
                          Last edited by muskokaandtahoe; 03-10-2011, 15:00. Reason: Added EDIT message as postscript.
                          Dave Nelson
                          sigpic
                          Seldom visiting, posting less often that that.

                          Comment

                          • lindsayts
                            Senior Member
                            • May 2010
                            • 194

                            #88
                            I haved used V645 now for around 3 to 4 hours. I must say this release took some getting used to with its new facilities and being a development version new bugs. But I find it is growing on me. A number or people have talked about the view keys, I found these take a LOT of getting used to but I find them new layout quite usefull. I have so far tried 4 or 5 differnet routes and have OR installed on three different systems. Interstingly it behaves slightly different on each of the systems. I will put a more detailed report when I have spent more time on the issue.

                            I have an immpression the frame rate is lower than with V360 but as yet have not done a comparison. The processor load figures though are reasonable being around 98% render 47% update and 1% sound.

                            I have come up with a number of problems for which I will post bug reports but at first a minor problem.

                            One of my systems has a Samsung 305T 2560x1600 monitor. Now both V360 and V645 work OK on it but with the current version when you save the options it displays an error box saying the format of the screen size string is not correct. It being of course "2560x1600". Now it appears the save routine is scanning the options for errors and because 2560x1600 is not in the list its is having a minor nervous breakdown. Now OR is great on 2560x1600 and one would much appreciate being able to use it in the future in this resolution.

                            On the whole VERY good work on the part of the OR team it is MUCH appreciated I may say.

                            Lindsay

                            PS I was going to respond to the remark on why shadows took so much effort to do but some one beat me to it.
                            Last edited by lindsayts; 03-09-2011, 19:08.

                            Comment

                            • twpol
                              Senior Member
                              • Oct 2009
                              • 434
                              • United Kingdom

                              #89
                              Originally posted by lindsayts
                              One of my systems has a Samsung 305T 2560x1600 monitor. Now both V360 and V645 work OK on it but with the current version when you save the options it displays an error box saying the format of the screen size string is not correct. It being of course "2560x1600". Now it appears the save routine is scanning the options for errors and because 2560x1600 is not in the list its is having a minor nervous breakdown. Now OR is great on 2560x1600 and one would much appreciate being able to use it in the future in this resolution.
                              Turns out it is an odder bug that that - the code that checks the value requires the width and height to be <2000. I guess it was meant as a sanity check for silly values but unfortunately catches you out.

                              You should still be able to set it from the old menu, if you ever need to change any other settings.
                              James Ross
                              Open Rails Management Team
                              sigpic

                              Comment

                              • ericmp
                                Member
                                • Aug 2006
                                • 26
                                • .

                                #90
                                I'd like to provide my feedback here as well...

                                Overall, I'm amazed with what has been accomplished. The new shadows, the cab controls appearing in the cab, the new particle effects, and all of the other new features are welcome additions. I'm especially happy to see that the headlight positioning issue has been fixed. Performance wise, I haven't noticed any significant loss.

                                The only disappointment I had was the loss of the microtex on the terrain. The ground textures look low-res without it, but I read that the team hopes to implement microtex in a future version. With the number of improvements in V0.645 over V0.360, I can live without the microtex.

                                The new camera controls are taking some getting used-to, but that isn't really an issue, as I'll get used to them.

                                That's it, really. To the ORTS team, thank you for all of your efforts and keep up the excellent work. I'll be posting some screenshots of the new version soon.

                                Comment

                                Working...