Stutter with SP1 and FSUIPC 4.1

[Haldir]

Hi Pete. First off thank you for getting a working version of fsuipc up so quickly after the patch!

However I have a small problem. With the latest fsuipc installed on sp1 I get a regular decent sized stutter every 5 seconds. I tried removing fsupic.ini and letting it build a new one but it does it just the same. If I disable fsuipic in dll.xml the stutter is gone. I've included a picture of my performance graph, you can see the stutter clearly on core 0. Usage drops about 15% during each stutter.

FSX sp1 clean install, no addons except the realair sf260 and fsuipc. Happens with all aircraft, however the stutter is more severe in multiplayer, even with no one else in the session, than in singleplayer, not sure why that is. I have the registered version of fsuipc.

Core 2 Duo 6600

2 Gigs 667mhz

Nvidia 7900 GTO

All drivers up to date.

Any ideas? Sorry to be a pain. If I can send you any more information to help please let me know. Cheers Pete!

Mike

[/img]http://forums.simflight.com/files/fsuipcbugsmall_260.jpg

[Pete Dowson]

With the latest fsuipc installed on sp1 I get a regular decent sized stutter every 5 seconds.

The only 5 second timer in FSUIPC4 is a timeout for the connection to SimConnect. A stutter would most certainly be caused if it was getting no information and was doing a complete re-initialisation each time, as that involves thousands of little messages to SimConnect.

Please look at the FSUIPC4.LOG file. That is what it is for, to show things going wrong, if and when they do.

Regards

Pete

[Haldir]

Hi Pete. Ok I did some testing and saved the logs. The problem only seems to appear consistently in multiplayer, when I'm hosting a session. It was a locked session, myself in it only.

In singleplayer this is the log I get for a bell 206 jetranger flight at geneva. I've used fsuipc's aircraft specific joystick settings to reverse the throttle, no other changes......

********* FSUIPC4, Version 4.10 by Pete Dowson *********

User Name="Michael Johnson"

User Addr="ramasurinen@gmail.com"

FSUIPC4 Key is provided

WIDEFS7 not user registered, or expired

Running inside FSX (SimConnect SP1 May07)

Module base=61000000

DebugStatus=15

31 System time = 05:55:41

31 FLT UNC path = "C:\Documents and Settings\Lotus\My Documents\Flight Simulator X Files\"

31 FS UNC path = "I:\00FSX\"

1031 LogOptions=00000001

1031 SimConnect_Open succeeded: waiting to check version okay

2562 Running in "Microsoft Flight Simulator X", Version: 10.0.61355.0 (SimConnect: 10.0.61242.0)

25422 SimStart Event: Initialising SimConnect data requests

25422 FSUIPC Menu entry added

25453 C:\Documents and Settings\Lotus\Application Data\Microsoft\FSX\Previous flight.FLT

25453 I:\00FSX\SimObjects\Rotorcraft\Bell206B\Bell_206B_JetRanger.AIR

57234 System time = 05:56:38, FSX time = 14:55:45 (12:55Z)

57344 Aircraft="Bell 206B JetRanger Paint1"

62875 Advanced Weather Interface Enabled

115640 Weather Mode now = Theme

No problems there. This is what I get when hosting a multiplayer session, flying the exact same aircraft at the same location, same time of day, weather etc....

********* FSUIPC4, Version 4.10 by Pete Dowson *********

User Name="Michael Johnson"

User Addr="ramasurinen@gmail.com"

FSUIPC4 Key is provided

WIDEFS7 not user registered, or expired

Running inside FSX (SimConnect SP1 May07)

Module base=61000000

DebugStatus=15

31 System time = 05:55:41

31 FLT UNC path = "C:\Documents and Settings\Lotus\My Documents\Flight Simulator X Files\"

31 FS UNC path = "I:\00FSX\"

1031 LogOptions=00000001

1031 SimConnect_Open succeeded: waiting to check version okay

2562 Running in "Microsoft Flight Simulator X", Version: 10.0.61355.0 (SimConnect: 10.0.61242.0)

25422 SimStart Event: Initialising SimConnect data requests

25422 FSUIPC Menu entry added

25453 C:\Documents and Settings\Lotus\Application Data\Microsoft\FSX\Previous flight.FLT

25453 I:\00FSX\SimObjects\Rotorcraft\Bell206B\Bell_206B_JetRanger.AIR

57234 System time = 05:56:38, FSX time = 14:55:45 (12:55Z)

57344 Aircraft="Bell 206B JetRanger Paint1"

62875 Advanced Weather Interface Enabled

115640 Weather Mode now = Theme

150109 I:\00FSX\SimObjects\Airplanes\Aircreation_582SL\Aircreation_582SL.AIR

150109 I:\00FSX\FLIGHTS\OTHER\FLTSIM.FLT

175890 I:\00FSX\SimObjects\Rotorcraft\Bell206B\Bell_206B_JetRanger.AIR

216312 **** Simconnect Data Stalled! Re-connecting now****

216312 SimConnect_Open succeeded: waiting to check version okay

216312 Failed on SimConnect_CallDispatch for Message, return = 0xC0000120

216312 Running in "Microsoft Flight Simulator X", Version: 10.0.61355.0 (SimConnect: 10.0.61242.0)

216312 Initialising SimConnect data requests now

216312 FSUIPC Menu entry added

222203 User Aircraft ID not supplied -- trying default

225312 **** Simconnect Data Stalled! Re-connecting now****

225312 SimConnect_Open succeeded: waiting to check version okay

225312 Running in "Microsoft Flight Simulator X", Version: 10.0.61355.0 (SimConnect: 10.0.61242.0)

225312 Initialising SimConnect data requests now

225312 FSUIPC Menu entry added

231203 User Aircraft ID not supplied -- trying default

234312 **** Simconnect Data Stalled! Re-connecting now****

234312 SimConnect_Open succeeded: waiting to check version okay

234312 Running in "Microsoft Flight Simulator X", Version: 10.0.61355.0 (SimConnect: 10.0.61242.0)

234312 Initialising SimConnect data requests now

234312 FSUIPC Menu entry added

240187 User Aircraft ID not supplied -- trying default

That repeating section just goes on and on as long as the session runs, each time giving about a 0.5 second stutter.

I hope this helps? Thanks Pete!

Mike

[Pete Dowson]

Hi Pete. Ok I did some testing and saved the logs. The problem only seems to appear consistently in multiplayer, when I'm hosting a session. It was a locked session, myself in it only.

There was a problem in FSX as originally released where SimConnect became disconnected altogether when a Multiplayer session was started. That was what first prompted me to add in the extra code to detect a lack of data arriving from SimConnect, and execute a re-connection.

This problem was said to be fixed in SP1. However, from this it looks like the fix isn't working correctly -- in fact it is worse, as the reconnection before worked and was only needed just the once.

This part is a bit worrying:

216312 Failed on SimConnect_CallDispatch for Message, return = 0xC0000120

I'll need to find out what that error number means.

I will report this to Microsoft and see what they say. I know there have been Beta testers successfully using FSUIPC with Multiplayer and Beta versions of SP1, so it may be system-specific -- so can you also report it with as much information as you can about your system. Send stuff to tell_fs@microsoft.com. A Simconnect log may be useful (see the FSX Help announcement).

Then you might like to try changing the timeout and let me know:

SimConnectStallTime=n

This is actually the timeout between items of data arriving, defaulting to 1 second. Since data should arrive on every frame, this is actually a very generous allowance. The range is 1 - 9. You could try something silly like 9, and if that appears to "fix" it, gradually reduce it.

Currently the 5 second retry interval isn't adjustable, but bearing in mind nothing of FSUIPC is working in the intervening seconds, there's not much point in making it bigger or smaller. There's only such an interval in the first place to stop it taking over the system.

I'll check into that Error number, in the mean time.

[LATER]

I need to see what is happening from SimConnect's point of view, so could you please do that second test again for me, but first set up to get a SimConnect log -- the instructions in the FSX Help will tell you how. It will be quite large, so don't let it go on too long, then ZIP the logs (SimConnect and FSUIPC4) and send to petedowson@btconnect.com.

Thanks!

Regards

Pete

[Haldir]

Cheers Pete. I did the same test flight and have emailed you the simconnect and fsuipc logs.

I wish you luck, thanks again. :)

Mike

[Pete Dowson]

Cheers Pete. I did the same test flight and have emailed you the simconnect and fsuipc logs.

Okay, thanks. I'll take a good look. I'm out tonight so it may be tomorrow before I get through it.

Meanwhile, could you do a test with that timeout changed as I mentioned. Also, I see you have TrackIR installed too. Does that continue working okay throughout the MP session? Could you try a test without it running, just to see if it is related to their being two very intense Simconnect clients?

Thanks,

Pete

[Pete Dowson]

simconnect and fsuipc logs.

The SimConnect log shows the explanation for the FSUIPC4 log entries.

FSUIPC4 times out arrival of User Aircraft and General data, but it doesn't worry about AI/MP traffic data.

t seems that user aircraft data was arriving to FSUIPC4 (client 63) okay till the last such data at 70.81075 (70+ seconds) which is possibly where you engaged MP mode? Following that the ONLY "ObjectData" return to client 63 was for AI/MP aircraft. Then, at 83.61249 I reconnect and re-initialise everything, to no avail ...

I look forward to seeing results without TrackIR.

Regards

Pete

[Haldir]

Hi Pete. Sorry I forgot to mention previously that I tried your timeout value suggestions, but I can't see any change in the behavior. Logs look the same. It seems that the stutters may have been spaced out a bit by setting the timeout to 9, but it was so far past my bedtime I'd have to say that was subjective at best haha.

I just tried disabling the track-ir, no change at all, same regular stutter. No change in the fsuipc log.

The strange thing is that my fsuipc joystick assignments, like my throttle reverse and assigned buttons all work just fine, so simconnect and fsuipc must be talking no?

Mike

[Pete Dowson]

The strange thing is that my fsuipc joystick assignments, like my throttle reverse and assigned buttons all work just fine, so simconnect and fsuipc must be talking no?

The event system is working, just not the aircraft and other data -- none of the standard SimVars feeding the FSUIPC offsets. This being FSUIPC's main purpose in life, it gets upset if it doesn't receive such data, hence the attempts to fix it by reconnecting 9which worked fine, and once only, before SP1).

I've provided the info to Microsoft and am in discussion with them about what might have changed to cause this in SP1. I'll be in touch when I've any news or something else to try.

Regards

Pete

[Haldir]

Ok, sounds good Pete. Thank you again for looking into this. Your actions define a standard of customer service that few can match. I'm *always* impressed.

I'll keep mucking about on my end. I don't really understand simconnect very well, but maybe I'll run across something that can help.

Cheers.

Mike

[Pete Dowson]

Ok, sounds good Pete. Thank you again for looking into this.

I have made a few small changes in FSUIPC4, as an experiment -- they were suggested by some things asked of me by Microsoft. Whether they will help or make things worse I dont know. But please donwload 4.102 using this link:

http://fsuipc.simflight.com/beta/FSUIPC4102.zip

This is the DLL only. Just copy it into the Modules folder (you may need to "unblock" the ZIP in its file properties first). If you could get me both an FSUIPC4.Log and a SimConnect log using this, as before, Id be grateful, no matter what the outcome.

Thanks!

Pete

[Haldir]

Ok Pete, tried the new version. Logs are in your email. No big recurring stutter, but lots of small stutters and accompanying errors in the logs. Track-ir is disabled, I'll leave it off for testing from now on so you're not spammed with data.

Cheers.

Mike

[Pete Dowson]

Ok Pete, tried the new version. Logs are in your email. No big recurring stutter, but lots of small stutters and accompanying errors in the logs.

I don't think the small stutters are related to fSUIPC, but 4.102 evidently isn't working in any case. The problem revolves around ambiguities in how to identify the User aircraft in Multiplayer mode.

I'm still talking to MS about this, but the weekend is going to get in the way. I'll let you know when there's any progress. Meanwhile I'm withdrawing 4.102.

Regards

Pete

[Pete Dowson]

Okay. One last try before I really do have to await more detailed help from MS.

Please download 4.103:

http://fsuipc.simflight.com/beta/FSUIPC4103.zip

and test that in the same way. FSUIPC4 and SimConnect logs to me, please.

If it is more or less okay, stop the SimConnect logging (just rename the SimConnect.INI file) -- that may be contributing to any stutters by its disk accessing etc.

Regards

Pete

[Haldir]

Hi Pete. Version 4.103 looks good so far! Logs are in your email. No errors in the fsuipc log, AC specific joystick settings are working again. Stutter is gone. I'll keep testing tonight in real multiplayer sessions and report back tomorrow.

Cheers!

Mike

[Haldir]

I've put 4.103 through its paces with about 8 hours of multiplayer flight and no errors at all so far. Tried every aircraft I have, hosting as well as flying in others' sessions and no big stutters at all. Looks like you fixed it!

I have a hunch simconnect is still going to sleep in multiplayer though despite ACES' claim to have fixed that problem. I re-enabled 3wire again just to mess with some carrier traps. It uses simconnect also and works like a charm in singleplayer, but it isn't reading aircraft position data in multi at all. I installed the latest sdk update but it didn't help. Strange.

Cheers Pete.

Mike

[Pete Dowson]

I have a hunch simconnect is still going to sleep in multiplayer though despite ACES' claim to have fixed that problem.

If it was going completely to sleep it could cause FSUIPC to timeout and re-connect, so it can't be, really. At least not completely.

I re-enabled 3wire again just to mess with some carrier traps. It uses simconnect also and works like a charm in singleplayer, but it isn't reading aircraft position data in multi at all.

Is this some other code, not using FSUIPC? If so, possibly it is using the non-MP ID of the User Aircraft to read the data. I think the ID is 1 normally, but it could be anything, apparently, when MP mode is started. This is what caused the problem in FSUIPC 4.102 that I got you to try earlier.

I always actually asked SimConnect for the ID, and used the one which was returned. Before SP1, because enterring MP mode cut off SimConnect data and FSUIPC4 re-copnnected, it asked for the ID again, so, once reconnected, it worked okay. With the fix in SP1 it wasn't reconnecting initially, so used the wrong ID.

There's a note about this in the SDK which isn't terribly prominent:

Note: Multiplayer Mode

When developing a client for use in multiplayer mode it is not safe to use the ID number for the user aircraft returned by the function SimConnect_RequestDataOnSimObjectType, as the actual number can change depending on several factors, including the number of users involved in the multiplayer flight. Always use the constant SIMCONNECT_OBJECT_ID_USER for the ObjectID parameter if the SimConnect client is to work in multiplayer mode.

Please do check that FSUIPC4 is getting the aircraft location in MP first, just to be sure. Run TrafficLook, for example -- the User aircraft position is in the first line listed.

Regards

Pete

[Haldir]

Hi Pete. I checked out trafficlook, everything seems to look fine. Planes seem to show up in the correct order in the sdk traffic map tool as well, but helicopters don't show, just says "non-aircraft" for my helo, and no players who join the session with helos show up. Still everything seems to work fine there regardless.

No worries on 3wire, I erroneously assumed it would start working in multiplayer after the simconnect fix in the patch, there's probably more to it. Can you tell I never fly singleplayer yet? haha.

3wire doesn't use FSUIPC at all to my knowledge so I'll ask the author about it, see if he's had the same experience, I know it's still very much a work in progress.

I do have one other observation though. I can't seem to get Helo trim to work properly in fsuipc anymore. It definitely worked in your pre-patch versions of fsuipc. It 'sort' of works at the moment which I'll briefly explain.

I can apply pitch trim commands either through the normal fs assignments or through fsuipc, and the helo will trim out, but the second I move the stick the pitch trim is neutralized again. However, if I then center the stick and hit the trim just once again it picks up where it left off, resulting in a big trim change at once. Seems it's remembering the amount of trim I had dialed in, but it's not applying it to the main controls if the joystick is in use. This is both single and multiplayer. Hope that description makes sense. :)

Cheers Pete.

Mike

[Pete Dowson]

Hi Pete. I checked out trafficlook, everything seems to look fine. Planes seem to show up in the correct order in the sdk traffic map tool as well, but helicopters don't show, just says "non-aircraft" for my helo

What exactly says "non-aircraft"?

My code does check for and accept

SIMCONNECT_SIMOBJECT_TYPE_AIRCRAFT and

SIMCONNECT_SIMOBJECT_TYPE_HELICOPTER

so if they aren't getting through to me with those checks they must somehow be provided as something else. The only other types listed are:

SIMCONNECT_SIMOBJECT_TYPE_BOAT

SIMCONNECT_SIMOBJECT_TYPE_GROUND

and I eliminate those for obvious reasons.

No worries on 3wire, I erroneously assumed it would start working in multiplayer after the simconnect fix in the patch, there's probably more to it.

It'll be the user object ID.

I can apply pitch trim commands either through the normal fs assignments or through fsuipc, and the helo will trim out, but the second I move the stick the pitch trim is neutralized again. However, if I then center the stick and hit the trim just once again it picks up where it left off, resulting in a big trim change at once. Seems it's remembering the amount of trim I had dialed in, but it's not applying it to the main controls if the joystick is in use. This is both single and multiplayer. Hope that description makes sense. :)

I've no idea what is going on there -- I've made no changes at all. Maybe MS have included a new bug -- or they've added some sort of "autotrim" mechanism?

I'll have a look when I can, but it'll be in about two weeks' time -- I've got a load on here then I'm off for a week. I have made a note to check into it. Meanwhile if you can collect any more specific data it may be useful.

Regards

Pete

[Haldir]

Hey Pete, sorry my mistake, I didn't word my email properly, the perils of being up too late haha.

Trafficlook does show my aircraft in slot 00, whether hosting or flying in someone else's session. It shows my aircraft tail # as my user ID and all the data next to that looks correct.

It was the FSX SDK's tool, traffic toolbox>map that showed my helo as a "non-aircraft", not trafficlook, sorry about that.

Thanks for looking into the helo pitch trim as well, its such a wonderful feature of FSUIPC. I was the one who asked you about a way to do that about a year ago, and you implemented it about two days later haha. Cheers for that. :)

Mike

PS: Enjoy your vacation!

[Pete Dowson]

I can't seem to get Helo trim to work properly in fsuipc anymore. It definitely worked in your pre-patch versions of fsuipc. It 'sort' of works at the moment which I'll briefly explain.

This afternoon I did done some tests on the helo trim facility, and I have found a bug, but it doesn't seem to fit your observations. Not only that but this is an original bug -- it was in the FSUIPC4 from the moment the Helo trim option was included.

I can only think that some time differences with SP1 has made it become more of a problem than it was before.

The bug is that when the trimmed value for the axis was computed (whilst the elevator axis value was "en route" via my Calibrations to FS as an "axis event"), the code was also writing the value back to the FSUIPC axis offsets but with the sign reversed!

Now, I can't see why this could do any harm to anything except programs designed to manipulate axes via the FSUIPC offsets. It should definitely not interfere with the values actually used by FSX otherwise.

I cannot reproduce the symptoms you describe at all, no matter what I do. But maybe the bug I have fixed will also fix your problemmaybe.

The fix will be in interim version 4.104 which I'll provide via the FSX Downloads announcement tonight. There's another fix in it too -- for the PFC menu when FSCopilot is being used.

If you still get the problem, please do some logging of it, using 4.104, as follows:

Add:

Debug=Please

LogExtras=1024

to the [General] section of FSUIPC4.INI.

Load FSX, go to FSUIPC4's Logging page, and set monitoring, to the normal log, of offsets 0BB2 and 0BBE, both as type S16.

Thanks,

Pete

[Haldir]

Hi Pete. I logged the two offsets you requested with 4.104, just sent them to your email. No change in helo trim behaviour. I'll try some different methods of mapping my joystick through FSUIPC and see if anything changes.

Cheers.

Mike

[Pete Dowson]

Hi Pete. I logged the two offsets you requested with 4.104, just sent them to your email. No change in helo trim behaviour.

The log seems to show something rather different to what you describe, but perhaps it's a matter of interpretation. It shows the Helo Trim (0BBE) and the Elevator Axis (0BB2) being EXACTLY THE SAME all the time, as if they are the same axis!

How are you setting the trim? I tested using NUM keys 1 and 7 for all trim changes. I'm afraid i didn't ask you to log all axes -- but something is really very very strange as shown in the log.

The log seems to show no time whatsoever with "neutralised trim". Instead the trim = elevator setting all the way, as if the elevator axis was always zero (centred) and you are using trim exclusively.

More information needed. I cannot reproduce anything like that here (and the FSUIPC code hasn't changed, of course). Maybe I need to set up my joystick input exactly like you. I assume you are using the default Bell helo? If not, what?

Pete

[bfels]

whats the issue with PFC when FSCopilot is running ? i saw the user, but didnt understood what was the issue...

Benjamin Fels

[Haldir]

OMG I am such a moron. The helo trim does work Pete, my *sincerest* apologies for wasting your time on that. When I removed my fsuipc.ini file with the release of 4.1 I guess it was, when I was trying to figure out what was causing the multiplayer stutter, I completely forgot to remap my elevator and aileron axes through fsuipc again. It was using the FSX defaults, so the trim wasn't sticking. It works fine now.

Hangs head in shame. :(

Thank you once again for your patience and help Pete.

Mike