IMPORTANT: for those good folks testing Wind Smoothing

[Pete Dowson]

I had a minor brainwave this afternoon, and went looking for wind values in FSX through a different path. I found the three Ambient winds (X, Y, Z) in m/sec (the others were in ft/sec) and it looks like these are the source of the ones I've been changing, so I've now made some more changes to FSUIPC4 to deal with those first.

It looks good here, but, then, I never managed to get all of the weird wind problems reproduced here, so it needs testing (of course) by those who have seen the wind problems and can get them to occur.

So, please ignore version 4.238, which was yesterday's changes uploaded earlier today, and get 4.239 when it is available, soon.

Please note the changes remarked in the downloads Announcement -- please change the Logging parameters, otherwise this version will clobber your performance.

There are four things still outstanding, needing testing:

1. Does the PMDG A/P heading hold still drift when in tirbulence?

2. Are there still those nasty wind shifts at some levels?

3. Does this version cause FSX to crash at all, ever?

4. Is it better or worse, or no difference, with the Wind Smoothing thread enabled (as it is by default) or disabled (via "WindSmoothingThread=No")?

I think this is about as close as I'm going to be able to get, so I do hope it is okay.

Thanks & Best Regards

Pete

[Pete Dowson]

So, please ignore version 4.238, which was yesterday's changes uploaded earlier today, and get 4.239 when it is available, soon.

Something else has come up. At least two testers are getting this reported at the beginning of the FSUIPC4.LOG file:

Running inside FSX (SimConnect Acc/SP2 Oct07)

Module base=61000000

Wind smoothing fix is only partially installed

That "partially" means that most of the work on smoothing I've been doing is not actually being applied -- you are getting the same results as an SP1 or RTM user!

I don't understand why this should be. Looking at the logs received so far, Bill's and Al's have had this problem. Only Kyle's logs are from a fully-working smoothing installation. (This makes his FSX crashes more suspiciously linked to FSUIPC! :-( ).

I am now concerned that there might be more than one version of SIM1.DLL floating around for SP2 / Acceleration, not the one as I had always assumed. I have Acceleration installed on my test machine, but I will go install SP2 on a PC which has currently only got SP1 installed.

Can each tester tell me whether he is using SP2 or Acceleration, and check the Version number on his SIM1.DLL, please? (Right click on it, select Properties-Version). Mine is:

10.0.61637.0

and it is 669,408 bytes long (892,928 on disk), dated 26/09/2007 15:09 (though that may just be the install date).

Thanks,

Pete

[Pete Dowson]

Mine is:

10.0.61637.0

and it is 669,408 bytes long (892,928 on disk), dated 26/09/2007 15:09 (though that may just be the install date).

Aaaarrrggghhh!

The SP2 SIM1.DLL and the Acceleration SIM1.DLL are not the same beast!

Oh dear. I've got some work to do now to work out whether I can cope with the SP2 version. This is terrible. I'm sure Microsoft said that the actual FSX program was the same in both.

The SP2 version is:

10.0.61472.0

... which is weird, because Acceleration was out first yet has a much later build number!

It is only 797,760 bytes long (798,720 on disk).

For a moment I thought it might still be the SP1 version, but that is build 61355 and only 796,568 bytes.

But it doesn't bode well for the smoothing. I couldn't find the code I need to hook in the SP1 version, and it looks like this SP2 version is not much different.

I'm out tonight but I'll work on this over the weekend. Meanwhile, testers who are NOT using Acceleration, please give it a rest for now. Thanks!

Regards

Pete

[cshark172]

Pete,

I am using Acceleration and my Sim1.dll is version 10.0.61637.0, 868 KB on disk, dated 9/26/2007. As you said, my logs have been reporting full smoothing on. Maybe that is why my experience with the last version was a bit different. I will try a couple flights with 4.239 with and without smoothing to see if I get any crashes and how smoothing goes. I've got a touch of the flu, so it depends on how long I can remain upright in front of the computer :lol:

[jordanal]

testers who are NOT using Acceleration, please give it a rest for now. Thanks!

Understood and standing by...

FSX + SP2 (patch only) sim1.dll:

Modified: ‎Monday, ‎December ‎10, ‎2007, ‏‎9:30:16 PM

File Version: 10.0.61472.0 (fsx-sp2.20071210-2023)

Regards,

Al

[Pete Dowson]

I am using Acceleration and my Sim1.dll is version 10.0.61637.0, 868 KB on disk, dated 9/26/2007. As you said, my logs have been reporting full smoothing on. Maybe that is why my experience with the last version was a bit different. I will try a couple flights with 4.239 with and without smoothing to see if I get any crashes and how smoothing goes.

Yes, thanks. You are already good to go with 4.239. Maybe the only one other than myself at present! ;-)

I've got a touch of the flu, so it depends on how long I can remain upright in front of the computer :lol:

Oh, sorry to hear that. Don't over do it. Best to rest up and get better first.

Regards

Pete

[cshark172]

Hi Pete,

I've tested 4.239 with my KSEA-KDFW route. I can say that the wind smoothing is working real well. I flew with all smoothing options enabled, and I can report that there were absolutely no IAS issues on climb out and cruise. I did have the strange spurious winds showing up in WeatherSet2 (319 switched to 117). The smoothing at the aircraft worked! I was watching the wind on the progress page of the FMC, and it slowly ticked down to the new direction, and did not upset the aircraft's speed at all. A little later, the wind returned back to where it was supposed to be and again the wind at the A/C slowly caught up with no IAS jumps. Unfortunately, I got the crash again--still in the geographic area of the last ones. This time it was in weather.dll, instead of ntdll.dll. I have posted the error report below, don't know if it will help.

I then reflew with the WindSmoothingThread=No setting, and the results were the same. The winds were very smooth from take-off to cruise. Unfortunaley, I also crashed with this flight--ntdll.dll again.

I will try the flight I completed from Dallas to Philadelphia later with smoothing on to see if the crash might be a localized problem. One thing I wanted to ask is, how long did you run my saved flight from the crash I sent previously? I can load the save, and run anywhere from 20 minutes to an hour before it crashes again. I do have the logs and autosaves from the two crashes today if you would like to look at them--I know the previous ones weren't too helpful.

Error log from first crash in weather.dll (vista uses this new XML format in the event logs):

Log Name: Application

Source: Application Error

Date: 2/8/2008 2:42:24 PM

Event ID: 1000

Task Category: (100)

Level: Error

Keywords: Classic

User: N/A

Computer: poseidon-vista

Description:

Faulting application fsx.exe, version 10.0.61637.0, time stamp 0x46fadb14, faulting module weather.dll, version 10.0.61637.0, time stamp 0x46fadb59, exception code 0xc0000005, fault offset 0x0000a6be, process id 0xe68, application start time 0x01c86a8ac94bb5ad.

Event Xml:

1000

2

100

0x80000000000000

7414

Application

poseidon-vista

fsx.exe

10.0.61637.0

46fadb14

weather.dll

10.0.61637.0

46fadb59

c0000005

0000a6be

e68

01c86a8ac94bb5ad

[Pete Dowson]

I've tested 4.239 with my KSEA-KDFW route. I can say that the wind smoothing is working real well.I then reflew with the WindSmoothingThread=No setting, and the results were the same.

Yes, I think I might default the Smoothing Thread off. I don't think it'll be needed now.

I will try the flight I completed from Dallas to Philadelphia later with smoothing on to see if the crash might be a localized problem.

Yes, please. I am most concerned about this. If it is my smoothing hacks causing the crash then, ultimately, they aren't releasable. I must get that nailed down one way of the other. When I've sorted out the SP2 implemetation and there are more testers actually using the whole facility we should find out pretty quickly.

On that front I think I've found the right places in the SP2 edition of SIM1.DLL. Not only that, but that version is so similar to the SP1 and RTM editions than I think I can make it work on all FSX versions -- provided I can sort out the crash, if that is me.

One thing I wanted to ask is, how long did you run my saved flight from the crash I sent previously? I can load the save, and run anywhere from 20 minutes to an hour before it crashes again.

Are, maybe I'm not leaving it long enough then. I see if I can run much longer tests tomorrow. I'll need to set it up on another machine.

I do have the logs and autosaves from the two crashes today if you would like to look at them--I know the previous ones weren't too helpful.

Well, except that they seem to both point, at the log end (which is where I assume the crash occurred) to the same geographical area, almost precisely.

Yes, please do ZIP and send them. You never know, I might spot something.

Error log from first crash in weather.dll (vista uses this new XML format in the event logs):

Well, a crash in WEATHER.DLL is at least more easy for me to look at than buried deep in NTDLL like the others. Thanks. I'll have a quick look, but there's not much to go on. I do have to repro it here.

Thanks,

Pete

[cshark172]

Pete,

Files are on the way. Sure hope this crash can be cleared up. The smoothing was so good that I had to double check that winds aloft was on.

[Pete Dowson]

Files are on the way. Sure hope this crash can be cleared up. The smoothing was so good that I had to double check that winds aloft was on.

But the winds were changing, I hope. Not fixed, the same for the whole flight? That would be a tad too smooth! ;-)

Pete

[Pete Dowson]

Understood and standing by...

Okay. Solved it, I think. And the good thing is that solving it for SP2 enabled me to implement it all for SP1 and RTM too -- the SP2 SIM1.DLL is very similar to the SP1 and RTM versions.

So, please try version 4.241, now available.

Things to watch out for:

1. Check the start of the Log file, make sure it says the wind smoothing is FULLY installed.

2. The only tester so far doing lots of tests with full smoothing (because of the only-just-discovered differences with SP2) is getting FSX crashes in pretty much every flight, but we currently don't know if this is down to FSUIPC or not. It happens anything from 20 to 60 minutes into a flight, and so far I don't have a clue. I'm about 35 minutes into such a flight with his FLT + WX files, and it's okay so far ...

3. Still getting sudden wind shifts? Does the PMDG 747X heading still drift in turbulence?

4. I don't think the Wind Smoothing thread is doing anything useful now, so I might remove it, but please test with and without.

Thanks,

Best Regards

Pete

[cshark172]

But the winds were changing, I hope. Not fixed, the same for the whole flight? That would be a tad too smooth! ;-)

Pete

Yes they were changing--right on my 2 seconds per knott/degree. I should have used a smiley--I was speaking toungue-in-cheek there :wink: .

[Pete Dowson]

Yes they were changing--right on my 2 seconds per knott/degree. I should have used a smiley--I was speaking toungue-in-cheek there :wink: .

Ah, good!

Kyle, there's one thing about ALL of the Logs which end in an FSX crash, they ALL crash with the aircraft almost at the same place, every time.

Here is the last line from each log, preceded by the last Weather Report request before that:

  2723466 Weather Read request (At Aircrft) to area 4: Lat=44.62, Lon=-117.31, Alt=11750.5, Req=2
  2728146 WX Received in 0 mSecs, WX request type 1, ICAO=GLOB

  2303542 Weather Read request (At Aircrft) to area 4: Lat=45.16, Lon=-118.53, Alt=11770.3, Req=2
  2308207 WX Received in 0 mSecs, WX request type 1, ICAO=GLOB

  2665948 Weather Read request (At Aircrft) to area 4: Lat=44.67, Lon=-117.43, Alt=11756.0, Req=2
  2670613 WX Received in 0 mSecs, WX request type 1, ICAO=GLOB

  3541332 Weather Read request (At Aircrft) to area 4: Lat=43.55, Lon=-115.07, Alt=11715.4, Req=2
  3542736 Ambient wind: 279/32 (Act 279/32) OKAY

  2690986 Weather Read request (At Aircrft) to area 4: Lat=44.67, Lon=-117.43, Alt=11755.8, Req=2
  2695510 WX Received in 15 mSecs, WX request type 1, ICAO=GLOB

I included the very last line so you can see there's only a few seconds separating that position and the crash. (The altitudes are in metres by the way, in case you were wondering).

I am really wondering if there's something up with the scenery in that area that is responsible, so please do any further flights far away if possible just in case. Thanks.

I have had a flight continuing from your saved FLT now for 90 minutes. It's still cruising happily at 39000. This is on the PC I just installed SP2 on. No add-on scenery or similar files. I'll do some more such tests tomorrow.

There's one other noticeable thing in those logs, as you can see above. The last line in the log in every case is a Received GLOB weather -- except one: that was the time the crash was in WEATHER.DLL instead of NTDLL. Hmm. Suspcious. I am still pretty worried that it is in FSUIPC.

Oh, one other possibility -- that it is to do with the Logging itself. On the test I have running at present (which I might leace running all night) I've only got Weather logging and LogExtras=1, not 67. I think you could do the same now. I don't think I'll be needing the complex Smoothing logging any more. You don't even need the weather logging. Maybe if you have success in other areas you could try that on the crash-prone route?

Thanks & Regards

Pete

[jordanal]

Pete,

I will test .241 first thing in the morning, starting with the autosave files I sent you this morning. The last half of Kyle's KSEA-KDFW flight had a pretty good mix of winds and should give me about an hour's worth of flight time as well.

Regards,

Al

[Pete Dowson]

I will test .241 first thing in the morning, starting with the autosave files I sent you this morning. The last half of Kyle's KSEA-KDFW flight had a pretty good mix of winds and should give me about an hour's worth of flight time as well.

Thanks!

I'm off to bed now. I've left that flight running...

Regards

Pete

[cshark172]

Hello Pete,

I am happy to report that I completed a flight from KDFW to KPHL in the PMDG 747x with wind smoothing on/smoothing thread off, FSUIPC version 4.241. I made it gate-gate with no crash--2.5 hours. I did turn off logging for this flight. The wind smoothing was superb. No abrupt IAS swings at all. Climb out was smooth to FL370 for cruise. The winds aloft were generally from the west (244-283 degrees) with speed varying from 75-109 knotts. The wind smoothing kept them in check. They moved ever so slowly at the set 1 knott/degree every 2 seconds. I performed a VNAV descent into Philly and the plane was spot on. Winds at top of descent were 284/90 and calm for landing. The smoothing kept the plane on vertical path and speed held all the way down!

I will try a couple more flights away from the area where I was crashing before going back there and flying with logging off.

Before this flight, I went to KSEA with my archived weather to check PMDG 747 heading hold--still not holding in the low level turbulence.

Kyle

[Pete Dowson]

I am happy to report that I completed a flight from KDFW to KPHL in the PMDG 747x with wind smoothing on/smoothing thread off, FSUIPC version 4.241. I made it gate-gate with no crash--2.5 hours. I did turn off logging for this flight.

Okay, so either it's realted to the Logging, or it is something related to your scenery or terain or landclass in Idaho somewhere.

I left the flight running last night and it was still going this morning. Many hours, no problem. I will try the same with full logging enabled, just to double check. (The hard disk is big enough for the HUGE log that will generate! ;-)

Before this flight, I went to KSEA with my archived weather to check PMDG 747 heading hold--still not holding in the low level turbulence.

I'm in PMDG's hands on that one. I can't figure it out. Unfortunately the chap who is helping has his hands full at present.

I'll be changing the Wind Smoothing Thread option to "No" by default in the next version, and maybe remove it altogether for general release.

Thanks & Regards

Pete

[jordanal]

Ohhhhhh, so this is what wind-smoothing is supposed to look like! :lol:

Started Kyle's FP at BINKI @ FL390 and am currently 334nm from KDFW. No problems so far. Will try with windsmooth-thread=on after completion of this flight. Have already descended from FL390 to FL380 and much better in this version; no jumps in IAS.

********* FSUIPC4, Version 4.241 by Pete Dowson *********
Reading options from "E:\MS Flight Simulator X\Modules\FSUIPC4.ini"
User Name="*** Provided ***"
User Addr="*** Provided ***"
FSUIPC4 Key is provided
WideFS7 Key is provided
Running inside FSX (SimConnect Acc/SP2 Oct07)
Module base=61000000
Wind smoothing fix is fully installed
Wind smoothing thread is disabled

Regards,

AL

[Pete Dowson]

... much better in this version; no jumps in IAS.

Good. After having it run all night with no crash (but no logging), I started again with Kyle's flight but with full logging. However I'm now in a 737-700 with full Project Magenta instrumentation and it's autopilot. 2 hours into the flight and no problems so far. Looks good. It looks more and more as if Kyle's crashes are indded more likely to be related to scenery or similar.

Best Regards

Pete

[jordanal]

Hi Pete,

I'm sending my logs from my first round of testing with v4.241 without the wind-smoothing thread. Everything was fine except:

1) As soon as I hit the top of descent (TOD), the ND immediately indicated that I was 2500ft above my VNAV descent path. VNAV aligned with the proper descent path after ~ 10K feet of descent.

2) Approaching the first descent waypoint BIODS; I got large LNAV swings by the 744X to almost perpendicular to the FP (magenta line). Winds & speed were rapidly changing by 1 or 2 degrees and 1 or 2 knots it what appeared to be some kind of flutter, which I think is what confused LNAVs ability to follow the flight-path. These were very small changes but very rapid. This continued until capturing the rwy 18R ILS at KDFW upon which; a smooth CATIII ILS approach and landing was experienced.

3) This was using ASXsp2 historical weather downloaded; and the metar was date-time stamped Feb/3/2008-10:27pm.

4) No CTD was experienced after about 90 minutes of flight-time.

[Pete Dowson]

I'm sending my logs from my first round of testing with v4.241 without the wind-smoothing thread.

Thanks. Got them. Not so much needed unless there are smoothing problems, but I'll take a look.

As soon as I hit the top of descent (TOD), the ND immediately indicated that I was 2500ft above my VNAV descent path. VNAV aligned with the proper descent path after ~ 10K feet of descent.

Er, sorry, I'm not sure about this. Does it relate to the weather smoothing or anything FSUIPC is doing? I'm puzzled. It sounds like some sort of VNAV bug which should probably be reported to PMDG?

Approaching the first descent waypoint BIODS; I got large LNAV swings by the 744X to almost perpendicular to the FP (magenta line). Winds & speed were rapidly changing by 1 or 2 degrees and 1 or 2 knots it what appeared to be some kind of flutter, which I think is what confused LNAVs ability to follow the flight-path. These were very small changes but very rapid.

That is the turbulence simulation applied by FSUIPC, unless you disable it. That is why I am involving PMDG, to find out why minor wind speed and direction changes affect their A/P so much. It doesn't affect Project Magenta's A/P (though you can see it acting to keep to the path), nor does it affect FS's default A/P. I've also got earlier reports that it is okay with the LevelD 767 A/P. So, it is something in PMDG's algorithms I think. It looks like they have a sign wrong and instead of tending towards a value they veer away from it.

If you review the other threads here on the smoothing you will see that this PMDG heading problem is now the only major problem outstanding, and if it cannot be resolved (probably by PMDG) I will have to simply recommend PMDG users disable turbulence (both cloud and wind).

BTW no need now to test WITH the smoothing thread. If it is okay without I'd rather do without! ;-)

Thanks & Regards

Pete

[cshark172]

Hi Pete,

I can report another successful flight with version 4.241. I decided to try a trans-Atalantic flight today, KPHL-EGLL in the LDS 767. Completed with no crash--starting to look like I have something up in that area in Idaho where my problems were. Wind smoothing worked like a charm. Very smooth climb, cruise and descent. No IAS jumps more than about 2-5 knotts the entire 6.5 hours of flight. Couldn't really tell how temperature/pressure smoothing were working. I wasn't watching them too closely, like the winds.

One thing I did notice was there were no wind, temp or pressure changes over the N. Atlantic. I assume this is due to the lack of stations. I was using ASX live weather. Once I got away from the Canadian weather reporting stations the winds were at 271/50 and remained there until I was about 265 nm from Ireland, when station EIKY updated. ASX had pseudo-stations in between reporting different winds, but there was never an update in WeatherSet or at the A/C. Once the weather started changing again, from 265 nm off Ireland to London, the smoothing kicked in and kept everything in check.

[Pete Dowson]

One thing I did notice was there were no wind, temp or pressure changes over the N. Atlantic. I assume this is due to the lack of stations.

I think there's an option you can enable to get it to generate stations for you, but whether it gets "real weather" for those or makes it up itself I don't know.

ASX had pseudo-stations in between reporting different winds ...

Oh, you did have that enabled!

... but there was never an update in WeatherSet or at the A/C.

Sorry, I don't know what happened there then. To set such whether ASX has to use special calls to SimConnect to create those temporary stations. I've never used that, and I don't know how well it works. I should think the stations need creating well in advance as the weather takes time to be "merged in" and interpolated. Maybe you were going so fast the changes were too late.

Maybe HiFi Simulations support can advise us on this. (Jim, are you there?)

Regards

Pete

[JSkorna]

Hi All,

I am here quite a bit these days!

Here is what virtual stations do:

Check this to have ASX automatically generate stations within FSX in order to fill gaps and provide more weather coverage. Stations are generated at approximately 100nm intervals and inherit and/or interpolate the weather of nearby reporting stations. When using this feature, FSX should be closed before ASX in order to prevent a delay that occurs when FSX needs to remove Active Sky's generated stations on disconnect. Note that the use of virtual stations within FSX can cause a performance hit as well as potential wind shifts in some situations which can cause problems with certain add-on aircraft.*

*Note: Due to some inherent FSX interpolation issues, wind shifts can be experienced at any altitude with any option configuration but can be minimized depending on your altitudes and area of flight. Some experimentation of the Virtual Station and Route Coverage options are recommended to find the best balance of resolution, coverage and smoothness.

[cshark172]

Pete,

I may have found the source of my crash near Boise, ID. I was having windows stabilty issues last night, and discovered I had a bad memory module (tested all four DIMMS with MemTest, and 1 was erroring out like crazy). I have new memory ordered and on the way. I tried flying with 2 GB installed, but am having some issues with Windows freezing up. I probably have some corrupted data somewhere. I am restoring a backup from December that I had pre FSX, and will retest the area of the crash with a virgin FSX/Acceleration install, then start adding my addons 1 at a time to see if there is a problem somewhere. I have completed 3 flights now with wind smoothing on, away from the problem area and had no crashes.