Pete, look this please

[ulisses]

http://forums.avsim.net/dcboard.php?az=4606&page=

Not sure what is going on, but I think this thread can help you with some feedback. I get some stutters but not all the time.

Ulisses

[Pete Dowson]

Not sure what is going on, but I think this thread can help you with some feedback. I get some stutters but not all the time.

I've not really got time to chase other threads. If I started doing that I'd get nowhere! I'm sure if anyone has anything important to ask or tell me they'd do it to here, otherwise it cannot be all that important.

Very small occasional stutters are almost inevitable if there are any weather changes through the only interface I have for them. This is particularly noticeable if the 3D clouds are not set to 100%, as I note in my documentation, but they are substantially reduced with the latest weather updating facilities I provided and which are starting to be used in the weather programs.

If default weather is used (no outside weather program), and the FSUIPC option to change FS weather is not checked, then FSUIPC is really doing very little except responding to external programs on request.

Regards,

Pete

[Soarpics]

Hi Pete,

My testing parameters as follows:

FSMeteo 6.x and Weather Center Novice

FS Jepp weather and FS User Defined weather

Unregistered (FSMeteo) FSUIPC 3.04 and 3.07

Tested on a good FS9 install, one that has given me few problems (most user induced) in almost two months. I have a few years experience working with MSFS, and easily tweaked it and my system to make it fly the way I wish. It's thus far been a well running sim on my system.

I've thoroughly tested under numerous conditions as noted in the Avsim thread... even removing FSMeteo and WCN completely from my system. In ALL tests, under ALL conditions, the sim has the stutter if FSUIPC is in the Modules folder. Remove the .dll, and the stutters are gone completely.

Hope this helps,

Greg

[Pete Dowson]

In ALL tests, under ALL conditions, the sim has the stutter if FSUIPC is in the Modules folder. Remove the .dll, and the stutters are gone completely. Hope this helps

Not a lot, actually, at least not yet. I don't know anyone with consistent stutters when nothing is using FSUIPC -- yours is the first report of anything like this I've actually received. I run FS9 on three very different PC systems and the only stutters I see are from things like scenery and autogen coming into view (provided I have all the cloud stuff set full).

So I need to use you to test things, if that's all right? If you haven't registered FSUIPC then this might not get very far, so say so, and after a couple of easy tests (below), I may have to wait for a volunteer who has a user registered copy.

Basically, when nothing is accessing FSUIPC, which is the only test I'm interested in at present, then FSUIPC does nothing except:

1. If you have any weather options set AND you have the Technical option set to influence FS weather, then it will update the weather now and then. So please set all options off. If your FSUIPC is unregistered (what!?) then use the button to set minimum.

2. If does obtain numerous items of information from the main simulation engine, SIM1.DLL, but you cannot stop this. If it looks like an area which could be causing your stutters then I'll devise a way of bypassing this code, to try to narrow it down.

3. It regularly reads the "weather at aircraft" so it can report all the weather details in the appropriate offsets. Unlike in FS2002 and before, this is done by a call into FS's Weather DLL, which isn't as efficient as I'd like. But I've found no other way.

4.It scans the population of AI aircraft regularly, so it can provide TCAS data.

And that's about it -- any other functions are completely dependent upon options being set to use things in FSUIPC (I assume you aren't actually using any), or being called or used by other programs, DLLs or gauges.

So, for any tests, please don't run any FSUIPC applications, make sure there are no other add-in DLLs in Modules, and only use default aircraft and panels.

TEST #1

You can make FSUIPC bypass ALL the weather actions completely (it doesn't go near WEATHER DLL then) by adding the following two lines to the FSUIPC.INI file:

Debug=Please

LogExtras=16

so first, please try that and see if the same stutters persist.

You can actually then turn this on and off with FS running. Go to the FSUIPC Logging page and change the "Debugging Data" field to 0 for normal weather scanning, and back to 16 for no weather stuff at all -- that way you can compare things to your heart's content.

TEST #2

Turn the AI aircraft off completely (Air Traffic Density = 0). Do the stutters persist?

You can change the amount of time FSUIPC spends on AI aircraft via the parameter "TrafficScanPerFrame" -- set it real low, 1 say for 1% per frame, to see if that has an effect.

Okay, let me know please.

Also, please can you describe the stutters more precisely, there timing, frequency, persistence? I'm really at a loss to understand what you are seeing.

Regards,

Pete

[Soarpics]

Hi Pete,

I've completed the tests you requested of me. My test flight was the same I've used for a few years for testing anything I do is FS... 3500" above S.F bay, clear weather, no wind, heading due north. In this case I used the default C-172. Unregistered FSUIPC 3.07 is the dll I tested with.

First, an explanation of the "stutters". I timed them today, and the closest I could get was 1 1/2- 2 seconds apart (probably closer to 1 1/2). The pause itself only lasts for a fraction of a second. The net result is a subtle jerky motion, as if the sim stops for an instant and then catches up exactly where it should be. But there is that instant of frozen motion that presents itself as the stutter.

Before testing I made sure there were no other FSUIPC options running, and removed the only other non-MSFS dll from the Modules folder (FS Nav 4.6.0.0).

Test #1... added the "Debug" and LogExtras" lines to the .ini file. I started by making sure that "Debugging Data" was set at 0. Stutters present. I then set the field to 6. Stutters present. Went to 12, and the stutters were less noticeable. At 16 I could no longer see the stutters. Went in the reverse order, and the stutters reappeared in much the same fashion as they disappeared. Back at 0 and 6 they were present as always.

Test#2... I usually fly with AI set to 30-50% (depending upon my start and destination airports). I can't remember the last time I set AI to anything above 50% (too much of a resource hit). "Debugging Data" was set to 0 at the beginning of testing.

Before testing I checked my "TrafficScanPerFrame" setting in the .ini file... it was 10. I first flew with AI set to 0%, and observed the stutters as usual. I then tested at 50%, and saw the same stutters. Went to 100% AI, and the stutters were present. No change no matter what the AI setting was. Then tested the above parameters again with "TrafficScanPerFrame" set to 1. I observed the same result as with it set to 10... stutters present and consistant at the above 3 AI settings.

Sys specs:

AMD 2400XP @ 2700

ASUS A7N8X Rev. 2.0 w/ ASUS 1.16 drivers and 1005 BIOS

1 Gig Corsair XMS 3200

ATI 9700 Pro w/ Omega 2.4.78

Windows XP w/ SP1. Minimum start-up apps... Desktop, Virus Scanner and Video Card tweaker. Disabled VS and tweaker for testing purposes. All MS live updates disabled. This has proven to be a very stable system.

After running these tests I removed the FSUIPC .dll from my Modules folder and reflew the test. No stutters present, even after enabling FS's Jepp download weather.

Keep in mind that the stutters are rather subtle. One must be looking for them to see them. It was difficult to see the gradual changes in Test #1. But what I saw at settings of 0 and 16 was very easy to see... stutters and no stutters respectively.

Feel free to let me know if I can do any further testing,

Greg

[Pete Dowson]

Test #1... added the "Debug" and LogExtras" lines to the .ini file. I started by making sure that "Debugging Data" was set at 0. Stutters present. I then set the field to 6. Stutters present. Went to 12, and the stutters were less noticeable. At 16 I could no longer see the stutters. Went in the reverse order, and the stutters reappeared in much the same fashion as they disappeared. Back at 0 and 6 they were present as always.

Values 0, 6, and 12 won't do anything, except maybe add extra logging and possibly slow things down -- performance related stuff mostly. Only the value 16 as I said will bypass any of the weather code. Please don't play with other values, you may get some nasty surprises! :)

The routine used to read the weather is done roughly once every 4 frames or so, so I'm not sure how that can create stutters 1-1.5 seconds apart. What's your frame rate? It is a bit odd. I'm suspecting something odd with the WEATHER.DLL interaction, or possibly with the memory management calls in FS.

How much memory do you have? Is there plenty free?

It was difficult to see the gradual changes in Test #1. But what I saw at settings of 0 and 16 was very easy to see... stutters and no stutters respectively.

Only 0 and 16 were relevant in any case. Any other variations were probably random or imaginary, though you were probably enabling excessive logging, which may cause stutters if your disks aren't defragmented or if you've turn file write caching off.

All this indicates is that it isn't a good idea to access any weather at all. Though it doesn't seem to matter on most folks' systems it evidently does on others. The routine to read the weather is a direct call into WEATHER.DLL which returns the data in structures which are allocated on the Heap. FSUIPC then copies the stuff out into locations for programs to read, then frees those structures. So there is some memory management involved, but comparing it to the numbers performed by FS all the time it is nothing -- in half an hour FS can clock up several million allocmem and freemem calls. The weather reading is a fraction of that.

In your case maybe you never have any programs that need to know the weather, but mostly that isn't the case. Supplying weather data has been a prime function of FSUIPC and FS6IPC before it for years. I could make it switchable, but I would be worried that folks would forget it was switched off and complain that programs like Radar Contact and so on were going wrong.

I'll just try, initially, slowing the read rate down somewhat -- say to once every 16 frames instead of 4. On my system that would mean weather data changing about once a second. Oh, that reminds me: what is the frame rate you get with FS2004, please -- during these tests, for example?

Do you have the FS frame rate limiter set to below "unlimited", as actually recommended by MS? The general recommendation, for smooth flight, is to gauge what your average frame rate is with it on unlimted, then set the limiter to something below that. the more additions you make, needing processor time, the more it needs restricting. I have mine at 20 fps even though it can attain 30 fps mostly, during flights (10-15 at detailed airports). But this is mainly because I also run WideFS with several clients, and 20fps keeps them smooth too.

So: TEST # 3, check frame rates and set the limiter.

After that, please write to me at petedowson@btconnect.com and I'll try to set up some experimental changes in FSUIPC for you to try. I can't make such attachments here, and any case I don't want any experiments going to a wider audience, not yet in any case.

BTW I have tried, in vain, to find the weather directly to avoid the procedural calls and messy memory management, but even after many many hours of searching and tracing through FS code I've failed miserably. One day I'll have another go. I did ask for help from MS, but they are not really interested these days and simply point to the SDKs, when they come out. There never was an SDK for weather access and control and I don't believe there'll be one now. :cry:

Thanks

Pete

[Pete Dowson]

Feel free to let me know if I can do any further testing,

Hi Greg,

I emailed a list of additinal questions and ideas to you on Wednesday night. Did you receive this?

I also emailed a special version of FSUIPC with several tests for you to try for me, if you would be so kind. These will enable me to at least pinpoint the cause of your stutters, within the weather interface, and possible make improvements or at least provide options for reducing or eliminating them.

This was earlier yesterday (Thursday 18th). I was hoping to learn how they went before releasing Version 3.08, just in case there was anything I can do -- so if you can get back to me soon I'd be most grateful.

Thanks,

Pete