info for Pete on FSUIPC4.4 and FSX

[rhogen]

Pete,

The FSX file in your SDK has a question mark beside address 3070 in the "write" column. I just wanted to let you know that my ArrestorCables program successfully writes to that address and it works fine, just like in previous versions of FS, but only with a commanding rate change.

The key thing that I noticed about writing to that address is that I had to slow down the rate of commanding by a factor of 12 compared to FS9 and prior versions. My timer used to send real time changes to that address at an ambitious 10ms cycle (bear in mind that I'm doing real time changes to the aircraft speed to simulate catapulting and arrestor cables accelerations!). Using that rate of commanding in FSX seriously bottlenecks Simconnect (as far as I can tell), resulting in major delays in the commands making their way into FSX. All I had to do to make things work was to increase that interval by a factor of 12 (120 ms interval now) and increase the scale of the changes sent. One might be concerned about a stuttering effect, but 120ms seems to work very smoothly.

The key point being that after Simconnect is overloaded, the rate at which it recovers does NOT seem to be proportional to the commanding rate - it's like it has a clot and the artery just plugs up! I leave it to you experts to understand what's going on in Simconnect at a low level when it crosses that bottleneck threshold.

fyi,

R.Hogen

[Pete Dowson]

The FSX file in your SDK has a question mark beside address 3070 in the "write" column. I just wanted to let you know that my ArrestorCables program successfully writes to that address and it works fine, just like in previous versions of FS, but only with a commanding rate change.

Okay. Thanks. I'll mark it so.

...

The key thing that I noticed about writing to that address is that I had to slow down the rate of commanding by a factor of 12 compared to FS9 and prior versions. My timer used to send real time changes to that address at an ambitious 10ms cycle (bear in mind that I'm doing real time changes to the aircraft speed to simulate catapulting and arrestor cables accelerations!). Using that rate of commanding in FSX seriously bottlenecks Simconnect

Hmmm. I wouldn't be surprised, at least with FSX original and SP1, since the SimConnect action uses TCP/IP communications. Very ponderous, even with in-process operations. Do you get the same with SP2/Acceleration applied? That is using named pipes which, whilst not the bee's knees, should be an order more efficient.

The key point being that after Simconnect is overloaded, the rate at which it recovers does NOT seem to be proportional to the commanding rate - it's like it has a clot and the artery just plugs up! I leave it to you experts to understand what's going on in Simconnect at a low level when it crosses that bottleneck threshold.

I think it is just what it looks lie, queuing problems, exactly the same as on overloaded motorways/freeways or trying to force too much water through a small pipe.

Regards

Pete

[rhogen]

To answer your question, this experience was with Acceleration/SP2. I never ran FSX without Accel/SP2.

thanks,

RH

[Pete Dowson]

To answer your question, this experience was with Acceleration/SP2. I never ran FSX without Accel/SP2.

That's rather sad, then, to find the pipe mechanism cannot cope with such rates of update. :-(

Pete