MS Flight Improvement Wish List

[mgh 89]

The whole point is that you don't have to do that -- settings are adjusted while playing to keep framerate smooth.

You suggest that the computer decides how much of X, Y, and Z the user gets to maintain a smooth (constant) frame rate?Surely, that implies that if the load imposed by X and Y becomes too great then Z will just be dropped and then if the load decreases Z will be picked up again? That would introduce a time gap in Z which it might not be possible to deal with. If Z were AI aircraft it would result in the aircraft not being updated making it impossible to continue their flights.The alternative is that the system gives X, Y, and Z a fixed fraction of the available time and varies them all up and down together. Either the system determines the fraction based on priority or the user has to set it. In the first case the user has reduced control: in the latter the user has to make decisions as with sliders. Either way I don't see any significant advantages.There are other ways to make the best use of the available processing time. I have no idea if they are used or not in FSX.At the heart of the simulation is the numerical integration routine that advances time step-by-step. There are variants that can estimate the error and adjust the step length accordingly to minimise the amount of time needed for that part of the calculation. Another is to identify those features by their nature vary more slowly than others. Those may be processed as a slower rate than the main simulation loop. The simulation may need to run at 50 fps for visual appearance but there's no need to update all variables at that rate - for example fuel flow.

[n4gix 4,955]

Hello BillFSAssist 2002 by Lago is the one you are thinking ofHere it is courtesy of the wayback machine.http://web.archive.org/web/20070503105415/www.lagosim.com/default.asp?act=scheda&p=100002〈=eng

Thanks! Yes, the ever-so-useful "Framerate Wizard..."My point of course is simply that since LAGO clearly accomplished this from outside of FS as an addon, it would so much simpler for such a feature to be coded into the core platform from the beginning!

[Guest veeray]

n4gix: you have no idea whether it works. You're basing your entire argument on an archived page from 10 years ago. Little faulty logic there.

[JSkorna 137]

Ummmmmmmm.........many of us were around a lot longer than 10 years ago and we DO know what worked and what didn't and FSA DID work.Who remembers when we were able to update cloud textures during a flight WITHOUT having to re-start FS?

[Guest veeray]

Yeah well some of us are real programmers and can tell you in an instant that you can't solve an equation with 15+ Unknown(sliders) and 3 constraints(preferences) in a short time or for that matter in real time. Add all that up and you now have something far more complex then the physics engine itself.

[duckbilled 310]

It seems like FSX already does this but the side effects are horrendous.My understanding (limited I must admit), is that the HIGHMEMFIX=1 tweak reverses the Aces attempt to accomplish what you guys are talking about. Without the tweak, FSX will dump start to dump things if it fears that there is not enough memory remaining to keep flight going. The tweak stops FSX from dumping things (scenery, ac exterior, ai, VC textures...). After applying the tweak, I have not had any crashes. Before the tweak, I had issues with almost every flight.I guess my point is that all of those sounds great but we must be able to control the enhancements. Without *******' research, FSX would have been a completely different experience for me. To stop this Aces enhancement would have required nothing more than a check box.

[Guest veeray]

Duckbilled: Not exactly. What you are speaking of is an optimization technique used within the code. Obviously there is some benefit to kicking things out of ram, but when it's things like 10mb liveries reloading then it is a show stopper and obviously something Microsoft didnt' anticipate. What the others are suggesting is taking the sliders and somehow coming up with the best combination among the over 10 million possible combinations. Where as I claim along with others it's a waste of processing time that could go to better use in actually rendering more graphics. :)

[n4gix 4,955]

n4gix: you have no idea whether it works. You're basing your entire argument on an archived page from 10 years ago. Little faulty logic there.

That is a rather arrogant assumption, especially since I introduced the product with the question "...does anyone else remember..."I used LAGO's product from the day of its release until finally moving on to FS2k4. It worked very, very well and had slightly under a 5k memory footprint.I've been a professional programmer and systems analyst for nearly forty-five years. I cut my eye-teeth on assembler... :)

[january 19]

-----------I've been a professional programmer and systems analyst for nearly forty-five years. I cut my eye-teeth on assembler... :)

Hah- another young whippersnapper! Humbug- the only way to cut eyeteeth was using machine language on an IBM 650.Assemblers and compilers were in the future for the softies- Real He-Men coded in machine language using a huge desk size paper template of machine memory- all 2000 10 character words!In fact at that time we were amazed when SOAP (Symbolic Optimized Assembly Programming) came out. Just imagine- a program that could actually modify your program and change the addresses of where you had stored data!!!And that spelled the end of the He-man era! Wow- that was 55 years ago!Alex Reid

[n4gix 4,955]

Hah- another young whippersnapper! Humbug- the only way to cut eyeteeth was using machine language on an IBM 650.

Alex, my very "first" program was in binary... toggled into the console of an IBM 1450. What a thrill! Fifteen minutes (and three mistakes later!) seeing "Hello World" being hammered out on the ancient line printer... Whee!

[january 19]

Alex, my very "first" program was in binary... toggled into the console of an IBM 1450. What a thrill! Fifteen minutes (and three mistakes later!) seeing "Hello World" being hammered out on the ancient line printer... Whee!

Fr Bill- Those were the days! After about a year playing wth the IBM 650 (& 604 prior), the powers that be, realized that I had no future as a programmer and simply put me in charge of about 30 of them to try to figure out what the heck a computer could be used for! (The "Powers" were right- I still get all sweaty trying to download something. Nothing like a handful of 80 col. Hollerith cards for transferring data!)Alex Reid

[Tierborn 190]

Hah- another young whippersnapper!....Alex Reid

I knew dirt .. when dirt was a baby ... so i win  :(

[mgh 89]

My first coding was on a Ferranti Pegasus with 8 accumulators, 56 words or memory and all of 5120 words on a magnetic drum. To keep processing speed up we had to keep tracvk of whjere we'd put o data on the drum in order to avoid rotational delays where retrieving it.

[january 19]

My first coding was on a Ferranti Pegasus with 8 accumulators, 56 words or memory and all of 5120 words on a magnetic drum. To keep processing speed up we had to keep tracvk of whjere we'd put o data on the drum in order to avoid rotational delays where retrieving it.

mgh- sounds much like the IBM 650 where we used a huge paper sheet with 2000 graphical slots corresponding to the 2000 10 character words in the drum memory. There were tables of instruction execution times and knowing the rotational/latency time for the drum, one could, if very clever, figure out where to place data or an instruction to avoid having the drum take an extra rotation! Equipment for a programmer included a number of pencils and a very large eraser. Those who got good at this went on to become highly valued programmers- those who didn't, like me, went on to esoteric activities such as devising schemes for using computers in the first place. (And also, sometimes, establishing that using a computer for some jobs would be wildly uneconomic.) Alex Reid

[mgh 89]

it sounds very much like the IBM 650. We were delighted when we got Aotocode which enabled us to write:

v215 = v0 + 1v215 = v213 / v215v218 = v213 - v215

One operation per line and no symbolic names - nddd for integer and vddd for floating point