Here's one of the tools that was used, and I used it today with the 780.. I had remm'd the TextureMaxLoad and SwapTimeOut, so there was no external interference with the outcome:

 

Memory Clock=6000; GDDR5; Target FR=33:G.Texture Res.=2048, results were:

 

[Display]

MAX_TEXTURE_DATA=4364

TEXTURE_BANDWIDTH_MULT=213

UPPER_FRAME_RATE=33

 

[Graphics]

TEXTURE_MAX_LOAD=2048

 

Are you sure that you have to enter the "final" speed of your graphics card RAM? Or do you need to specify the "true" clock speed, which is "only" 1500MHz with a 780? I somehow learned that TBMs over 120 will anyway be ignored, with 1500MHz in your little tool you get a value of 107. Confusing. And what the heck is this "MAX_TEXTURE_DATA=XXXX" line? Never heard of, never seen and it is again something that you have to add to your FSX.cfg. As mentioned above, I really like to minimize the amount of additional lines in the FSX.cfg...

[MAIN]FIBER_FRAME_TIME_FRACTION 0.15 [TERRAIN]SWAP_WAIT_TIMEOUT=10 [DISPLAY]TextureMaxLoad =15 

 

The above setting can get some blurry effect because it is not an aggressiv setting but in the air it works pretty good on my system.

 

Michael 

 

I like your settings Michael!

 

One thing I noticed though was in areas of more complex add-on scenery, I'm getting occasional momentary occurrences of black texture tiles. Is there a parameter I can adjust to get rid of those?

Are you sure that you have to enter the "final" speed of your graphics card RAM? Or do you need to specify the "true" clock speed, which is "only" 1500MHz with a 780? I somehow learned that TBMs over 120 will anyway be ignored, with 1500MHz in your little tool you get a value of 107. Confusing. And what the heck is this "MAX_TEXTURE_DATA=XXXX" line? Never heard of, never seen and it is again something that you have to add to your FSX.cfg. As mentioned above, I really like to minimize the amount of additional lines in the FSX.cfg...

 

Final speed?  I don't know, Ankh - I didn't write the applet, it's not "my little tool" - and, to my knowledge there were no particular instructions with it.  It was published about five or six years ago, coincident with several other posts containing other slightly different and more complex calculations, plus a great long post at one of the other sim forums. Then, perhaps two years ago, an enormouse rebuttal was written on SimForums.com, quoting everything that the authors said, posted, calculated, postulated - by a "third party", and I suspect many people simply shut up about it, given the vehemence of this response. 

 

Regarding this little applet 's results - I'm not even sure that it's a valid calculation, as i said in my post - "I'm unsure of the calculator, given this result, as it was very similar to the stock TBM of 120 and default TextureMaxLoad of 6". i.e. It made no difference -moreover - the four or five flight tests which I carried out resulted in Michael's settings being validated, the "little tool" showing no advantage whatsoever. 

 

Additionally - I reckon your own tone, too, AnkH, is a little more aggressive than warrants my part in the discussion, considering that I'm simply responding to Michael's interesting settings, as it harks back to those early times with the new FSX. I don't claim any knowledge or expertise in the use of the "little tool", nor do I claim any spectaular results by it's use - neither am I suggesting you add any "Never heard of, never seen" additional lines to your config.

 

pj

Additionally - I reckon your own tone, too, AnkH, is a little more aggressive than warrants my part in the discussion, considering that I'm simply responding to Michael's interesting settings, as it harks back to those early times with the new FSX. I don't claim any knowledge or expertise in the use of the "little tool", nor do I claim any spectaular results by it's use - neither am I suggesting you add any "Never heard of, never seen" additional lines to your config.

 

pj

 

Paul, I really have to apologize if my tone is to aggressive, this was never my intention. I am really happy that I found this thread and some of the mentionened settings here actually led to an improved FSX experience now. The only thing I am always a little bit cautious about is adding lines to the FSX.cfg of which I do not know what they are all about. It seems that keeping as much as possible out of your FSX.cfg should really be the way to go with a modern, powerful system nowadays, that's why I mentioned this.

 

Update: adding swap_wait_timeout, TextureMaxLoad and Max_Texture_Data lines to the FSX.cfg did not improve anything on my build. So I deleted them again. I now only use TBM=213, together with unlimited frames inside FSX and the Inspector settings that were posted in this thread. Like that, I get the best results when flying over the terminals at LSZH with AI Traffic enabled.

 

So, I really appreciate the help I get here in this forums, with all the DX10 guides etc. and I am really sorry, if my tone seemed to be aggressive in some posts from me.

Thanks for that, AnkH: The problem we have with any forum, is the lack of face-to-face contact, where that all-important voice tone and body language is not present, and when writing rapidly, often emotionaly, one generally doesn't recognise this, and so the true intent may become masked. Emotion, which is more readily conveyed, can then incorrectly interpreted by the reader. 

 

OK. 'Nuff said. Life is good, once again. 

 

TBM of 213? Yes - that's what that calculator says, but that number is predicated upon the type and amount of texture(s) to be drawn, and the ability of the system to deliver them. The "120" "limit" comes from Nick Needham, and may well be a reasonable number - but it may also be low for us DX10'ers, given the speed and memory of the newer GPU's on the market. 400 is still valid for photoscenery (which I don't like, nor use. MS Earth/SOCAL was my first and last a long time ago), so I'm pretty sure that numbers higher than 120 can be used by FSX: if the DX10 graphics pipeline can manage the data volume, then why not? 120 has been "the limit" for a few years now, and - as with almost every quoted - all DX9 - guide, is based on DX9 graphics, and that is not what we are dealing with. Raf and Phil, in their blogs (2007, but still quoted as if they were current - even when taken from pre-SP2 blogs!) mention lower and different parameters than SP2 gives us, with very little detail on almost all of the parameters. Often-quoted BufferPools, FFTF and Swap time-out settings are also victims of oblsolescence because of the current speeds and change of architecture.

 

In this light - my thinking is that Michael's numbers, these numbers from the calculator or from any of the early formulae, can be valid for any given system, but only rigid test flights can confirm them. I have not done sufficient number of tests to disprove any of them.

 

Regarding the TBM of 213, AnkH - TBM is not used by FSX when unlimited fps is used, with Nick also stating thte number can only be in multiples of 5. Swap and mtl do make subtle changes to the performance of the sim, but I think FSX needs to be running as hard as possible in order that the changes stand out.

 

All the best,

 

pj

 

Current:-

Swap = 6

TextureMaxLoad=9

TEXTURE_BANDWIDTH_MULT=120
UPPER_FRAMERATE_LIMIT=33

TEXTURE_MAX_LOAD=2048

FIBER_FRAME_TIME_FRACTION=0.1

LOD 6.500

Ok, interesting. I still tend to forget some aspects, like that TBM is not used with unlimited FPS. Too many things to consider, but I try to keep up. I did a full test flight now from LFPG (Aerosoft Scenery) to LEBL (Aerosoft Scenery) with the Overland 737-800 in the VC, UT2 set at 75%. Guess what: without this ominous AffinityMask=14 entry, my FPS (limited or unlimited) went down to 11-13 while taxiing to the runways at LFPG. Unplayable. So I did put this AffinityMask line back in and stunningly, my FPS while taxiing went up to 18-20FPS. This is a MAJOR difference, which I again did not expect. Really confused now, why does this AffinityMask entry has such an impact on my machine (i5 maybe the reason?)? I do not know, but I will stick to that now. I also again realized, that limiting FPS via FSX.cfg pushes down my FPS in all situations by 2-3FPS. Taxiing around LFPG with 18-20FPS, I can simply not afford to loose those 2-3 FPS. But then, TBM is not functional. Hmmm... I will give your settings a try, maybe the combination of limited FPS, the swap line and TextureMaxLoad is the trick. Who knows... :lol:

Simplified:

 

Try my setttings with swap=10 with or without Limited FPS. I use unlimited FPS as it gives me so smooth experience with my GTX780SC.

 

For now i am testing TML=27,SWAP=4,TBM=80 and its not as smooth but close, Going back to SWAP=6 for test 

 

Michael

Ok, I did some additional tests with strange outcomings. My mentioned test scenario seems to give the most massive hit on performance on my rig. I tested several configs now and in contrast to my previous post, I do not get more than 12-13 FPS when taxiing on LFPG with the SMS 737-800 and UT2 traffic set to 75%. No matter which setting I use, obviously there must be something wrong with this scenery, AI traffic and DX10. However, test flights above Switzerland Professional X also resulted in strange FPS numbers: while adding swap=10, TML=27 and TBM=120 or 215, I really get more FPS, between 50 and 60 now. BUT: if I fly above the scenery, I instantly realize that my rig cant keep up with the textures, resulting in blurries after some minutes of high-speed flight (250+ kts). Furthermore, there is the strange phenomenon that performing a 180 degrees turn results in a drop of the FPS down to below 20. Another turn does not fix it, it stays that low. Strange. Without swap and TML, I get slightly lower FPS overall, but turning back to where I came from does not result in a FPS drop. I also have a better and faster texture load, resulting in less blurries. Any ideas?

 

Next thing to test: I reduced now my AI traffic further down to 50%. As I still have a really filled LFPG scenery with 75%, I think I could live with 50%. Or do you think that it is an intrinsic problem of the scenery with DX10? I am pretty sure that I did not have that low FPS running the scenery in DX9, but I can be wrong.

Well, Michael, I have to hand it to you. You just pushed me back a few years to a slew of posts which calculated some of these figures - not started by me by any means, but I quickly became a "convert, as "the science" and math were quite logical, and - moreso - gave some good results. There was a whole webpage/XL spreadsheet, too, devoted to this on the DigitalThemePark site at that time (it has since been pulled) which listed dozens of GPU/graphics settings combinations. Unfortunately it was torn to pieces, and ridiculed by a (then) most respected "Guru" of FSX, and as a result, the whole concept died.... well, my friend - YOU just resurrected it! I went back to my "old figures" - TBM=157 (GTX580), TextureMaxLoad=21 - and guess what? With frames set at 32 in FSX, the A2A C172 at Anacortes gives an-almost-ROCK-SOLID 30 frames, with none of the "usual" dips to "24 - 27 - 26 - 19 -30 - 29 - 24 - 28 - 25 - 29 - 29 - 27 - 29", etc.,.... It sits at 31 - 31 - 30 - 32 - 30 - 32 - 30 - 31 - 3- 30 - 29 - 32 - very much more smooth, and completely without those dips. As you know - I run a pretty "Lean Machine", at 5 with a 780, and since following your lead, I have once again been completely surprised by another "discovery"!

This means a complete re-visit of the calculation texts, and sites, if they exist, plus more testing and testing and testing. It will result in a partial re-write of the "How-To" doc, and so - PLEASE - feel free to add your 20 cents!

 

TEXTURE_BANDWIDTH_MULT=157

UPPER_FRAMERATE_LIMIT=32

TextureMaxLoad=21  //  15  Michael Moe  //  6 = default

 

All the Best,

 

pj

 

 

So my friend, did i waste your time ? I have difficulty seeing anything from a saved point anyway. :-)(in fsx)

 

Michael