Some of you may have noticed in an earlier post that using photoscenery only and turning autogen off, I had managed to achieve excellent results in my quest to ensure all scenery was loaded at the highest resolution ultra quickly and was clear and crisp as far as you could see. I had gone as high as TBM 500, without noticing any problems. TBM 400 is the recommended setting for use in tile proxy. I have a question to the more technically minded here regarding this setting. What is the theorectical upper limit to this setting? What defines the upper limit IE, what would be the next bottleneck? And what are the likely effects of setting this too high? I have been having some ntd.dll crashes, usually when I quit the game, not too many, but crashes all the same. ( I have turned down my memeory timings to 7,7,7, 24 performance level 7 as suggested previously) I wondered if this was perhaps related and how the higher settings related to what is going on in the rest of the system, perhaps in cinjunction with LOD radius etting, to better understand the overall interaction of these factors.Thx,Mark

when I use bufferpools and a high tbm number my game crashes constantly, redraws textures, etc. i set it to 60-100 max

Mark,Former ACES team member and expert on the FSX graphics engine, Phil Taylor, gave the community a lot of tech info on his blog. Here is a post by Phil on the TBM setting: http://blogs.msdn.com/ptaylor/archive/2007...-week-or-2.aspxNote that settings of TBM on 400 is insane. A lot of simmers don't realise that the TBM setting does not work in the same way as with FS9. I have TBM set to 80.OT: Phil Taylor has now joined up with Intel and is working on the Larrabee project (blog here: http://www.futuregpu.org/)

Mark,Former ACES team member and expert on the FSX graphics engine, Phil Taylor, gave the community a lot of tech info on his blog. Here is a post by Phil on the TBM setting: http://blogs.msdn.com/ptaylor/archive/2007...-week-or-2.aspxNote that settings of TBM on 400 is insane. A lot of simmers don't realise that the TBM setting does not work in the same way as with FS9. I have TBM set to 80.OT: Phil Taylor has now joined up with Intel and is working on the Larrabee project (blog here: http://www.futuregpu.org/)

Hi Ulf, just gonna head over to that link now and read. However, please note, that I am not running the normal FSX settings, i am not using the normal scenery, only photoscenery and NO autogen. AS I said, when using Tile proxy, which is a streaming photoscenery utility, the recommended setting for TBM by the developer is 400. You say it is insane, but I think you are referring to normal FSX usage, as aside from a few crashes, which may or may not be related, I get good frame rates and generally very smooth performance. Plus crystal clear, quick loading scenery. Luiz Felipe has also pointed out my hard drive set up is no doubt also helping.Edit: Just had a quick look and I think i had read most of that quoted in various thread before. I think the key point here is that photoscenery is treated and acts very differently to the normal FSX sceneries with autogen and 3D objects. Does anyone know for instance, if using the bufferpools tweak is of any merit when using photoscenery? I have not used it up until now and have as I said have had good results. Although , not perfect but I still feel that there is a lot of undiscovered or at any rate , very unknown to the common crowd, ground out there relating to the usage of photoscenery within FSX and the effects of the various tweaks. My little discovery of using what was described as "insane " level of TBM for example has resulted in the clear, fantastic looking scenery I had been seeking for months. Yet, no one ( that contributed to the discussion) realised that using photoscenery on its own without autogen, opens up a whole different world of tweaking possibilities, as the parameters are so different.Thx, Mark

Thanks for pointing this out for me, Mark! :( I didn't realize the difference between FSX processing normal textures compared to photoscenery.

Thanks for pointing this out for me, Mark! :( I didn't realize the difference between FSX processing normal textures compared to photoscenery.

Hi Mark:If you are in an experimental mood, try a TEXTURE_BANDWIDTH_MULT (TBM) of 25600 (yes, that reads "twenty-five thousand six hundred" ...use no commas or it will bumped back down to a lowly default in FSX.Cfg).YES, I know it's "an incomprehensibly large value", and it will not work on all systems; but I'll be interested as to whether it works for you.NO, I don't have time to deal with responses as to how "insanely inappropriate" this number is; but I am interested as to whether it works for you.This assumes, in your active copy of FSX.CFG:1.) You are not using a TEXTURE_MAX_LOAD value larger than 1024 (not to be confused with "TextureMaxLoad" without underscore character delimiters)2.) You are not using a BufferPools statement with any visibly assigned value ("//" at the start of that line in FSX.Cfg disables any more than FSX's internal default)NOTE: This has not been tested with high velocity or long duration flights, nor with a fixed frame rate lock setting.PS: I'd be particularly interested in knowing if you see both faster and sharper display of photoreal texures draped onto your terrain mesh.Hope this helps with your "ONLY Photoreal, NO autogen" flight scenario !GaryGB

Hi Mark:If you are in an experimental mood, try a TEXTURE_BANDWIDTH_MULT (TBM) of 25600 (yes, that reads "twenty-five thousand six hundred" ...use no commas or it will bumped back down to a lowly default in FSX.Cfg).YES, I know it's "an incomprehensibly large value", and it will not work on all systems; but I'll be interested as to whether it works for you.NO, I don't have time to deal with responses as to how "insanely inappropriate" this number is; but I am interested as to whether it works for you.This assumes, in your active copy of FSX.CFG:1.) You are not using a TEXTURE_MAX_LOAD value larger than 1024 (not to be confused with "TextureMaxLoad" without underscore character delimiters)2.) You are not using a BufferPools statement with any visibly assigned value ("//" at the start of that line in FSX.Cfg disables any more than FSX's internal default)NOTE: This has not been tested with high velocity or long duration flights, nor with a fixed frame rate lock setting.PS: I'd be particularly interested in knowing if you see both faster and sharper display of photoreal texures draped onto your terrain mesh.Hope this helps with your "ONLY Photoreal, NO autogen" flight scenario !GaryGB

Hi Gary, you have engaged my curiosity now !! Why pick such a huge number ( Obviously, I dont know your reasoning or knowledge behind it)But I am definitely gonna try, right now...I was also not aware of the TextureMaxLoad value without underscores. What does this do that is different to the other one. And why should | have a value of no higherthan 1024 when I try your suggestion, I usually use 4096 , for my HD REX textures.Sorry for the questions, I like to know the rationale behind the instruction...reporting back soon..

Hi Gary, I just tried your suggestion.I can report that it was difficult to notice any radical difference, although I felt ( and this was just a feeling, not measured) that the sim was quite smooth. Frame rates seemed a little lower, but more steady. The scenery looked pretty crisp, i have to say, but I did manage to find areas not fully focused, I was flying at around 150kts. interesting..... Why did such a huge number have so small an effect? Obviously more to this than meets the eye....

Hi Gary, I just tried your suggestion.I can report that it was difficult to notice any radical difference, although I felt ( and this was just a feeling, not measured) that the sim was quite smooth. Frame rates seemed a little lower, but more steady. The scenery looked pretty crisp, i have to say, but I did manage to find areas not fully focused, I was flying at around 150kts. interesting..... Why did such a huge number have so small an effect? Obviously more to this than meets the eye....

Hi Mike:Am I correct your video card is the model "BFG GTX285OC" with 1 GB video memory ("VRAM") on it ?http://www.bfgtech.com/bfgrgtx2851024ocbe.aspxAlso, feel free to try it again (if you haven't already !) with the TEXTURE_MAX_LOAD ("T_M_L") value at 4096... but be sure to test these FSX.Cfg parameter settings over a "definitive" photoreal scenery area (ex: FSAddon's Plum Island or a "good" FSX-file-format MegaScenery based on native color aerial photos etc..Additionally, I'll be interested in whether you see a change in responsiveness of FSX when cycling between various sequences of camera views ("A" key) and between inside-cockpit / outside-locked spot views.I would assume that for terrain ground cover, "HD REX textures" (like most of the default FSX scenery textures) although in part originally derived through special processing of photoreal aerial imagery, may perhaps be in their final distributed form, a type of "hybrid" land class scenery tile (reused throughout the FS world).This should be distinguished from custom photoreal imagery tiles based on minimally processed aerial photos in the strictest sense of SDK definition.It is a "true" FSX photoreal ground tile scenery flight environment (aerial photo tiles used in only "1" place) that we are trying to optimize system settings for in this test case; the twin Scenery/Texture folder pair for such a specific scenery package should likely have "*.AGN" files in the \Texture folder.Thanks for your feed back on this investigation of parameter settings for "ONLY Photoreal, NO autogen"; I'll be interested in reading how the "T_M_L" of 4096 worked for you with the "TBM" of 25600.PS: Be sure you don't turn up the FSX Level Of Detail (LOD) radius slider beyond 1/2 to 3/4 maximum for initial photoreal testing purposes.And when you get an "unfocused/blurry" tile, go to slew mode, top-down view and zoom repeatedly in/out from the aircraft to ground to see if the tile sharpens up quicker.GaryGB

Hi Gary, in answer to your points, yes that is my video card.The rex textures are usually just airport runways and water ( as well as clouds) Interestingly, when I just used used 1024 for thexture max load, I did not notice my clouds looking any less attractive....I will test with 4096, but I did not notice any better responsiveness in the first test, just a feeling of more smoothness. With my lower settings of 400 and 500 for TBM I was finding virtually no low res tiles displayed anywhere, and the few areas that did , snapped up to hi res pretty quickly. But it is really difficult to quantify these things accurately, the 25600 setting you suggested did not seem very differnt as I said.All the photosceneries I use ( Megascenery , Blue sky scenery ) use .bgl files. The texture folder generally only contains .agn files, which I presume are related to the autogen, and therefore not used. in fact many of the packages for blue sky have no texture folder, some even had .bmp files in the texture folder. I think Gottfried is now using a differnt system that just produces .agn files, but you would have to ask him. yesterday when i tested, I had the LOD radius at my normal setting of 8.5 I will reduce to 4.5 tonight and report back.Can you enlighten me as to your background and interest in this and why you want me to test these settings ? Have you not tested this yourself ? Just curious!!!! Oh, and its Mark, not Mike !! cheers, speak later

This is interesting. I have copied and pasted a section from Christian Buchners usage manual ( which incidentally is a goldmine of other scenery related information, read it here: http://www.edtruthan.com/tileproxy/manual.htm#6Here is the particular section of interest in the context of this discussion:

Starting with FSX SP1 Microsoft's Flight Simulator X spawns multiple child threads that willoccupy all of the CPU cores multi-core PCs. These additional threads are used to accelleratescenery and tile loading and processing. The first core always performs the graphics renderingand each additional core will trigger load requests for terrain tiles. Therefore a quad core chip may show up to three times the tile loading speed of a single and dual core PC.

However I recently found a really unfortunate bottleneck in this system: The other cores sendtheir I/O requests back to the first core, where they are responded to by a thread fiber that performs all non-graphics rendering related tasks. So essentially the other cores may end up waiting for the first core to return the graphics data. You can nicely see this in Task Manager when you enable the

Anything over a TBM of 400 is IGNORED400 is the max hook value possibleAnd we have had this conversation about TBM befrore... Photoscenery is not handled in render like textures

Hi Nick, hope you are well!!Thats interesting, I wondered if there was an upper limit to the value. Why do think GaryGB is asking me to set it at 25600? ( To show me that higher values do nothing perhaps? ) I cant work out what he is getting at really..., hoping he will explain a bit more.Did you see my question re the fibre frame entry ( we cross posted I think) Have you read through Christian's instructions? Its pretty interesting stuff! Also Nick, he recommends using the in game anisotropic filtering selection, can you see any mileage in this ? ( perhaps he was not aware of advantages of 8 x s filtering? ) I still use the settings as recommended by yourself with nhancer. Slowly, little pieces of the puzzle are falling into place for me! I know we have discussed before, but there are many gaps in my understanding of all this, hence my persistent questions!!!!best, Mark

MarkUnlike settings in the config such as LOD_RADIUS=X and TEXTURE_MAX_LOAD=X which do in fact have a range above their default value, and, unlike edits in the config such as MipBias=X which do not exist but can be added BECAUSE the hook exists, TBM has a hard limit which is 400. Anything happening with a value above 400 is anyone guess in terms of what the sim is doinga. Defaulting to the max of 400b. Defaulting to the default value the PerfBucket=X sets for the installed hardware via Aces predefined rangle for hardware ability (between 10 and 40, typically 30)c. Defaulting to the lowest value possible in range of the hook value (Which I recall from my notes is 10 but do not have them in front of me right now)I did not ask about that when I inquired about this subject however I did ask about TBM MAX and a few other settings in which I was informed was the same hard limit as FS9.. 400 and that anything above is ignoredPlease see Phil Taylors post and quote from Adam at Aces about FFTF changes with SP1/SP2 http://blogs.msdn.com/ptaylor/archive/2007...f-the-week.aspx==================================Here are 2 quotes from Adam, verbatim:"FIBER_FRAME_TIME_FRACTION determines the maximum amount of time per frame that we will run fiber jobs on the primary thread. We measure how long it took to simulate and render and then multiply that time by FIBER_FRAME_TIME_FRACTION to determine how long to run the fibers. For example, if it took 10 milliseconds to simulate and render and FIBER_FRAME_TIME_FRACTION was set to the default value of 0.33, then we would allow the fibers on the primary thread to run for up to 10 * 0.33 = 3.3 milliseconds. For fraction values of 1.0 and 2.0, the time given to fibers would be 10 milliseconds and 20 milliseconds, respectively.The operation of FIBER_FRAME_TIME_FRACTION on single core machines has not changed since RTM. On multi-core machines in SP1, we moved many fiber jobs off of the primary thread and onto secondary threads. Since FIBER_FRAME_TIME_FRACTION only affects scheduling of jobs on the primary thread, it will have less of an impact on the performance of Flight Sim on multi-core machines. In fact, we moved so many jobs off of the primary thread that there probably isn

All very interesting stuff!!I am still not clear wether there is any benefit in A) Actually having the fibre frame entry in the config file on a machine such as mine ( I read Adams comments as saying " its there but it does not do much now on a multicore machine post SP1" - have I got that right? ) :( Changing it in any way ( up, or down ) IF you run photoscenery ( the logic being that due to that type of scenery being handled differently by Flight sim, there may be mileage in it, as lets face it, whenever anyone talks about these tweaks, they assume you are using normal FSX textures and autogen) Christian, presumably knows what he is talking about, so for him to suggest using that value, he must have found a good reason for it. The same as the high setting of TBM for using photoscenery, which has apparently been "out there " for a while, but is definitely not common knowledge. Edit: I dont know why the "cool" emoticon came up ther, it should read "B"Also I think there is still much to find out about the interaction of TBM, LODradius, FFRT together in realtion to photoscenery usage. Finally Nick, what do you think GaryGB is getting at? ( in the absence of a reply from him on the subject!) cheers, Mark