I use the same Riva you mention, and turned off the IO setting.It won't recognise the driver, maybe cause I use the latest version: 197.15.But then again, just wanted to experiment, I use EVGA with no problems and issues, excellent control of clock and fan...

I can only say, tried the latest Rivatuner with the 197.13 driver, installation went ok, checked off IO, driver recognized, all set up correctly. Though it didn't help really - but maybe it helps someone else. My sim crashed nevertheless...Another observation I got to share (same as someone else here I think):When BP @ 500,000 or 1,000,000 or even higher, I get stuttering, that's known. I also noticed that GPU is not being used as much when BP is NOT 0.When I set BP to 0 (FPS unlimited of course), it's nearly all the time 98%+.

Hey allFound I had to roll back to 196.75 as 197.13 was giving me some problems with switching display modes with alt,tab & alt enter where the display was sticking on the last render frame,back to 196.75 no more sticking....hummmSorry to hear your still crashing Word Not Allowed,just thought id post what id found as it was working on my system but im sure we'll get to the bottom of making this tweak stable for everyone one way or another ;)LatersGizmo

I'm talking about allocated, a fixed, software managed pool. We are not talking about a mirrored, if that were the case you will see an increase in BOTH Video Ram AND system RAM. And, interestingly enought, the system memory, particularly, the fsx.exe process grows 'proportionally' to the number of bytes you set in BP!! easily tested, just see for yourself ;)Yeah, I'm aware of what he said, and ALSO about what Rafael said. The 'behavior' of the fsx.exe process using BP=0 vs. BP=xxx is CLEARLY visible to anyone using perfmon, BP=0 memory usage is LOW, and if you changed BP to a value like 500 000 000 you'll see how the fsx.exe grows 'exactly' 500MB change that to 200 000 000 and check again... it IS a static, allocated, system/CPU/FSX managed memory pool.Its actually, the other way around.. BP=0 reduces system memory utilization, CPU utilization, and GPU utilization and increases 'slightly' Video Ram utilization, so, lets suppose you do your own measurement and observe exactly the same, (assuming my observation was correct of course) what would you make of the above?via perfmon counters for system memory (including some CPU counters as well) for GPU usage and memory utilization I'm using the GPU-Z tool (which is the only way to read video ram usage on my nvidia card)this is (so far) observations, but the good thing is, ANYONE can confirm or deny what I'm seeing.

This is interesting, not so much that FSX.exe would grow with BP, it probably should as its trying to read or calculate a head of time to be ready WHEN the GPU needs it, but when does the GPU make use of it and should we expect that it will all be used at once? Not sure if we should think so, but I dont think it would be that far behind the pool either, also looks like the GPU only takes what it needs, not the whole. yeah the math does looks funny in light of what Aces have posted about it, but your on to something.Along with that thought this is also what I have observed about different BP:After setting my system back to somewhat "normal" I7 state from running high-speed memory/uncore back down to running memory @ 1600 7-7-7-20 & uncore/QPI at stock 8:9, and keeping everything else the same in my setup with high High image IQ 16xAF 8xAA-combined + Gama and medium large LOD of 6.5 AG on dense, scenery at Very dense and all settings maxed as I usually run (full mesh, water etc) This: - Any low BP Settings of =0 or just a few kilobytes =4096 along with the smooth flight, the >terrain does not keep up with the flight and if faster then say about 220knts and above turns to a blurrie mess. With default BP or higher it stays sharp and can fly it fast down at ground level. Anyone else?Also because of the maxed IQ settings BP=0 is quickly back to corrupted or balck screens, very small BP=4096 still allowed for longer Smooth flights, but again Blurries mess if flying any kind of med-high speed...So a question then:This makes me wonder where I got the thought that BufferPools had something to do with SP2 new AG Batching proccess, now after digging arround in all the info I dont see the connection. Sure we may gets spikes from AG with BP=0 when panning etc but instead of being directly related LIke I think some of us have assumed, now I think its more of a symptom. So now I am starting to lean more towards the idea that BP =index buffers to store vertex and geometry that tries to keep the terrain mesh geometry and texture indexes at the ready to reduce blurries. ??? and not just limitted to indexing AG and maybe not anypart of the SP2 batch proccessing of AG.BTW RivaTuner has the ablility to display GPU and memory usage, or at least it used to.

I installed and benchmarked the new nVIDIA 197.13 drivers yesterday. Good drivers on my system with a significant perfomance gain. I got inspired by a post by David Roch and tried the new drivers without using nHancer.These are my settings in the nVIDIA control panel with the 197.13 beta drivers

If you are only using the NV CPL then your not getting the 8xAA you selected as well as some other NV settings in FSX.The driver connection has been broken for sometime, you will still need a RivaTuner/Nhancer type of program to overide if you want to use those settings. Maybe explains your FPS increase? Very unlikely NV has done anything new to support FSX since way way back and frankly there just isn't much FSX GPU code there to optimize. Mainly AA, AF and a bit of shader, as you know FSX geometry is mainly done on the CPU unlike other types of games.

197.13 works fine on my system. Very Dense Scenery and Autogen, FPS lock @30 FPS, 10,000,000 BP setting and a 70 TBM = Nice smooth flight

197.13 works fine on my system. Very Dense Scenery and Autogen, FPS lock @30 FPS, 10,000,000 BP setting and a 70 TBM = Nice smooth flight

Say, did you ever compare the FPS between 10MB and 0? I don't know how's with you, but running 10MB BP even on my sys I don't get steady 30fps... that being over FTX or anywhere busy.

Before I try the new drivers, I just wanted to re-affirm to those who are not believers:Performance doubled (fps wise) since upgrading my drivers from the 182.50s to the 196.21 and adding BP=0.I will try the latest driver just to see if it gets rid of the flashes and spikes. I'll post my results.

100% increase in FPS for changing drivers? Those are really big numbers. Did you go from 24 to 48, 12 to 24? Doubled at low end, or high end, average, or what, if you don't mind me asking?Stephen

100% increase in FPS for changing drivers? Those are really big numbers. Did you go from 24 to 48, 12 to 24? Doubled at low end, or high end, average, or what, if you don't mind me asking?Stephen

Well, I can nearly believe those numbers. BP=0 itself adds about 60-70% in the FPS arena, and add to that a good driver change, from a really old 182 driver, it's possible...He DID write "196.21 and adding BP=0."

This makes me wonder where I got the thought that BufferPools had something to do with SP2 new AG Batching proccess, now after digging arround in all the info I dont see the connection. Sure we may gets spikes from AG with BP=0 when panning etc but instead of being directly related LIke I think some of us have assumed, now I think its more of a symptom. So now I am starting to lean more towards the idea that BP =index buffers to store vertex and geometry that tries to keep the terrain mesh geometry and texture indexes at the ready to reduce blurries. ??? and not just limitted to indexing AG and maybe not anypart of the SP2 batch proccessing of AG.

Another interesting observation, check CPU usage when using 'any' poolsize other than 0... it goes to 100%. PoolSize=0 cuts CPU utlization in half! (note that, in order to actually 'see' any CPU reducion, you need to run the sim at UNLIMITED and LIMIT the frame rate with a tool like Frame rate limiter or via VSYNC) otherwise, it always show at 100% (I run FSX with VSYNC Off by the way)If you think about it, and you observe that, in fact, BP=0 reduces CPU utilization considerably, there is enough juice to do other things like terrain mesh, AG, scenery objects which will explain the sudden increase in performance when using BP=0There are exactly three options that can be used inside [bufferPools] one is, of course PoolSize, the other two are UsePools (which can be 1 or 0) and RejectThreshold which defaults at 1 Million (no idea what it does)Some other hidden settings I played with: they go in the [Graphics] section:ALLOW_SHADER_20ALLOW_SHADER_30SHADER_CACHE_VERSIONSHADER_OPTIMIZESHADER_PARTIAL_PRECISIONSTALE_BUFFER_THRESHOLDHIGHMEMFIXDPUPBUFFERSIZEI have tested almost all of them, they really make no difference... but the only ones worth investigating are the last 3.

Say, did you ever compare the FPS between 10MB and 0? I don't know how's with you, but running 10MB BP even on my sys I don't get steady 30fps... that being over FTX or anywhere busy.

I am not seeing the huge FPS increase some are seeing using 0 BP. I actually get better performance using 10 for some reason than with 0. With 0 I get alot of micro stutters while flying. With a setting of 10 it is smooth as silk. In heavy hitting areas I get lower FPS using 0 BP. I tested this at JFK at I get 5 FPS higher not using 0 BP. I run with UTX/GEX/FEX for my scenery and clouds/water.

Another interesting observation, check CPU usage when using 'any' poolsize other than 0... it goes to 100%. PoolSize=0 cuts CPU utlization in half! (note that, in order to actually 'see' any CPU reducion, you need to run the sim at UNLIMITED and LIMIT the frame rate with a tool like Frame rate limiter or via VSYNC) otherwise, it always show at 100% (I run FSX with VSYNC Off by the way)If you think about it, and you observe that, in fact, BP=0 reduces CPU utilization considerably, there is enough juice to do other things like terrain mesh, AG, scenery objects which will explain the sudden increase in performance when using BP=0There are exactly three options that can be used inside [bufferPools] one is, of course PoolSize, the other two are UsePools (which can be 1 or 0) and RejectThreshold which defaults at 1 Million (no idea what it does)Some other hidden settings I played with: they go in the [Graphics] section:ALLOW_SHADER_20ALLOW_SHADER_30SHADER_CACHE_VERSIONSHADER_OPTIMIZESHADER_PARTIAL_PRECISIONSTALE_BUFFER_THRESHOLDHIGHMEMFIXDPUPBUFFERSIZEI have tested almost all of them, they really make no difference... but the only ones worth investigating are the last 3.

Some of that stuff rings a bell, Shader20/30 where for backward compatibility with older graphic cards as I recall (you can see the same command in the display.cfg for older cards).I wonder if its all still being used or is just left-overs in the dll. Back to using FPS limiter, not saying its impossible but are you sure your getting a real report on CPU usage, and that the limiter isn't interfearing? But maybe not, like you say its less with BP=0, but then thats why the system can't keep up with the terrain and we get blurries, so BP=XXXXX stutters maybe related to more thread collissions???Well not sure where this will all go, BP=0 and other stuff was explored more than two years ago and I dont recall near the same results as we are seeing now. If we could get find a better balance for those of us who like to run with the higher Image Q settings with many add-ons without the black screens and artifacts we'd be in business.Thinking about the above, I wounder how this test out on something other than Vista/Win7? Have you explored this?I dont have XP64SP2, only 32 but I may set it up because I dont recall this much of a problem on an old 4Ghz AMD quad and have attributed much of this to the character of the I7 IMC not being 1:1, but now I wounder if it was the OS.I'm glad you've kept going with this despite what I have said before, and so I must eat some humble pie and also say I'm glad the thread is still open.. :( If I can find the time gonna see what other tools are lying about the place to see what else is lurking here, dig into some of those other settings and also putting XP on for a spin or two.

Thinking about the above, I wounder how this test out on something other than Vista/Win7? Have you explored this?

I used BP=0 in Windows XP 32Bit a year ago with great results (there is a post here in Avsim about it)

I'm glad you've kept going with this despite what I have said before

Thats why its so important that other people CONFIRM what others have seen so far, I guess you and I are more interested in the 'Why' rather than the go fly and enjoy :) I appreciate your time, and I'm pretty sure others do as well, specially with constructive comments and opinion that 'help us' find some answers. at the end, we all benefit from it :)One more thing.. I can't confirm it, because I have not tested it extensively, but, there is one thing that you could test and see what results you get.. it 'seems' as if, the newer nVidia drivers work GREAT with VSync set to OFF or with the 'Use the 3D Application setting' maybe WDDM 1.1 and the 196.xx changed 'something' related to FSX and VSYNC because forcing (in the nVidia CPanel) VSync to ON its the ONLY WAY I seem to replicate black screens and get artifacts, Forcing vsync to ON practically KILLS the card! I even get snow dots! so, there is something weird going on with VSYNC here as well.Also, for some reason the 'Water' setting, when set to ANYTHING below High 2.0 does produce tearing (but not artifacts) I guess it could be collisions or the schduler object batching going out of sync because of some hidden obscure setting. In fact, it doesn't surprise me, because in the ESP docs I remember seen something about the ShadowMappingQuality filter changing to High or Low depending on the 'Global' scenery settings, what I would recommend (as a test scenario) to those able to 'replicate' artifacts and/or black screens is to:1) Disable vSync2) Set Water to HIGH 2.0 or MAX 2.03) Use de nVidia Control Panel (and not nHancer) to control Aniso/AA (remember that uninstalling nHancer is not enought to get rid of it, you need to DELETE the global profile it creates inside nHancer) and remember to do that BEFORE uninstalling nHancer.It would be wonderful if someone can 'confirm' that the above settings completely eliminate artifacts/tearing/black screens for them.I have set Aniso to (Application controlled) and also AA is set to application controlled in the nVidia control panel, Clamp and High Quality, and there is NO WAY I can replicate the artifacts, so, the ABOVE steps to try are in order of importance, I would try disabling vsync first and see if it makes a diiference to you, specially if running the 196.xx or 197.xx driversit doesn't surprise me why so many people have problems with BP=0!! *IF* (big if) Vsync, Aniso, AA, Dual/Clone monitors, ENBSeries mod all contribute to stress your video and BP=0 does in fact makes MORE draw calls to the driver, the command buffer will fill.. the card can't keep up with so many requests and you'll eventually get artifacts and/or black screens.. by the way, Black screens are nothing more than a driver crash what you are seeing is TDR (timeout detection & recovery) in action, and it only happens on Vista and Win7If you suffer from Dissapearing textures, corrupted textures and/or the top FSX menu dissapearing read about the shader tweak, it efectively solves that problem, however, it DOESN"T WORK with old ported FS2004 aircraft, in order to use this tweak you have to use NATIVE FSX AC.EDIT: Forgot to mention... if you have two monitors either in DualView or Clone mode try disabling it for testing and make sure only ONE monitor is used, altought minimal, there IS a performance penalty (even in clone mode)

I used BP=0 in Windows XP 32Bit a year ago with great results (there is a post here in Avsim about it)Thats why its so important that other people CONFIRM what others have seen so far, I guess you and I are more interested in the 'Why' rather than the go fly and enjoy :) I appreciate your time, and I'm pretty sure others do as well, specially with constructive comments and opinion that 'help us' find some answers. at the end, we all benefit from it :)One more thing.. I can't confirm it, because I have not tested it extensively, but, there is one thing that you could test and see what results you get.. it 'seems' as if, the newer nVidia drivers work GREAT with VSync set to OFF or with the 'Use the 3D Application setting' maybe WDDM 1.1 and the 196.xx changed 'something' related to FSX and VSYNC because forcing (in the nVidia CPanel) VSync to ON its the ONLY WAY I seem to replicate black screens and get artifacts, Forcing vsync to ON practically KILLS the card! I even get snow dots! so, there is something weird going on with VSYNC here as well.Also, for some reason the 'Water' setting, when set to ANYTHING below High 2.0 does produce tearing (but not artifacts) I guess it could be collisions or the schduler object batching going out of sync because of some hidden obscure setting. In fact, it doesn't surprise me, because in the ESP docs I remember seen something about the ShadowMappingQuality filter changing to High or Low depending on the 'Global' scenery settings, what I would recommend (as a test scenario) to those able to 'replicate' artifacts and/or black screens is to:1) Disable vSync2) Set Water to HIGH 2.0 or MAX 2.03) Use de nVidia Control Panel (and not nHancer) to control Aniso/AA (remember that uninstalling nHancer is not enought to get rid of it, you need to DELETE the global profile it creates inside nHancer) and remember to do that BEFORE uninstalling nHancer.It would be wonderful if someone can 'confirm' that the above settings completely eliminate artifacts/tearing/black screens for them.I have set Aniso to 16 and AA to 8xS in the nVidia control panel and can't replicate the artifacts, so, the ABOVE steps to try are in order of importance, I would try disabling vsync and see if it makes a diiference to you, specially if running the 196.xx or 197.xx drivers

Just have time for quick reply: NV drivers have been and still are broken with regard to FSX in game filtering control. Thats why you need a hack like Nhancer or RVtuner and also why your not getting the artifacts, the GPU isn't really being pushed anymore without the hack, because its not giving you 8xS and not much else. So using the NV CPL will get a lot of confirmations of less problems no doubt about it, because its not really giving us the selected filtering.

Just have time for quick reply: NV drivers have been and still are broken with regard to FSX in game filtering control. Thats why you need a hack like Nhancer or RVtuner and also why your not getting the artifacts, the GPU isn't really being pushed anymore without the hack, because its not giving you 8xS and not much else. So using the NV CPL will get a lot of confirmations of less problems no doubt about it, because its not really giving us the selected filtering.

I don't think I want to give up 8xS. Giving FSX 1-2MB of BUFFERPOOLS seems to do everything 0MB did but with no artifacts. Nice to have the 8xS I feel.Noel

So using the NV CPL will get a lot of confirmations of less problems no doubt about it, because its not really giving us the selected filtering.

True, and yes there is a HUGE difference between in game AA and nHancer, because I tested the one in nHancer and it looks beatiful. Well.. Fermi is around the corner, probably they will play along well with BP=0