Thanks to this thread I now understand the RejectThreshold tweek. I never tried it before as I never really understood it, but this thread clearly explains it. With the RejectThreshold=131072 added and no additional BufferPool setting added I am getting GREAT performance. The areas I am really seeing the benefits of it are at large airports with alot of AI around. I just flew the Level D 767 into Mega Airport London Heathrow with 80% AI in UT2 and it was the best I have seen it. Thanks to all involved in this thread.

What is the lowest you can go on the reject threshold? as I understand it the lower you go the closer you are to the BP=0 tweak right?

What is the lowest you can go on the reject threshold? as I understand it the lower you go the closer you are to the BP=0 tweak right?

Im not sure but 131072 is pretty low. The value is in bytes so 131072 converts to .125 mb.

I dont recall making any staments about "it" working or not, so your deffending a possition that you created, not me. The "it" (oh no, quotes) works just as you see (well if you had a faster system), push the IQ with complex add-ons on a state of the art system and we all know what will happen, hence all the neccassary caps on certain settings.You sure it wise to make assumptions about how someone else came to a conclusion when you dont have any info about where their knowledge comes from?:( Sorry again to the original poster, who really probably just needs to dial things back a bit, increase Poolsize back from zero to 100000 etc a bit at a time or use the reject rejectThreshold method as Ryan suggest. Cheers!

All your posts in this topic are wasteful and unnecessary, sometimes some people need to learn the phrase "quiet while the adults are talking"Trying to one up and shush Tabs after he gave valuable information was unbelievable.

Spike and Scatter pixel may occur when Usepools = 0Try this , it resolve problem for me- use dx10 mode- ground shadow ON- aircraft shadow on itself - decrease your CPU speedchoose only one just enough , will lose a few Frameratep.s. problem happen because graphic card can not handle cpu power , many workload to gpu . I wait for 485 but now it is only smoke. Ahhh 580 just a good idea.

Im not sure but 131072 is pretty low. The value is in bytes so 131072 converts to .125 mb.

In fact, that value is around the 'optimal' value. It should be between 96Kb and 256Kb to take FULL advantage of the hybrid approach. 131072 is 'exactly' 128KB, which is a good value if your card can handle it. If you experience problems, you can increase it in 64KB blocks. You can completely ignore the other two settings (UsePools and PoolSize) and only use RejectThreshold=131072 if you wish.The 'problem' with this setting is that is VERY system specific (like the UsePools=0) the good news is that RejectThreshold can be 'increased' slightly to improve stability. Now, if you increase it beyond 500KB there is absolutely no reason to use it, that means your system can't handle it and you better forget about tweaking using this setting or get a faster PC. The absolute minimun (for best performance) I recommend is 96KB.. 'Technically' what the setting does is set a 'limit' to the amount of 'data' that can go directly to your Video Card vs. the data that goes to the bufferpools. BufferPools are 'software managed' meaning they take CPU, thus reducing performance. But they also help 'store' data in memory so it can be used inmediatly (the vertex data) to render scenes.RejectThreshold=131072 tells FSX to send vertex data SMALLER than 128KB DIRECTLY to the video card ring buffer (very fast) and to send vertex data larger than 128KB to the bufferpools .. this is an hybrid approach that mixes the performance benefits of UsePools=0 with the stability of using BufferPools.I'm pretty sure all this have been covered plenty of times, and both Ryan and audiohavoc explained it very clearly.

Lazza:Your entire FS.cfg is irrelevant. What IS relevant are the following:1. You have a i7-920 @ 4Ghz. You have an overclock of over 1.4Ghz from the stock 2.66Ghz2. Your ram timings @ 7-7-7-20 are tight however, you don't mention what speed you are running your RAM AND what your DRAM:FSB ratio is3. You have a GTX 285 but is it overclocked and if so what are the specs?4. Tell me about your Case and CPU cooling systems.What you are seeing is a textbook case of artifacts...nothing new. I don't know what bufferpools does but I do know that whatever it may be doing its causing either your CPU, GPU or memory to suffer and the result is, you guessed it, artifacts.I suggest you download prime95 and memtest (if you haven't already since you are an overclocker) Take prime95 and run a stress test over 8 - 24 hours. If it passes with no errors then your CPU and Memory are just fine. If it fails run memtest. If Prime95 finds no errors then the issue with with your video card...either its overclocked and you need to pull it back or your have bad video memory...OR...the combination of your CPU overclock, memory speeds AND GPU are not working.Let me know what you find out.

In fact, that value is around the 'optimal' value. It should be between 96Kb and 256Kb to take FULL advantage of the hybrid approach. 131072 is 'exactly' 128KB, which is a good value if your card can handle it. If you experience problems, you can increase it in 64KB blocks. You can completely ignore the other two settings (UsePools and PoolSize) and only use RejectThreshold=131072 if you wish.The 'problem' with this setting is that is VERY system specific (like the UsePools=0) the good news is that RejectThreshold can be 'increased' slightly to improve stability. Now, if you increase it beyond 500KB there is absolutely no reason to use it, that means your system can't handle it and you better forget about tweaking using this setting or get a faster PC. The absolute minimun (for best performance) I recommend is 96KB.. 'Technically' what the setting does is set a 'limit' to the amount of 'data' that can go directly to your Video Card vs. the data that goes to the bufferpools. BufferPools are 'software managed' meaning they take CPU, thus reducing performance. But they also help 'store' data in memory so it can be used inmediatly (the vertex data) to render scenes.RejectThreshold=131072 tells FSX to send vertex data SMALLER than 128KB DIRECTLY to the video card ring buffer (very fast) and to send vertex data larger than 128KB to the bufferpools .. this is an hybrid approach that mixes the performance benefits of UsePools=0 with the stability of using BufferPools.I'm pretty sure all this have been covered plenty of times, and both Ryan and audiohavoc explained it very clearly.

If I understand this correctly without rejectthreshold in place the whole bufferpool block amount would be sent to the card all at one time (ie 4mb default or whatever BP value is in place) ?

If I understand this correctly without rejectthreshold in place the whole bufferpool block amount would be sent to the card all at one time (ie 4mb default or whatever BP value is in place) ?

No, without rejectthreshold EVERY 'data' related to graphics and needed for rendering (vertex data) will go FIRST to a pool that is managed by FSX and THEN released to the Video Card.rejectthreshold will simply set a threshold, meaning that all data GREATER than the value you set in the threshold WILL be send to the pool (software managed), but the 'smaller' data will be processed instantly (directly to the card)

No, without rejectthreshold EVERY 'data' related to graphics and needed for rendering (vertex data) will go FIRST to a pool that is managed by FSX and THEN released to the Video Card.rejectthreshold will simply set a threshold, meaning that all data GREATER than the value you set in the threshold WILL be send to the pool (software managed), but the 'smaller' data will be processed instantly (directly to the card)

Got it! Thanks!

I've actually appreciated this thread, though it would be nice to leave things at a technical level, but I understand that doesn't always happen. My understanding of poolsize is that it doesn't limit the total amount of buffer space, just the fragmentation. i.e., you could have 1 single 4Mb buffer or 4 - 1Mb buffers. ACES determined that having too many buffers ate up CPU so they increased the size of each buffer to 4 Mb. I guess that can waste some memory but the trade was considered acceptable. Now using the reject threshold will reduce the amount of data being managed in the buffer, so obviously the total amount of buffer space required goes down (if that's correct). That means a smaller poolsize might work because the demand is lower. In the limit as reject threshold goes to zero it results in no buffers ever being needed so is equivalent to usepools 0.I would propose for testing to leave out poolsize initially and just play with the reject threshold, starting large and reducing until you get artifacts, then backing off. Once reject threshold is OK you can experiment with adding a poolsize and reduce that from 4Mb and see if it makes any difference. My guess is that poolsize has less impact than reject threshold, but that's based on nothing but supposition.This is all based on trying to interpret what you guys have put out, so could be 100% wrong if I mistake what you are telling us.scott s..

Thanks Ryan, Shane and ******* for the replies.I agree with Scott, less supposition and more real testing is what opens the eyes.My own drawn-out long testing of various Threshold values that destabilize the sim when IQ is pushed with complex add-ons resulted in FSX acting nearly the same way that smaller Poolsize values do, thats all that was intended, and based on other information I still feel the same, however some of the explinations are interesting.So I still stand behind what I posted, though I try to have an open mind so who knows that could change if these explinations are accurate it may be different with values I haven't tried.I really do appreciate the efforts in trying to explain by ******* as well as Ryan sharing the info makes for great discussion compared to just repeating, cut and paste what ever on info that has been available to all does nothing to add to the coversation.-->Just because there is disagreement does not mean its a personal childish attack and I really have no time for that kind of BS, there was no attack on Ryan or anyone else that kept this a objective technical discussion (though looking back I understand exactly why he took offense and appoligize).Take Luck

I agree with Scott, less supposition and more real testing is what opens the eyes.

Agree.. the info Ryan provided comes directly from the MS team itself :) no suppositions, that I can assure you.In any case, what is important is what WORKS for you and not what people tell you, besides, no single tweak willhave the same effect on the variety of hardware out there, so its perfectly normal to find people posting infoabout X or Y tweak not working for them, nothing wrong with that.happy landings,

I had a feeling that this BP=0 topic will come up again. :( As hardware and drivers change, so will our cfg tweaks.*******, I hope you're doing well.

In fact, that value is around the 'optimal' value. It should be between 96Kb and 256Kb to take FULL advantage of the hybrid approach. 131072 is 'exactly' 128KB, which is a good value if your card can handle it. If you experience problems, you can increase it in 64KB blocks. You can completely ignore the other two settings (UsePools and PoolSize) and only use RejectThreshold=131072 if you wish.The 'problem' with this setting is that is VERY system specific (like the UsePools=0) the good news is that RejectThreshold can be 'increased' slightly to improve stability. Now, if you increase it beyond 500KB there is absolutely no reason to use it, that means your system can't handle it and you better forget about tweaking using this setting or get a faster PC. The absolute minimun (for best performance) I recommend is 96KB.. 'Technically' what the setting does is set a 'limit' to the amount of 'data' that can go directly to your Video Card vs. the data that goes to the bufferpools. BufferPools are 'software managed' meaning they take CPU, thus reducing performance. But they also help 'store' data in memory so it can be used inmediatly (the vertex data) to render scenes.RejectThreshold=131072 tells FSX to send vertex data SMALLER than 128KB DIRECTLY to the video card ring buffer (very fast) and to send vertex data larger than 128KB to the bufferpools .. this is an hybrid approach that mixes the performance benefits of UsePools=0 with the stability of using BufferPools.I'm pretty sure all this have been covered plenty of times, and both Ryan and audiohavoc explained it very clearly.

Thanks for everyones comments so far, It is a little difficlut to answer everyones questions sorry, and at the moment I am having most sucess with Reject Threshold so I have one more question relating to reject thresholdIn Bojotespost above he mentions ." The absolute minimun (for best performance) I recommend is 96KB.. "I understand that 128kb = RejectThreshold=131072 which is the setting I am currently using. Bojote says 96kb may provide better performance which I want to try out.But I am struggling to work out what the 96kb figure would be for use pools...ie if 128kb = RejectThreshold=131072 then 96kb = RejectThreshold=???Any suggestions?Laz