Well, I'm cautious about accepting the assertion the Intel platform is buss limited. The traffic appears to free to zoom along at whatever speed its capable. The racetrack appears NoT to be impeding their progress. It is certainly a valid theoretical premise. Someday this will CerTainly be a factor, but not yet. Remember, the AMD platform cleaned Intel's clock in memory bandwidth for years. But that wasn't what allowed AMD to win that performance war. Intel's new Core2 platform only adopted AMD's CPU architecture but Did Not adopt their high bandwidth memory controller architecture. This was because Intel recognized the high bandwidth memory buss would not be needed for quite a while. Well into the DDR3 (or more likely DDR4) speed ranges. Now the Intel platform is cleaning AMD's clock with an almost identical (but faster clocked) CPU, but with their old FSB And memory controller architecture. Intel's platform is not (yet) Buss limited. Further, the test houses occasionally run head to heads with various FSB and memory speeds. Increasing FSB or memory speeds make no difference. Finally, Intel's Nehalem Platform will complete Intel's transition to AMD's archecture. The FSB will go away and Intel will include a "close-in" memory controller onboard the CPU die. But even this will Not Be Needed, for a while. This is just like any other buss evolution. We have No use for even SATA I. PCIE-v2? I guess 2010, there Might be a Vcard that will need it. or DDRFFFRist sut like I'm seeing a subtitle difference in the visuals, and I'm curious about that. However it appears ram speed or DDR version is only a marginal part of it (if any at all). Getting IrfanView's conversion function tweaked out right is a factor too. For instance, that AA fuzz in my initial post's screenies is not happening IRL. Here's another set: http://forums.avsim.net/user_files/187160.jpghttp://forums.avsim.net/user_files/187161.jpghttp://forums.avsim.net/user_files/187162.jpghttp://forums.avsim.net/user_files/187163.jpgSomehow, I'm still loosing it in the translation. However at the end of my propped-up feet, I'm laying the original screenies over my live flight on a 42" monitor and there's not much difference. A bit (I guess), but my live flight is nothing like the "before" shots. I took it down to my 19" at 1280x1024 and I was sooo impressed about how well that darn 42" holds up. It might have been a bit tighter, but it was arguable. I dunno. What do ya'll see. 'Peers to me there's something other than ram involved 'tween the 'fore and afters.

Your images do not display the UTX LUX Urban Layer.. the Port of Seattle is full of vegitation when it should look like this if UTX and the scenery library are correcthttp://forums.avsim.net/user_files/187164.jpgYou either have a 3rd party landclass in the wrong place are you have disabled the UTX LUX Urban Layer.you are also missing quite a bit of autogen, a very -LARGE- amount of autogen (distance is very poor) is missing for a level 5 scenery slider setting. http://forums.avsim.net/user_files/187165.jpghttp://forums.avsim.net/user_files/187077.jpgAs a matter of fact, all your shots are missing autogen for a level 5 slider setting.Either you do not have it on 100%, or, as I have said many times, FSX is dynamic and will skip renders to deliver its priority, therefore you are no where near my performance. Perhaps you do not realize how much the sim is not rendering in your shots even though the sliders may be pushed up?Perhaps you have the autogen restriction line tweak edits in the config file? I have none.Your UTX roads show low LOD (jagged edges) and that is not compression, that is priority to the vector roads not being applied until you are on top of it.Also, my frame rate over Seattle is 10 frames higher than yoursAll 3rd party landclass products when used are to be layered with UTX as specified:UTX - URBAN<------ 3rd Party LandclassUTX- VEGETATIONThe benchmark flight which was used to produce the presentation travels the jet at a set speed of 250kts for a reason.

NoelThere is more CPU and FSB ability than the memory you have will provide for unless you can run the CAS timing in conjunction with the memory speed and the FSB/divider (STRAP) based on this chart...http://forums.avsim.net/user_files/187166.jpgYour targets are the GREEN The LOWER tRD value represents the fastest and most efficient(first line in the chart) At 1:1 CPU and DDR2 800 (400MHz memory speed), you need to be running CAS3 with the proper sub-timing @ 400FSB and the CPU as high as it will go in multiplier with that, stable, and, that is still not as fast as 5:4 DDR2 1000 (500MHz memory speed) 400FSB @ CAS4In the example of DDR2 1000, that is not set in stone.. the higher you go past DDR2 1000 with the memory running CAS4 on the 5:4 divider, the faster the system assuming you keep tRAS in line and do not need to loosen other timings more than +1 to maintain it. 1T CMD will only improve that result further. I always strive for 1T and will usually sacrifice some memory speed to maintain 1TSame with DDR2 800 @ 1:1 and CAS3, the higher you push the memory past 400MHz the more productive you will be. Even small amounts such as 25-50MHz can have a profound effect if CAS3 is maintainedthe main problem is the sec of the memory, what are its abilities, and, can it handle a Vdimm (voltage increase) if needed to hold stability. Good clocking memory is not cheap for a reason. If you purchase low latency, high speed memory you do not have to work so hard to hit the numbers, and, usually without any voltage increase that would cause any issues.

NoelScroll down the thread and look at the compares of imagesWith FSX its not just about frames... of course smooth flight is the objective, however, the more you have in resources, the more you see on the screen, and, the sharper it renders.. with FSX that includes having sliders maxed out.Just becuase a slider is at 100% does not mean you are seeing everything. Thats my point. If you have the resources, you get the show.. if you dont, you may get smooth flight but it will also come with a cost in loss somewhere. In the case of the images I compared futher down and assumning the sliders are at 100% with no tweak in the config file for autogen, that cost is scenery to get frames.My expereince is FSX will remove autogen and scenery for the priority of flight and do its best to deliver what it is being asked but if the resources are not there, it will degrade your visual experience to try and deliver the needed frames, period.FS9 was the same way.That is why I was trying to explain in my presentation it is very difficult to call a frame number a performance value unless you know what is missing on the screen, and/or how fast/slow it was rendered. Those elements are performance as well.

Great info as usual Nick, but I have a couple of questions. One even though, you have road traffic, set to 12% on your FS sliders, I don't see any in your screenshots. Are you sure you have road traffic activated in UTUSAX? While I see the scenery load lag you note with DDR2 memory, (I use PC2-8500 Corsair Dominator memory @1066Mhz)I see similar results to what you have posted, I don't Overclock either my system or GPU, yet, I am getting better frame performance than your system (DDR2 settings) even though yours is OC'd 20% over what my system runs at 3Ghz. Here are my bechmark results (FRAPS)My normal settings - Which are the same as you set, except I have no Lens Flare, Clouds set to 100Miles, and Traffic set to 100% airline traffic, but no GA or Ground or ship trafficInside VCFrames, Time (ms), Min, Max, Avg 7445, 300000, 17, 32, 24.817ExternalFrames, Time (ms), Min, Max, Avg 8258, 300000, 13, 41, 27.527With your test settings Inside VCFrames, Time (ms), Min, Max, Avg 6115, 300000, 13, 27, 20.383ExternalFrames, Time (ms), Min, Max, Avg 6589, 300000, 14, 30, 21.963Only variable I could see if traffic. I use WOAI with all available US airlines, as well as some user installed, including all regionals. Most major European and Asian airlines and their regionals. No GA though, still that should be significantly more than Default Traffic. KSEA is loaded. What traffic files do you use?

I did not alter anything. The settings are exactly as they were during the flights. I do see traffic but during the benchmark tests flights I did also notice car traffic volume was not showing as much as it usually does and as it shows here, although still somewhat difficult to see due to compression of the image...http://forums.avsim.net/user_files/187164.jpgI do know the sim runs things in priority and the areas where I grabbed those shots during the benchmark are very high impact. It would not be unusual to see an automatic reduction of a lower priority element such as car traffic in that case. Case in point, if you scroll up 3 posts or so you will find a compare between one of my benchmark images and one from someone who says the sliders are 100%. As you can see in that compare, and all of the shots, there is a massive amount of autogen missing for a level 5 autogen slider setting in Sam's shots compared to mine. His slider being 100% does not mean the sim will render it all, something I have seen many times in FSX and why its very difficult to compare using sliders and FPS numbers... a visual is required in order to compare correctly.As for AL traffic, what you are running does not come close to the same load on the system as default. Default AI, beside the models being heavy compared to WoAI, has gate and jetway code data which will force very high CPU use in calculation of all the gates at airports in a 50-120MN radius of the aircraft when the scenery complexity slider is VERY DENSE and above. That load is magnified by the AI slider value, greatly. The replacement AI removes all that.I also placed a note in my presentation in section one under CONSIDERATIONS that removal of the default AI aircraft and flight plans was in fact causing an offset with my preliminary tests and was throwing off accurate results with the benchmark, which is why default AI had to be restored.Again, and as I said above to Noel, if the settings and the configuration file edits are not the same as the ones I applied, and, the user does not use the default AI, the results will be skewed if people are trying to compare what I posted to themselves. If those settings, config file edits (including TEXTURE_BANDWIDTH_MULTIPLER=80) and addons are not exactly as I posted with default AI enabled, the result will not be the same or anywhere near a compare, and, images are required in order to compare even if all test settings are matched to mine.The only settings that may be altered assuming all others are the same are WIDESCREEN and the monitor resolution to accommodate the system being used, and, those changes will also play into the result.AND driver settings MUST be the same.Its very difficult to simply post numbers and compare to my results and completely impossible to to so with default AI removed and the config settings different, especially TBM.

Thanks Nick, I missed where you mentioned you were using default AI with the jetways, When I set them up with my WOAI traffic awhile back it brought the system to it's knees, and I had to dump it right away. You got me curious though, (Which is easy to do!) So I removed my WOAI traffic files, and restored my default, and got even better results, plus the texture load seemed better. (Not as good as your DDR3 tests though!) Now I'm really puzzled why there is a difference, considering the difference in system specs.VCFrames, Time (ms), Min, Max, Avg 6436, 300000, 12, 31, 21.453 ExternalFrames, Time (ms), Min, Max, Avg 6933, 300000, 12, 36, 23.110

You must look at images to compare for missing or unclear elements. The FPS is not the target observation, and, the results are subjective to the user and what they see or acceptyou are also overclocked to 1066 according to your specs.. it is possible you are seeing the result of that if the system happens to be running the memory correctly or somewhat close in that respect.Different monitor resolutions may have an impact.I have no idea what is in your config file either, or, what may or may not be tritely missing because of toying with the AI back and forth. Assuming you saved the current config file, allowed FSX to write a new one and set up the sliders exactly as I show AND make sure REALISM is set to HARD or 100%, and then make the exact config edits I posted.. the resulting flight and images from it should reveal what you seek. At the same time that all assumes the elements of the UTX add-on are enabled the same as mine such as intersection night lighting, etc. And it assumes there is nothing stuffed in the FSX install. And please note I do have Acceleration installed.

I do run with FSX SP1 in Windows XP SP2, so maybe that is the difference, but I thought SP2/Acceleration improved performance over SP1? I do run Acceleration in Vista, but that runs slower. As for my memory, As I said it is PC2-8500 memory, while it does run standard at 800Mhz, it is fully rated at 1066Mhz with EPP and command rate at 2T. (5-5-5-15-2T). I had just recently rebuilt my FSX.CFG file, the settings I use are .23 fiber frame, TBM of 100, Widescreen, MipBias=6, Autogen 4500 trees, 3000 buildings, and of course AffinityMask=3. My graphics settings are 16X AA Coverage sample, AA Transparency set to Super, 16x AF, VSYNC Off, Texture Quality High Quality, Trilinear on both in game and driver.

Then you are completely different than me and there is no true compare that can be made with a number. Baring the SP1/SP2 differences, you would have to dump the config and set up the same if you are trying to compare. You also must provide screenshots for compare.There is no other way to even begin to say one is faster/better/smoother than another. The SP1/SP2 changes alone will throw a compare completely off/And comparing is not what this is about.. its about finding the best you can personally do with the memory and CPU you are running.I know the cars are hard to see especially when you compress a high res image with a large amount of data in it down to a smaller res from altitude.. even with the low altitude showing the 1m textures being stretched, since the roads are 7cm and vector drawn, you can see the little buggers running here...http://forums.avsim.net/user_files/187180.jpghttp://forums.avsim.net/user_files/187181.jpghttp://forums.avsim.net/user_files/187182.jpg

Here are the charts with respect what I am running on that P35 systemhttp://forums.avsim.net/user_files/187183.jpgThis is base for which to find the lower tRD relationship to FSB/STRAP (divider) CAS/MEM_SPEED. I am running tRD7 and not 6 due to a P35 issue and need to bump memory speed up just a bit more to force the threshold.That chart above just gets the ball rolling. The next chart shows typical performance of memory and what/where I should be running to get the most out of it at default limitsThese charts are crude since I do not have time to plot things exact but do demonstrate the point. These charts reflect the P35 and the 333 STRAP I am using and vary based on the formula/setting being used (ie; A 266 STRAP @ 400FSB will reflect a different tRD scale at DDR2 1100).http://forums.avsim.net/user_files/187194.jpgThis next chart shows the same, except it removes the typical manufacture default CAS (GREEN) areas and replaces them with where the user wants to be to maximize the system IF the memory/system will allow for ithttp://forums.avsim.net/user_files/187186.jpgAs you can see that old DDR3 memory is still falling quite short in being optimal but the tRD formula allows me to make use of what I do have. I am right at the threshold for tRD 6 and because the P35 is not quite in sync with the math its holding me at tRD7 unless I boost the memory speed up a bit more. Unfortunately, I can't with that older DDR3 memory and remain stable.. it is topped out. I have seen tRD6 with it, at just under DDR3 1475.There is more to this and that comes with understanding the sub

I guess your right, with the 2 different versions you really can't make a fair comparison. Here's just a sample of what I see on my system. Just noticed, while Fraps reports a higher avg FPS, your individual screenshots show higher. Which does make sense. http://forums.avsim.net/user_files/187188.jpghttp://forums.avsim.net/user_files/187189.jpghttp://forums.avsim.net/user_files/187190.jpghttp://forums.avsim.net/user_files/187191.jpghttp://forums.avsim.net/user_files/187192.jpg

Thanks for posting that TomIt

I was going to try and OC my system, but of course, as you are well aware, Dell doesn't trust it's customers, so the only way I could do it is with NTune. Which isn't the best way to do it, and some say avoid it like the plaque!:-) Since I'm getting decent performance with 100% traffic, even at locations like NYC. I don't really need to do it anyway. I do try to keep my system well tuned and defragged, although I use Ultimate defrag not O&O like you suggest. (Their new version that is coming out looks real interesting!) As such I don't disable Last Access Update. While I may get a little boost with it, I don't think it's worth the risk of some errand app, trying to use it and ends up mucking up the system! If my performance was worse, I might have tried it.

Oh, the list?Well, all I can say is I would not fly or use a system without that one. If the Dell install disk is SP2 and was produced after December of 2004, you are most likely fine and will not have the MS bug in ntfs.sysIt all comes down to what works for you! :)the 100% trafffic is WoAI, we know that! LOLAI is the #1 killer running the scenery complexity above DENSE with default AI installed.