Seems like a fresh FSX install may have to be the standard. For FSX.CFG tweaks, we'll just make a straight forward use Bojote's online tweak tool - no modifications otherwise. Any objections?

I can not see any way around it, not if we want accuracy. As far as FSX.CFG modifications, we need to agree on adoption of Bojote's written recommendations 100% for the particular machine. Software to be a freshly installed FSX plus Acceleration only, without a single addition or deletion with exactly specified settings. There should be a standard format protocol for any posting of results. Anything else posted in the official thread that does not conform 100% should be deleted. All this without exception, strictly enforced. (David would have to be willing to be a strong armed policeman!)Kind regards,

I did some testing yesterday and today, and found out that, no matter how much you disable, you can never get your FSX to "vanilla" standard, as like it is when you ONLY install FSX, without anything else.So I decided to go a bit a try-out route.I backed windows and fsx partitions with an Acronis True Image and did following:Renamed main FSX folder, FSX folder in my User Profile and FSX folder in ProgramData folder. Renamed all 3 to FSX2.Then started a FSX installation from the DVD - it asked me to repair or uninstall. Went for repair. It installed everything from the two DVDs.Then SP1. SP2 asked to be uninstalled. So I did, and reinstalled SP1 again, and then SP2 successfully.That's it. I had a vanilla situation. Including drivers and driver profiles.Did my testing.When I was done, I deleted newly created folders (all 3), and renamed old ones back.Created new driver profiles - I could have probably saved it somehow, but this was a lesser task at hand.I think this is a way every user could undergo, without really compromising his currently installed FSX installation and STILL provide very accurate results.

For those who havent tried it, I ran some tests at a 4.7Ghz (much healthier for CPU) to compare with 5Ghz.5Ghz:2011-02-27 18:19:20 - fsxFrames: 11550 - Time: 300000ms - Avg: 38.500 - Min: 15 - Max: 472011-02-27 18:25:36 - fsxFrames: 11401 - Time: 300000ms - Avg: 38.003 - Min: 25 - Max: 482011-02-27 18:33:58 - fsxFrames: 11290 - Time: 300000ms - Avg: 37.633 - Min: 25 - Max: 48 4.7Ghz2011-02-28 09:18:59 - fsxFrames: 11198 - Time: 300000ms - Avg: 37.327 - Min: 20 - Max: 462011-02-28 09:25:04 - fsxFrames: 10996 - Time: 300000ms - Avg: 36.653 - Min: 24 - Max: 462011-02-28 09:34:20 - fsxFrames: 11174 - Time: 300000ms - Avg: 37.247 - Min: 26 - Max: 45I dont think the extra stress on the CPU (1.365V versus 1.45V) is worth the extra frame!Another observation is that the first run of the test after starting FSX is much slower than the following tests (typically 10 FPS average) so I have to run the test four times and ignore the first result. In fact I think the poor perfomance on the first flight occurs on all flights and not just the test.

Thanks for the figures Howard.No need to go to 5 ghz then. I keep mine 4.7-4.8 and find it runs perfect for me. The step from a 920 at 3.8 was a big one indeed for me. Could not be happier:)

All those that still have [email protected] or similar! Please run the FSXMark11 test so that we all can get a clear idea how much gain we would get by going to to the 2600K - it might be almost twice the FSX performance! we are seeing about 40 fps for the 2600K (or more) and alainneedle1 is seeing over 50 fps with his superduper 980X. I think we will see about 25 fps with the older generation.

All those that still have [email protected] or similar! Please run the FSXMark11 test so that we all can get a clear idea how much gain we would get by going to to the 2600K - it might be almost twice the FSX performance! we are seeing about 40 fps for the 2600K (or more) and alainneedle1 is seeing over 50 fps with his superduper 980X. I think we will see about 25 fps with the older generation.

Frames, Time (ms), Min, Max, Avg8635, 300000, 18, 45, 28.783

Hi,I hope that my RMA'd g.skill SSD will be returned to me this week in perfect working order. I think they shipped me a new one, but not sure yet. When it arrives I will do a fresh FSX install and benchmark test it accordingly.Kind regards,

Just for fun and conversation, I charted some expectations of the general FSXMark11 results to provide a quick perspective of how far we have come with the last few generations of CPUs. There has been a significant jump in performance, mostly due to CPU speed, but I think to other elements - principally memory speeds - and also disk performance, since there is a dramatic need for scenery to be paged in as you fly along. These plotted estimated results are meant to be fuzzy areas - every PC is just simply different - mobo settings, memory latency, etc. It is interesting that this Avsim group has decided to use an fresh FSX install to try to get repeatable results . The FSX process when executing is full of calculation and graphics pipelines, threads, and fibers, and they all must align to get the best performance. I added a couple of "tuned" points where I still feel that some of the "outlier" results show that a few of us have been extra fortunate to find the tipping point to another 20% or so performance that is felt a little in fps but also in real fluidity - nearly total lack of stutters. It's almost like shifting gears in a sportscar, third gear has some stutters, fourth gear is quite good, but some of the PCs find the overdrive position. This special tuning can come from a multitude of sources - memory speed and timings, fsx.cfg tweaks, reductions in scenery setting, etc. The really good news is that with the advent of the 980X (triple channel memory is excellent) and the new 2600K, those with these systems are finally getting immersive performance that can really bring flight simulation to a place of satisfaction. Please keep providing actual data so we can replot this type of chart.

Just got finished with some testing after a fresh FSX install following procedures from the soon to be coming updated guide. It was not until I did a fresh install that I realized REX had not reloaded all the default FSX textures. That should reinforce just why a fresh FSX install will now be a requirement for the benchmark. I also noticed that regardless of which graphics card you select in the dropdown, Bojote's online tweak tool does not add the line "UsePools=#" or "PoolSize=#" into the fsx.cfg. I performed three benchmarks just for educational purposes. The first with nothing but Bojote's tweak tool applied, the second with the addition of UsePools=0, and the third with the addition of UsePools=0 and PoolSize=8388608. Don't let me confuse you, however. The new guide will no longer allow fsx.cfg tweaks other than what the online tweak tool does automatically... unless you guys convince me otherwise. This is with nvidia inspector disabled and nvidia control panel restored to defaults, by the way. Other than what you see here, I did no tinkering with the cfg or anything else. These benchmarks were run within an hour of installing FSX.Looking at the results, I considered making a provision in the guide to allow the addition of "UsePools=0" for graphics cards equal to or greater than the GTX 460 1GB. I see some minor artifacting with my GTS250 and UsePools=0, hence the requirement of GTX460 or greater... but then I figured, why add any more confusion to the benchmark? What do you guys think?

I agree on keeping it simple Corey. *******' tool doesn't add the UsePool=0 line, very true, and I think this comes from the convo we had about this. Well, my bad, UP=0 was never added but neither removed, because it was already there: I had my bufferpols previously set to UP=0 and the tool simply added the rejectThreshold.Actually I can see in your results, like in mine, that UP=0 overrides any other BP setting, at least performance wise (once the UP=0 line is there, the rest of the section seems to be ignored)

Hi Corey and Jazz,The Bufferpools tweaks seems to be quite inconsistent between machines. I would say let the user determine what best fits his/her Bufferpools setting or not to use it at all, but to report exactly what the settings were for the test. Afterall, the bufferpools is in default (4,000,000) whether it is modified in the cfg file or not. If it works better one way vs another, why not let the tester use what works best? We want to see how well any given hardware can perform with a fresh installation right? What do you think about having a standard reporting format, so that the readers do not have to interpret the interpretations? I like the way you just reported yours Corey. My RMAed SSD still isn't back, although G.skill say they shipped it last week. Whenever it gets here, I will reinstall SFX and run a number of tests with bufferpools modified and unmodified fsx.cfg files. Maybe that information will help us, or make matters that much more foggier. Lots of options I guess, so do what you think best.Sorry if I repeated myself as I tend to do that from time to time! Sorry if I repeated myself as I tend to do that from time to time! Kind regards,

Stephen, if i may, if you can change your memory for faster one like the one in the link you will see an improvement with your 2600K at 5.0GHz... Here's the new one for SB, 2300MHz at cas 7 "POW" http://www.techpowerup.com/141392/G.Skill-Exhibits-New-Memory-Products-at-CeBIT.html

Just for fun and conversation, I charted some expectations of the general FSXMark11 results to provide a quick perspective of how far we have come with the last few generations of CPUs.

Great post... thanks for the chart! In the original FSMark we all plotted this stuff on graphs. Numbers are boring, let's see some graphs yay!

Nice work guys...

Trust me, graphs are the eventual goal. Right now we are more concerned with getting this benchmark nailed down. I will have the guide finished very soon. In the mean time, check this out!