I will test any resolution (max 1920x1080x32), AA, & AF that anyone else tests to compare, and we could also try one run with minimal resolution (1024x768x32) no AA and no AF to isolate the CPU.I can also easily set up another flight with no clouds, but all the tests I run so far where quite similar even with the little clouds that appear scattered in fair weather.With no traffic you mean 0% traffic or just default traffic? I already disabled mytraffic X (it just takes 1 minutes to disable MyTraffic / UT2 and restore default traffic) so if someone doesn't want to disable their traffic addons, well, ok, but I'd still rather have some results with addons than noneJust my opinion

Results with different addons are nonsense. There must be a vanilla FSX with specified addons and nothing else.Let's agree to two tests, I'll adapt the spreadsheet:1280x1024 AA/AF in FSXand1024x768 no AA/AF in FSXNO use of nhancer or inspector. Best simply to restore profile in nvidia control panel.And no traffic addon. Best would be to disable traffic (aircraft) in FSX completely.Am I forgetting anything?

Fair enough. No, you're not forgetting anything that I'm aware.Maybe we could add a third test with no traffic & 1280x1024 AA/AF in FSX then, so those who are not willing to disable their traffic addons can still compare some of their results to others'I'll restore my nVidia CP profile now, and I'm ready to run the tests if you guys agree on using my flight.

Results with different addons are nonsense. There must be a vanilla FSX with specified addons and nothing else.Let's agree to two tests, I'll adapt the spreadsheet:1280x1024 AA/AF in FSXand1024x768 no AA/AF in FSXNO use of nhancer or inspector. Best simply to restore profile in nvidia control panel.And no traffic addon. Best would be to disable traffic (aircraft) in FSX completely.Am I forgetting anything?

Let's do this internally here, before we go make a big post about it in FSX forums - where this should be actually. But I think we have to make a general post to invite everyone to take part in this.

Fair enough. No, you're not forgetting anything that I'm aware.Maybe we could add a third test with no traffic & 1280x1024 AA/AF in FSX then, so those who are not willing to disable their traffic addons can still compare some of their results to others'I'll restore my nVidia CP profile now, and I'm ready to run the tests if you guys agree on using my flight.

Too many different tests, we risk people not wanting to make many tests. 1 or 2 are IMO enough. But you probably ment "with" traffic. With traffic differences are always way too high.I suggest you make a new test file - and my suggestion would be to set autogen and scenery a bit lower, maybe Dense for AG, and Very Dense for Scenery? I think those or similar settings are what most are using... we need real world tests and not some theoretical "all sliders right" tests.And turn off traffic. Leave rest as it is?

Well, I fear if we lower the sliders someone will come saying that they don't care for some dumbed down sim settings haha, but actually I use normal AG (like my scenery maxed out though)There's already a compromise made with water at Low 2.x, light bloom off, GA traffic 0%, and no ground vehicles.Personally I would not turn off traffic, that's one of the most demanding things in FSX as we all know and this test was ment to test FSX on a strainAgain, that's just me and I'm sure we can all agree on the settings here... although so far it looks like no one else is interested :Sigh:I was thinking about posting in the PMDG section and see if we can get more people to chip in

OK, let's then leave AG and Scenery on max, let's see how it really goes. I made a preliminary test, and I got something around 24fps, and my machine is (still) right there in the tops, so I don't see how people could really complain. But let's go with it.You can leave ground vehicles though, just no aircraft.Also put in the readme the thing about resetting the Nvidia control panel, about recreating fsx.cfg, those are really important things.Tell me when the files are ready.It is indeed true what you say though, that traffic eats most, BUT - people have different traffic addons, many have them customized, if one would have 10 aircraft more than the other, it would already give a wrong benchmark, and there simply is NO way to test traffic. ONLY if you could, and you can't, make sure everyone has the same product and same setting in it, and really same amount of AI, then you could be sure that the result is not biased. Let's just for now go without it... the point of the benchmark is to see if similar computer makes similar performance, what GPU differences make, and what CPU is the best, and how well your FSX system is configured.I'm sure when we post it in FSX (or PMDG section for that matter, or both), we shall get some results.

Ok, I'll edit the readme to include all the rules, it was quite incomplete but I'm so laid back.. heheI actually run my tests already and included some 0% traffic ones. The ones with traffic as per included settings file were on default traffic, so I will make it clear that you need to disable MyTraffic or UT2 b4 running the benchmark (not too big of a deal anyway IMHO)Would you change the flying pattern too? maybe a U turn to head back to the airport after flying past it?1920x1080 AA/AF in FSX / Traffic ON as per settings file (100% Airline, 0% General Aviation)Frames, Time (ms), Min, Max, Avg 7533, 300000, 13, 34, 25.1101280x1024 AA/AF in FSX / Traffic ON as per settings file (100% Airline, 0% General Aviation)Frames, Time (ms), Min, Max, Avg 7561, 300000, 14, 34, 25.2031024x768 NO AA/AF in FSX / Traffic ON as per settings file (100% Airline, 0% General Aviation)Frames, Time (ms), Min, Max, Avg 7550, 300000, 14, 34, 25.167NO TRAFFIC:1920x1080 AA/AF in FSX / Traffic 0%)Frames, Time (ms), Min, Max, Avg 8594, 300000, 19, 37, 28.6471280x1024 AA/AF in FSX / Traffic 0%)Frames, Time (ms), Min, Max, Avg 8555, 300000, 20, 37, 28.5171024x768 NO AA/AF in FSX / Traffic 0%)Frames, Time (ms), Min, Max, Avg 8744, 300000, 20, 38, 29.147

Be a good guy and put the result into the spreadsheet :-)

Done, sorry about that. How about this? : Would you change the flying pattern too? maybe a U turn to go head back to the airport after flying past it?And my results show very little difference when changing resolution and AA/AF (no surprises really, it's the CPU bottleneck in all it's might)

Ok, I edited the instructions and uploaded the files here: Benchmark FilesAlso included the instructions in the spredsheet:Spreadsheet

OK, looks very nice! But, why don't we make this test a bit more challenging? All the aircraft does is fly a straight line... maybe do an 8 over the Heathrow? Would stress a system quite a bit more than straight, don't you think?

Yeah, that's what I ment a couple of posts above I'll try to keep it flying right over EGLL as much as I can. Give me a couple of minutesEDIT: Ready now: http://www.divshare....ad/13049680-426

Edited November 2, 201015 yr by dazz

Yeah, that's what I ment a couple of posts above I'll try to keep it flying right over EGLL as much as I can. Give me a couple of minutesEDIT: Ready now: http://www.divshare....ad/13049680-426

Erm, you forgot to change the airline density to 0, it still loads to 100%.Running tests now.EDIT: Results are in.Btw. your results, are those based on the old straight test, or the new, turn based test?

My results currently are the ones I got in the old benchmarkI thought we agreed on leaving airline traffic on? (I explicitly asked to disable traffic addons and use default AI in the instructions)We can always disable traffic and test it like that. Looks to me like no one else is going to join us anyway -_-I'll run the tests using the new benchmark nowThen I'll try again at 4.2GHz to make a clock for clock comparison, but 4 vs 4.2GHz is more of a real life comparison between our chips, since usually Bloomfields overclock that much better than LynnfieldsSomething that seems quite evident is that changing the resolution or AA/AF has no impact whatsoever. Sort of pointless to repeat the test on vanilla graphics

Erm, you forgot to change the airline density to 0, it still loads to 100%.Running tests now.EDIT: Results are in.Btw. your results, are those based on the old straight test, or the new, turn based test?

My results currently are the ones I got in the old benchmarkI thought we agreed on leaving airline traffic on? (I explicitly asked to disable traffic addons and use default AI in the instructions)We can always disable traffic and test it like that. Looks to me like no one else is going to join us anyway -_-I'll run the tests using the new benchmark nowThen I'll try again at 4.2GHz to make a clock for clock comparison, but 4 vs 4.2GHz is more of a real life comparison between our chips, since usually Bloomfields overclock that much better than LynnfieldsSomething that seems quite evident is that changing the resolution or AA/AF has no impact whatsoever. Sort of pointless to repeat the test on vanilla graphics

And I thought we agreed on 0% traffic. :biggrin:In all honesty - is it easier to mess up disabling the traffic externally, but rather just put the slider to 0%? Do what you think, but with the backup of fsx.cfg, nothing goes wrong in this field...I just think testing AI is pointless.I believe we will have people testing. Just have to see what we really do with traffic :Big Grin: I had virtually no difference between 4 and 4.2. Left it running at 4.2 just for the sake of it!Yeah, I think we can leave it at one test, 1280x1024 + AA/AF in FSX. Other test is pointless.