I had a slight improvement with overclocking the graphics card in the VC, but it could also be too subjective and within "statistical error".I forgot to mention before that I too have default.xml disabled. Just a note about the 7600GT. You are right about the relative performance, but I'm sure you at 256MB and I'm at 512MB. It's just possible this scenario is pushing over the 256MB limit, which might slow you down a little.Having FSX control the AA and AF gives me a horrible looking screen. Seems to work a lot better when the card controls it.I'm a little despondent about the 8800 results. Perhaps at higher resolutions it would stand out more (relatively speaking). Was planning an upgrade.It sounds like it's worth having a number of benchmark scenarios which may expose different performance factors in our machines.

>>On the other hand overclocking my 4800+ X2 doesn't seem to do>anything to the benchmark, >I think that's because your 7600 is your bottleneck. At least that's logical to assume based on the effect o/c is having for you.In my case, overclocking the cpu does indeed impact fps, so I think in my case the 7800GT is handling everything it's getting from the gpu.>What do you guys think? Would a benchmark WITHOUT any traffic>be more valuable after all?>That is part of the problem people had with this when it was suggested in October. Everyone said there were too many variables. Traffic is one of them. I, for example, have CUSTOM traffic files I have made (American Airlines, Continental, Copa, Grupo Taca so far) and you see, those will adversely affect the bench for me, unless I disable them.Especially American Airlines, which will have lots of Los Angeles area aircraft flying around the test area.I probably would disable the traffic for testing. Then maybe do a 2nd set of benching with traffic enabled.Also I noticed you want bloom enabled. My gosh man, do you WANT to kill my frames? I never run with bloom.RhettAMD 3700+ (@2310 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2 GB Corsair XMS 2.5-3-3-8 (1T), WD 250 gig 7200 rpm SATA2, CoolerMaster Praetorian case

>>I forgot to mention before that I too have default.xml>disabled. That's another variable...I too have default.xml disabled.I also use Matt Fox's Autogendescriptions.spb file.RhettAMD 3700+ (@2310 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2 GB Corsair XMS 2.5-3-3-8 (1T), WD 250 gig 7200 rpm SATA2, CoolerMaster Praetorian case

>I probably would disable the traffic for testing. Then maybe do a 2nd set of benching with traffic enabled.>Also I noticed you want bloom enabled. My gosh man, do you WANT to kill my frames? I never run with bloom.I agree, if you want to use some traffic, limit it to ground traffic, as that shouldn't be different between systems, but ideally I think a moving variable like that should be off on a benchmark.As for Bloom, I also think that should be off, as it isn't representative as the way most currently use FSX. Most only turn it on to take screenies but turn it off normally, because of the severe perf hit!

Thanks for the input guys. I pretty much agree with all what you said.Ok, maybe we should do something like this: Take this default scenario with bloom and air traffic disabled and add maybe another scenario at night with bloom, but still without traffic. What would also be interesting would a spot view benchmark at an airport with bad weather. I don't want to make it too complex or noone will use it. Or should we just leave it at one benchmark location? Maybe I'll pick one location who has it all: mountains, dense city scapes and water. Not sure.I enabled bloom in the first place to really stress both the CPU and GPU.Pat

>Thanks for the input guys. I pretty much agree with all what>you said.>>Ok, maybe we should do something like this: Take this default>scenario with bloom and air traffic disabled and add maybe>another scenario at night with bloom, but still without>traffic. What would also be interesting would a spot view>benchmark at an airport with bad weather. I don't want to make>it too complex or noone will use it. Or should we just leave>it at one benchmark location? Maybe I'll pick one location who>has it all: mountains, dense city scapes and water. Not sure.>>I enabled bloom in the first place to really stress both the>CPU and GPU.>>PatSounds good to me! One other suggestion. Maybe it might be a good idea to do this using the recorder feature in FSX, instead of a saved flight. FSRecorder would probably be better, but not everyone will have that! This will eliminate any drifting, that can happen in free flight, the recorded flight will track the same every time. EDIT: Scratch that idea of the recorder, I don't think you could save things like weather that way.

Alrighty, I just ran the benchmark a few times with different CPU and GPU speed settings and got:1. CPU @ 3.5GHZ, GPU @ 650/400 - 37.5 (14.5 & 23)2. CPU @ 2.13GHz, GPU @ 650/400 - 29.5 (10.5 & 19)3. CPU @ 3.5GHz, GPU @ 395/243 - 35.5 (14 & 21.5)Comments: 2.13GHz is CPU stock speed and 395/243 for the GPU was the same proportional reduction for the GPU as I did for the CPU (ie. approx 40% slower)Analysis: The 40% slower CPU speed resulted in a 21% reduction in benchmark score. The 40% slower GPU speed resulted in a 5% reduction in benchmark score. ie. CPU speed has a 4 times greater impact on FPS than GPU speed does. CPU overclocking is 50% efficient (ie. your FPS changes by half the change you make to your CPU speed), and GPU overclocking is 12.5% efficient.Conclusion: For this particular scenario and FSX setting on my system, CPU speed is the prime determinant of FPS achieved.Gary

I don't know about with FSRecorder, but you can certainly load the same weather when using FSX's inbuilt recorder. The important thing is to have a saved scenario with the weather included, then playback the recorder after loading this scenario.Gary

I just re-ran the benchmark with default.xml renamed and my mark shot up to 30 (10+20). Not bad.I ordered the 8800GTS anyway, but I don't expect a lot of earth-shaking performance in FSX. At least, I'll be all set for DX10. ;)Pat

Thanks Gary,I see your spot view is virtually the same as mine. At a guess I would say my larger cache and lower o/c is compensated by your better and faster memory.

Come back to us when you have your 8800GTS running with your impressions. I've been considering such a purchase too.

Updated benchmark data:################################################CPU: AMD64 3500+ 90nm 2.21 GHzGPU: ATI 9600XT - 256Mb 527MHz/250MHz Catalyst 5.10Memory: DDR400 512x2 Mb 2.0/3/2/6 1T 200.9MhzOS: Win XP SP2FSX Benchmark: 18.8 (6.8 + 12.0)Remarks: Default install no tweaks no addonsCPU/Mem per CPUZFrames per FRAPS99#################################################################################################CPU: AMD64 3500+ 90nm 2.21 GHzGPU: ATI 9600XT - 256Mb 500MHz/250MHz Catalyst betaMemory: DDR400 512x2 Mb 2.0/3/2/6 1T 200.9MhzOS: Win Vista RC2 build 5744FSX Benchmark: 15.4 (5.8 + 9.6)Remarks: Default install no tweaks no addonsCompatability mode (Aero off)CPU/Mem per CPUZFrames per FRAPS99#################################################scott s..

>I just re-ran the benchmark with default.xml renamed and my>mark shot up to 30 (10+20). Not bad.>>I ordered the 8800GTS anyway, but I don't expect a lot of>earth-shaking performance in FSX. At least, I'll be all set>for DX10. ;)>>PatDo you know if there is still this talk out there about how the 8800 series cards do not implement all of the DX10 features?That was a netrumor that I heard. Don't know if there is/was anything to it though.RhettAMD 3700+ (@2310 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2 GB Corsair XMS 2.5-3-3-8 (1T), WD 250 gig 7200 rpm SATA2, CoolerMaster Praetorian case

I doubt that that rumour is true.According to nVidia: "GeForce 8 Series GPUs include all required hardware functionality defined in the Microsoft Direct3D

CPU: Intel E6700 (stock)GPU: nVidia 8800GTS 640MB (stock)Memory: DDR2-667 4GB (stock)OS: Windows XP}FSX Benchmark: 37Remarks: RAID 0 HD; FSX cfg tweaks (Terrain Max Autogen Trees 600 per cell, Buildings per cell 3000)