It is salutory - and disheartening - to learn that not all cards with the same specifications perform equally. No wonder people report having such mixed experiences. It is one thing for the manufacturers to rely on advertising/marketing hype. But this sort of variance sounds to me like something different. It sounds as it you gamble when you buy a card: you might get a duff one and just live with the disappointment under the assumption that you have been a victim of hype, rather than of faulty manufacturing. These cards are not cheap. People really shouldn't have to take "pot luck" with them. I suppose there's no practical way for such a tiny interest group as FSX-users to force the industry into being more transparent with consumers. But I wonder whether we might set a ball rolling? I wonder what coherent argument the manufacturers could mount against this suggestion: package the cards with a simple diagnostic tool - or let it be downloadable from the internet - which lets consumers test the card (including performance) against all relevant parameters and against known benchmarks, so that they can tell at once whether they've got a duff one. Tim

Tim there are several wildcard variables in the manufacture of all electronic devices. 3 of the primaries when we are talking about circuit boards area. Component bin quality: Although there is a min spec required for the components used on cards, their quality can still vary and will usually create a +/- result in the final products. b. design of and ...c. installation of heat dissipation devices: This is where a very large variance in final products existsOne of 285's I got got back displayed a better result but not up to the 1st card. I noticed there was a 5c difference between the two and took the heatsink off the card, cleaned and reseated it and the result fell in line with card 1. 5c doesnt sound like much and according to the published specs it shouldnt make any difference, right? .. but it did/does and can with these devices especially if the memory chips, and a few other components, which are not monitored on video cards and motherboards, are getting hot. Thats just 3... and users can see anywhere from a 1-20% difference between cards and never know it. I consider anything above 13% a defect. One of things I do after I have tested cards and select one to keep (I give the others to a local tech school for their systems) I always remove the heatsink and either replace it with one of better design or do a bit of cleanup and surfacing on it. I never had to deal with the noise of the 8000 series because I use the Thermalright HR with a 90mm fluid dynamic bearing fan which made no noise at all unless the card was in 3D mode and even under a load never exceeded 70% of its rated speed due to the efficiency of the HS itself.I do the same thing with motherboards.. the heat rails come off, the rail and the chips/VR's are sanded and cleaned, then thermal epoxy applied and the rail reinstalled. A not for the faint at heart operation because if that epoxy kicks before the rail is secured, you got problems LOLBut it does work and it does make for a much better system since electronics manufactures really skimp when it comes to that area.

maybe it`s your statistical methods and not the manufacturers, ever think of that.. 4 machines is hardly enough samples to come to a `real world`probability, i believe the magic number would be 30.

maybe it`s your statistical methods and not the manufacturers, ever think of that.. 4 machines is hardly enough samples to come to a `real world`probability, i believe the magic number would be 30.

Of course 4 machines are not going to present a 100% real world probability.. thats a given. I think I would probably push that to 100 with 4x25 using the different OS's and also different hardware combinations but 4 machines does present a greater range of result that does reveal issues .. as does real world use in compare,.. which is far better than using one machine and assuming everything in the process.

Exactly and who has time for that.. I don`t believe in the hardware variation so much.. i think what`s needed is a large amount of test runs with regards to scenery and aircraft to come to a decent conclusion.

I don`t believe in the hardware variation so much..

If we are talking about IC's single components and proc's, I agreeIf we are talking about a total design including the PCB process and heat removing devices,... I disagreeand I would disagree abouth the systems those cards go into as well which present a wild card to results without being verified, as the cards should be verified, first. None the less, the best way to real world the situation to include all factors and find a stable average is to have a database where users add the result of several predefined tests.. it would be best to define the test(s) with strict instructions as to how the tests are to be executedOf course one must also must trust users are not going to upload exaggerated data because they want their system to look better than others too.

I know! . . . believe me I was astounded (and incredibly disappointed) when I had the nVidia card in, and I was flying into KBOS under heavy overcast and there it was -- spikes through the air and textures ripping themselves to pieces. It was the first nVidia card I've ever seen it happen with, and I'm no longer in any mood to try and push hardware around so that it works right so it went right back to the retailer. Problem eliminated with the 5870.

At the risk of incurring the wrath of my peers by posting a "reply" to my own post, another advantage I've noticed with the new 5870 is that it seems to be running a bit cooler, and that allows my CPU to run 4 - 5 degrees C cooler as well. The 4870 runs HOT, passing that ambient heat throughout my case. When cooling only with air, that was a constant concern for me. This may allow me to goose the O/C up even a bit more now -- maybe that's how I'll capture a few more frames from this purchase.

Unfortunately I don`t think there`s a single FS playback software that doesn`t bring in it`s own issues, especially with regards to shimmering. Heck even the MSFS video replay is broken. Regardless I think a good number of repeatable test runs for a minute or two would be far better than the current approach of `flying over new york`` or heavy weather versus light weather... individually they have merits, but as a whole that`s not enough. I think that`s the one thing I really liked about Crysis, the fact they gave you a program to benchmark with, so you could easily find the sweet spot no matter what combination cpu gpu you were using. A fix to the playback and shimmering of FSX would have to be fixed frame rates of 20 or 30, and conclusions would be based on if they can in fact maintain the FPS under higher levels of AA and AF and resolutions. Sure these would be ordinal results but I think small differences should be ignored in the first place, and it would definitely provide a more qualitative comparison, and FPS would just be based on flicker free performance.

Exactly and who has time for that.. I don`t believe in the hardware variation so much.. i think what`s needed is a large amount of test runs with regards to scenery and aircraft to come to a decent conclusion.

The different batches of C0 Core i7 920 and 940 processors showed up to be very different when OC:ed. Some could OC to 4GHz without any problems while others had big problems getting the processor to 3.8GHz. Article at AnandTech on this if you care to read it.Why should the manufacturing processes regarding other type of chips differ regarding quality? You get what you pay for and hardware is maybe too cheap.

I think that`s the one thing I really liked about Crysis, the fact they gave you a program to benchmark with, so you could easily find the sweet spot no matter what combination cpu gpu you were using. A fix to the playback and shimmering of FSX would have to be fixed frame rates of 20 or 30, and conclusions would be based on if they can in fact maintain the FPS under higher levels of AA and AF and resolutions. Sure these would be ordinal results but I think small differences should be ignored in the first place, and it would definitely provide a more qualitative comparison, and FPS would just be based on flicker free performance.

Of this we both agreeAs long as the manufacture of the product applies a complete set of simulation variables for the hardware to deal with and lock that test so it will always produce the same in repeat, then we have established a baseline for accurate compare. But there comes into the mix that Crysis is not a simulator, and, its a local area map game where MSFS is in some respects loading the world up. But with a proper design of such a test which is tied directly to the core of the product I do think your suggestion would be the best in terms of a benchmark, especially one that does not need such externals as FRAPS which also has a high potential to vary the result.

When I use my favorite aircraft the Level-D 767, it seems that I get the same FPS as with the 8800GTX. So does this show, that the VC is very depended on CPU power?

VC's payware brings my fsx from 30 to a crawly ~6[/] fps (in fs9, from 60 to 20 :D) so i'd say you're right on that one.

So at the moment it's obvious to be the 5870 and I'm sure the GT300 will be the new king soon. But should I wait? Unfortunately I have not too much patients :(

For me there is no spoon. Only the ATI offers 3 displays @ 5760x1080 without the need of a tripplehead, so we SIMMERS can fully enjoy our amazing setup, earned with blood and tears :)have a nice night \o\

VC's payware brings my fsx from 30 to a crawly ~6[/] fps (in fs9, from 60 to 20 :D) so i'd say you're right on that one. For me there is no spoon. Only the ATI offers 3 displays @ 5760x1080 without the need of a tripplehead, so we SIMMERS can fully enjoy our amazing setup, earned with blood and tears :)have a nice night \o\

Just setup my ATI 5870 last night for 3 displays. Had to get a Dell Displayport-to-DVI adapter to make the third work which just came in yesterday.I have to say I'm impressed. I used to have twin Nvidia 9800GTXs. Running just one monitor at 1920x1080, I averaged around 25fps at SeaTac Airport during taxi and takeoff. Adding a second or third crippled the frame rates-- dropped to about 6-8fps. On the 5870, I'm running at full 5760x1080 and averaging a surprising 27fps under the same conditions. Once in the air, I get around 50fps. The card handles the additional resolution very well.

Newbie, your FPS sound mighty impressive. What CPU are you using? What are sliders set at on average? Is this FS9 or FSX? Oh, and what size are your displays?Thanks,Clutch

Newbie, your FPS sound mighty impressive. What CPU are you using? What are sliders set at on average? Is this FS9 or FSX? Oh, and what size are your displays?Thanks,Clutch

Here's the specs:Core i7 920 Quadcore on a Gigabyte EX-58Diamond ATI 5870 6GB RAM2TB 7200 RPM Hard DriveAsus 23" DVI/HDMI LCD @ 1920x1080FSX SettingsDetail: LargeComplexity: MaxMesh: MaxTexture Res: MaxWater effects: Low 2.xScenery Complexity: DenseAutogen: NormalSpecial Effects: HighTraffic: very lowAnisotropic & Antialais on Catalyst settings: DefaultDX10 Preview on -- this seems to have the most impact compared with my previous card. Not sure why but seems like the ATI handles DX10 better. On the 9800 GTX, turning on DX10 didn't make much difference. On the 5870, turning it off reduces the performance to 8-10 fps.Also, the ATI Catalyst is set to "application settings"Have not overclocked the 920 yet but that's next on my list.

Played around some more tonight. Tried spanning monitors in portrait mode. Looks amazing! At a glance you can see all your instruments and a great outside panorama.

Played around some more tonight. Tried spanning monitors in portrait mode. Looks amazing! At a glance you can see all your instruments and a great outside panorama.

Played around some more tonight. Tried spanning monitors in portrait mode. Looks amazing! At a glance you can see all your instruments and a great outside panorama.

That looks pretty dandy and also what I'd like to do too! However, I'm waiting for new monitors to be released with a thinner "frame".As for overclocking the i7 920, this is what I did with my Asus P6T V2 Deluxe Mobo and it's rock solid running at 3,8 GHzAi Overclock Tuner => ManualIntel Speed Step Tech => DisabledIntel Turbo Tech => DisabledCPU Ratio => AutoBCLK Freq => 190 (= 3800 MHz)PCIE Freq => 100DRAM Freq => DDR3-1523MHzUCLK Freq => AutoQPI Link Data Rate => AutoVoltages-----------CPU Voltage => 1.25CPU PLL Voltage => 1.80QPI/DRAM Core Voltage => 1.25DRAM Bus Voltage => 1.64CPU Differential Amplitude => 800mV... all other voltages to Auto