>Where did I say the test was run without filtering? I tried>some tests myself and adding AA of 4 and AF of 8 incurred no>performance hit at all. If I increase them to 8 and 16 there>is still no performance hit.You didn't, but the tests performed by Legit Reviews with 3DMark06 seems to be run without any AA. Seems that this ATI card is very good. I wonder why Nick doesn't thing that the tests done by Toms Hardware are reliable. I hope that he responds to my question.Ulf B

<>The benchmarking of the games appears to have AA set at 4X. Not sure about AF.Their review of the 9800GX2 shows it to be slightly faster overall than the 3870X2 but with a price premium of $200. That review also shows the 3870X2 to outperform the 8800 Ultra on many games.It's a nightmare choosing cards for those who want to play many games. One card can be brilliant at game A only to suffer with game B. Fortunately I only use FS so it's not a problem for me.All I would say about TomsHardware review is that if the 8800 Ultra is 6fps faster than the 3870X2 you should be able to put Scenery and AutoGen Complexity to max and get 25+ around London. Hopefully Nick will reply.Yes, it is an excellent card and should be on the shortlist of anyone considering a new system. PC Pro (UK) described it as "the world's fastest graphics card". Debatable maybe but certainly fast.Cheers,

Because Tom's site is unreliable and will manipulate their "tests" to get the results their sponsors/advertisers like best.I don't know who pays Tom these days.

>Because Tom's site is unreliable and will manipulate their>"tests" to get the results their sponsors/advertisers like>best.>>I don't know who pays Tom these days.OK, I get the picture. Are there any site thats reliable? A site that isn't dependant on sponsors/advertisers? And have enough capital to invest in several reference hardware plaftforms to perform benchmarks on?Do you know of any sites that are better than Toms Hardware? I'd really apreciate to be able trust in benchmarks.Ulf B

There is a note here http://computershopper.com/reviews/nvidia_...2_graphics_card ,under Cons, that says when using multiple monitors one of the gpus has to be turned off.

That's not good. But that should be a driver issue, which could be corrected if nVIDIA want's to.Ulf B

Its like this.. Tom got caught (documented) with his hands in the cookie jar many years ago when he started a deal with Intel well before the Core2 was ever on the market or conceived. There was a time when I could check the main page of all the high traffic areas of his domain and whom-so-ever had the most advertising posted, I could tell the result of any article or benchmark, before reading it... hilarious tooSo Tom is worthless for trusted results.Anyway, benchmarks and articles with controversy are all about web hits. In that they all have a little "Captain" in them. http://www.youtube.com/watch?v=kum2woR1KqgWhat I do is read about 7 primaries such as AnTech, HardwareZone, HARD, Trusted Reviews, Xbit Labs, etc, etc and look for the common denominators. Many times they don

I don't think there are any reliable sites.The number of people who can do proper testing is low and most of them will be working for commercial interests.It's rather expensive and time consuming to buy everything that gets released where it comes to PC hardware and test it.So those people will need someone to pay them to do their work, and the only people willing to pay for it are either those same manufacturers or magazine editors who themselves depend on those manufacturers for their own advertising income.And it's hardly limited to this industry.Effectively no magazine or website publishing such comparative reviews can be trusted to not be biassed.Just about the only trustworthy sources are companies like the BBC that don't rely on advertising for their income (The Beeb is paid out of taxes).Worst case I've seen was a "review" of a dirt cheap low quality Tamron lens which was compared to 2 professional top quality (and expensive) Nikon lenses.The Tamron was the winner in everything, the "review" completely blasted the Nikon lenses. Of course the only things that matter for serious users, image quality and durability, were not even mentioned as criteria as testing on those would have utterly destroyed the Tamron.That same magazine contained 2 full page advertisements for that Tamron lens, the only Nikon advertising was a small logo printed by a dealer.

Nick,Thanks for the insight about Tom's. I'll bookmark the sites you mention and make it a habit to scan all of them when investigation news on new hardware.Ulf B

That was my pointTrust NONEUse them all and look for the areas one may highlight or skip over the other.They are all in business to make money, not bring you the best and most honest reportsI have even been able to trace a manufacture rep from one site to the next by reading key words and statements because I know the rep and I know the 'terminology" they push... there have been times when I have seen the same review posted at 3-4 'reputable' sites that were clearly skewed by the manufactures rep, no doubt about it.Tom was arrogant and that

<>As we are only interested in how FSX performs with processor A vs processor B or graphics card A vs graphics card B we have the capability to organise our own testing.Create a static scenario and save the flight. London City airport Rwy 28 is not dissimilar to Meigs which I used to use for testing. Then push all option sliders to max and save the configuration. Upload the files with instructions on how to setup FSX to AvSim's library.Those who want to run a test do so with a controlled situation. It's not perfect but it doesn't need to be. It will show what cards perform better at a given resolution which is the main reason for this discussion.Use the frame counter in ATI Tray Tools or the nVidia equivalent for measuring fps. Get a volunteer to receive the results and store them on a spreasheet or whatever.The two most important things are we all use the same test flight and we all use the same option settings and those are 100% controllable.If I was retired I'd run it but working full-time doesn't leave me with any spare time.

There are indeed driver issues with the x2 3870 which are directly related to the 2nd core use.When you see 'patches' for games such as Crysis that come with the tag "SLi Patch" its another bull' routine. SLi and Crossfire is in the DRIVER, not the GAME. Phil already bloged about that well over a year ago. When they release a patch that enables or betters SLi or crossfire, its because THEY screwed up on their side of the DX API, not the hardware.ATi engineering is working on the last part of the x2 issue which is enabling the full function of the 2nd core... and part of that will come with a price. The PCi2 2.0 standard DOES IN FACT add more power to the slot for higher functions... those who say 2.0 is not needed are wrong. The standard increases power to the slot (and add other things too) which hardware needs to function at 100% ability. Its not all about bandwidth, its about meeting power requirements in the standard. There is backwards compatibility however full function will be restricted if the chipset and slot (and drivers) are not designed to the sec the card wants to see by default. Higher functions such as 2x 2x3780 will not be possible on a standard motherboard for the same reason. How that may affect performance is yet to be seen until the drivers have completed thier release dev cycle.

One other thing tooboth the x2 and the 9800 are the END of an era and the BEGINNING of the nextJust as the 7950Gx2 was the end of the 7000 series the 9800 is 2x the same old crap. The x2 is the setup and dev for the R700 xfire systemsI would wait till this May-June because all these cards are outdated for the next gen tech which will start to appear right around that time.For now, a 8800GTX clocked nicely will do just fine with FSX.. its about the CPU and the memory bandwidth and how it is setup/used with FSX more than the card. I would put the money into the platform and the CPU first..buy a cheaper card, and expect to upgrade the card in the next 6 months

the issue with FSX is that it is DYNAMICbenchmarks are therefore NOT COMPLETELY LINEAR in resultsthere are at least 10 factors I can list which will influence the same benchmark on equally built (so you think it is) towers.FSX is a very hard nut to crack and claim a benchmark is accurate for everyonethe key to FSX is CPU cycles, memory bandwidth and how it is applied to those cycles (NOTE: That does not mean HIGH CPU speed and HIGH memory speed, that means the RIGHT way to apply bandwidth, its CAS in relation to tRD and FSB/STRAP)Since a motherboard/BIOS may have completely DIFFERENT methods of setting up timing, sub timing and relating that to the CPU, you CAN NOT USE A BENCHMARK for linear FSX results BETWEEN SYSTEM A and B. You can get CLOSE if the tech sets up the BIOS and confirms ALL in special software such as CPUz and MeMSET, etc. Only then can linear results be examined for accurate benchmarks.. and even then, 2 different manufactures of the SAME video card may use memory with different spec internal timings, and MANY other electronic factors that play into results.You CAN use a benchmark to compare you OWN results on a linear level.. and that is as far as the accuracy goes.

Nick,I respect your opinion but FSX is CPU bound - just as FS98 was CPU-bound 10 years ago when I last ran a benchmark test on CompuServe.A test using a standard static test back then showed which graphics cards and CPUs performed best and I believe the same applies now.I'm not talking about simulating the setup when a magazine tests cards or CPUs. That is not possible. But a test as I have described would still show if one card performs better than another in FSX.Anyway, without a volunteer it's academic anyway but I still feel it would be more useful than controlled tests of games that have no bearing of how that hardware performs with FSX.No need for the capitalisation of certain words - I can understand what you're saying in lower case.