
... FSAA on Radeons is just a joke though, it might run slightly faster than the Standard IntelliSample setting (which is set to Quality over Performance btw) but even on High-Performance the FSAA on all GeForce at 2x will only cut your fps by 5 and at 4x cuts it only be 15.
personally i don't believe much of the online benchmarks i see of the Radeon cards - i have some here, and i keep them as far away from my machines as possible.
quite frankly GeForce are the best budget cards, and thier top end perfomance is just remarkable... and i have yet to see someone show me otherwise to my face.
i hear all of these statistics all the time, which is all well and good when i hear something like "aww man i had a Radeon 9700 running Unreal Tournament 2003 at 1600x1200x32 and it was running at a sweet 50fps" - so what did i do, i decided to go down to the shops get myself a Radeon 9700pro ... set it all up.
First thing that happened was the entire vsync for my monitor went into a mode it couldn't handle, after sorting that out i spent 2hours just trying to get Unreal Tournament 2003 to even load using the bloody card. And when i did finally get it all going ... fast would hardly of been the term i'd use to describe its speed.
Next day i find out this friend has a Dual Pentium4 2.5Ghz Rig with 1.0Gb of RDRAM ... whereas i was trying to see this thing work on an AMD AlthonXP 1200+ w/512Mb Ram.
and aparently he also told me he was running Windows2000 not WindowsME and they aparently ME isn't a viable operating system for Radeon cards.
I also have a 9800pro ... and no matter what i do there is no way in hell i can get the bloody thing to benchmark as quick as the Radeon9800pro on the 3DMark03 site, and thats on a Dual Optiron 3.0Ghz system - same setting the lot.
It also rejects alot of the Shaders i create within Cg, and has a tendacy to garble all of the DOS games graphics i play. It oftenly rejects OpenGL games suchas Quake3 ... one minute they work, next it doesn't want to load because of a memory error.
as far as stability goes, if the Radeons are as stable as a Rock then that Rock must be make out of Marshmellow, because i've had so much greif with the cards i have i ended up buying a whole range of budget cards to fill the gap.
i've never has so much bloody trouble with a range of card since SiS, and to be honest i have one of the new 680's and it doesn't even give me as much grief as the Radeons.
quite frankly the only viable arguments i've heard against the FX chipset is the noise and the size ... and to be honest it is barely any louder than a the Radeon, which whipser quiet is not a term i'd give its fan - its as loud as the standard CPU fans.
Taking up two slot, well you can actually edit the card not to do that ... it only does it to make sure it never overheats without the use of the 3" tall heatsink lik CPUs use. It isn't a nessity.
and at the end of the day the GeForceFX 5200 is cheaper, runs FAR FAR cooler, is possibly if not the most stable card on the market ... plus it runs on CineFX 1.5 (total shader version support) as well as HCT IntelliSample which is faster FSAA than previous cards.
Quite frankly if you want to see a list of supported features checkout both sites, and it is noticeable from when you first open each page that the FX's is longer - as for Speed, the DetonatorFX drivers edit the speed problems found within the Detonator4 drivers
as for noise i doubt you'll even notice it - the only people who do are those geeks who've gotta have the least noisy system in the world, even then i personally have edited my FX5800 (and probably do the same to the FX5900 soon) to use a passive heatsink so zero noise at all ... take up the same slot space, but still less noise.
Within the Epic battle of the fates the Shadow and the Angel will meet. With it will harbinger the very fight of good vs evil!