Anand's Smooth Vision Article is up!
- Thread starter Adul
- Start date
interesting, but I think there is some confusion on Anands part with regards to multisample and multi-buffer supersampling
flexy
Diamond Member
- Sep 28, 2001
- 8,464
- 155
- 106
well....it basically shows that the radeon is superior with 2 x Quality AA in comparison to a gf3 with aniso AND Quincunx...and it shows that the 4xAA algo of the radeon is much slower than 4xAA on the geforce.
But...i dont care about 4xAA...since i dont see a point in enabling 4xAA when 2xQuality AA gives great performance and picture quality on the radeon, shown also clearly in the screenshots. In reality you want the best possible picture quality *combined* with best performace (== 2x quality AA).
I've got to wonder.. is Anand accusing ATi of doing a 'simple' supersampling method of Anti-aliasing (simply upping the resolution rather than doing something similar to 3dfx's or nVidia's method)? does he have any proof besides the fact that the GTS is forced to do that type and achieves similar texture quality (what about the performance hit??)?
the fact that ATi is very secretive, is either becuase they don't want to let the secret out, or becuase they're don't want us to discover the truth.. the second of which doesn't make any sense to me, becuase AFAIK the Radeon was already doing the SuperSampling method found on the GTS beforehand, so why change the name and call it a different method, and cover it up?? what scenario seems more likely? perhaps with their new 'history' of forcing certain settings in benchmarks to get performance and not telling people about them (lol, nVidia doesn't tell us their default settings, some of which can be discovered by looking at the control panel, others which can't be found without extremely close examination..
the fact that ATi is very secretive, is either becuase they don't want to let the secret out, or becuase they're don't want us to discover the truth.. the second of which doesn't make any sense to me, becuase AFAIK the Radeon was already doing the SuperSampling method found on the GTS beforehand, so why change the name and call it a different method, and cover it up?? what scenario seems more likely? perhaps with their new 'history' of forcing certain settings in benchmarks to get performance and not telling people about them (lol, nVidia doesn't tell us their default settings, some of which can be discovered by looking at the control panel, others which can't be found without extremely close examination..
Anand mentions that he thinks that is what the Radeon 8500 is using, and ATI's response kinda points towards that. I don't know why it matters, just because it's not as complex doesn't mean it's worse. It provides higher quality right? Does it matter what technology it uses?
On a side-note, I remember back in the days of the Voodoo5 vs the Geforce 2 that someone mentioned that the Voodoo5's AA quality, being multisampling, actually achieved higher AA quality than both the Radeon and the Geforce 2. Just makes me wonder how come nVidia hasn't been able to implement that multisampling technology into their chips, having purchased almost all of 3dfx's patents and technology. Perhaps this superior and yet efficient AA method will appear in the next line of products.
On a side-note, I remember back in the days of the Voodoo5 vs the Geforce 2 that someone mentioned that the Voodoo5's AA quality, being multisampling, actually achieved higher AA quality than both the Radeon and the Geforce 2. Just makes me wonder how come nVidia hasn't been able to implement that multisampling technology into their chips, having purchased almost all of 3dfx's patents and technology. Perhaps this superior and yet efficient AA method will appear in the next line of products.
Anand is confused saying Radeon 8500 uses SSAA but it should use MSAA like geforce 3 but what he doesn't realize is that Radeon 8500 was alway SSAA using multiple buffered samples just like how voodoo 5 rendered it.
I think his confused about multisampling AA and multiple buffered samples.
He really shot himself in the foot on this one.
I think his confused about multisampling AA and multiple buffered samples.
He really shot himself in the foot on this one.
<< Anand mentions that he thinks that is what the Radeon 8500 is using, and ATI's response kinda points towards that. I don't know why it matters, just because it's not as complex doesn't mean it's worse. It provides higher quality right? Does it matter what technology it uses?
On a side-note, I remember back in the days of the Voodoo5 vs the Geforce 2 that someone mentioned that the Voodoo5's AA quality, being multisampling, actually achieved higher AA quality than both the Radeon and the Geforce 2. Just makes me wonder how come nVidia hasn't been able to implement that multisampling technology into their chips, having purchased almost all of 3dfx's patents and technology. Perhaps this superior and yet efficient AA method will appear in the next line of products. >>
appearently the inquirer thinks they'll implement this in the nv30 (okaay i know it is the inquirer, but they just might be right for once ...)
EMAN, you're confusing me.. you're saying that the FSAA algorithm that the Radeon 8500 has is really the one from 3dfx Voodoo 5's?
in case you want to know, that particular algorithm is called RGSS (Rotated grid super sampling I think). the card draws the same frame multiple times, just with a different grid, so the pixels end up slightly out of alignment from one another, which means that when combined, they have this FSAA effect.
the reason why that algorithm seems to be one of the best is simple, it works with nearly all games. rendering in high res and bringing it down simply doesn't work quite properly, especially when you have some sort of 2D control panel that cannot be scaled to 1600x1200+ resolutions. IF anand wants to find out what algorithm this card is using, he could have tried a game like that.
nVidia's GF3 method of anti-aliasing I think follows an SGI technique. I don't know much about it, but from what I remember the GF2 FSAA technique is simply scaling the resolution much higher and then downsampling.
in case you want to know, that particular algorithm is called RGSS (Rotated grid super sampling I think). the card draws the same frame multiple times, just with a different grid, so the pixels end up slightly out of alignment from one another, which means that when combined, they have this FSAA effect.
the reason why that algorithm seems to be one of the best is simple, it works with nearly all games. rendering in high res and bringing it down simply doesn't work quite properly, especially when you have some sort of 2D control panel that cannot be scaled to 1600x1200+ resolutions. IF anand wants to find out what algorithm this card is using, he could have tried a game like that.
nVidia's GF3 method of anti-aliasing I think follows an SGI technique. I don't know much about it, but from what I remember the GF2 FSAA technique is simply scaling the resolution much higher and then downsampling.
Soccerman,
voodoo 5 uses the same SuperSampling AA but with Rotated Grid using multiple buffer samples.
Radeon8500 uses SuperSampling AA but pseudo random sampling, and was done with multiple buffers instead of a super scaled backbuffer.
Old Radeon and GTS uses SuperSampling AA with Ordered Grid and using super scaled backbuffers.
voodoo 5 uses the same SuperSampling AA but with Rotated Grid using multiple buffer samples.
Radeon8500 uses SuperSampling AA but pseudo random sampling, and was done with multiple buffers instead of a super scaled backbuffer.
Old Radeon and GTS uses SuperSampling AA with Ordered Grid and using super scaled backbuffers.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter