Originally posted by: Sentential
Sounds like there are serious issues with 90nm. Not suprising, they should have known better than to do something that stupid.
Originally posted by: cbehnken
ATI plans to use Tile based rendering, therefore it should be very easy to allocate a correct amount of screen for the slower card to render without slowing anything down.
In all respects so far, AMR looks like a more thought out and arguably better solution than SLI.
Originally posted by: Rollo
Originally posted by: Sentential
Sounds like there are serious issues with 90nm. Not suprising, they should have known better than to do something that stupid.
There are always yield problems with a new smaller die size. A lot of people faulted the 5800s for that, I commended them for moving the tech forward. Same here, it needs to be done, there will be bumps in the road.
Originally posted by: Rollo
Originally posted by: cbehnken
ATI plans to use Tile based rendering, therefore it should be very easy to allocate a correct amount of screen for the slower card to render without slowing anything down.
In all respects so far, AMR looks like a more thought out and arguably better solution than SLI.
Think about what you just said:
You believe AMR will reduce the amount of the screen rendered by the slower card to maintain the speed of the faster card.
OK- where's the benefit then? With load balancing like that, how are you going to increase framerate? If one card is only doing 30% of the work because the other is a a lot faster, and there's overhead involved balancing it, how much will the fast card gain?
IMO, and I could be wrong of course, it would seem having card with identical chipsets would be the way to go so the load can be balanced as evenly as possible? Each card does half?
Originally posted by: Fenixgoon
Originally posted by: Amplifier
Oh yeah, and F U ATI. I have the money set aside right now to buy your stupid R520 but I might just go SLI when I need to upgrade...
at least for now.. X850 PE >> dual 6800's U's in SLI, except in a few cases
Originally posted by: cbehnken
Originally posted by: Rollo
Originally posted by: cbehnken
ATI plans to use Tile based rendering, therefore it should be very easy to allocate a correct amount of screen for the slower card to render without slowing anything down.
In all respects so far, AMR looks like a more thought out and arguably better solution than SLI.
Think about what you just said:
You believe AMR will reduce the amount of the screen rendered by the slower card to maintain the speed of the faster card.
OK- where's the benefit then? With load balancing like that, how are you going to increase framerate? If one card is only doing 30% of the work because the other is a a lot faster, and there's overhead involved balancing it, how much will the fast card gain?
IMO, and I could be wrong of course, it would seem having card with identical chipsets would be the way to go so the load can be balanced as evenly as possible? Each card does half?
Have you thought about what you said at all?
If you render 30% of the screen with the slower card then the faster card only has to render 70%.
Less work = higher FPS
Supertiling, in its current implementation, divides one frame of geometry into separate tiles. Unlike AFR where dedicated GPU geometry processing occurs for each frame, supertiling is unable to send ONLY the relevant geometry to each of the pipes. With Supertiling, each GPU is forced to do all of the geometry transform calculations, resulting in no geometry scaling. The result is a less-efficient architecture whereby each GPU is forced to spend extra cycles performing the entire geometry processing.
Originally posted by: ddogg
Originally posted by: Ackmed
Originally posted by: ddogg
how do u know it will be 2X and above the speed of a 6800U. So far all the specs on the R520 is all rumours and u cannot completely judge the performance of the card just by looking at specs.
Wow.. where to start.
Hows about with your own statement a little earlier?
Originally posted by: ddogg
Who knows...ATI might be giving this lame excuse of pushing their AMR marchitecture to cover up the problems their R520 might be having. I guess they did not learn from Nvidia's mistake by jumping on the bandwagon too early. I think Nvidia has got it right this time and from the little news the inq has posted it seems they have a gem in their hands. Only time will tell though!!
So now its ATi "giving an excuse"? When did we see something from ATi? We havent.
Now what the Inq has posted is "news"?
Before you say that the specs and speed of the R520 are just rumors. Then you take what the Inq has posted as fact? Hypocrite much?
i never said what the inquirer posted was fact or rumor, from where did u come up that i said the inq has posted "news"???...and all specs about the R520 that have been going around is just rumor, nothing has been confirmed!! read the fuken post properly before u start spewing crap from ur a$$!!
Originally posted by: cbehnken
Notice they said in its current implementation?
interesting...
I'm glad you would rather throw away your old cards instead of using, even if it's only a 20% gain. Big deal, better than nothing. It has the potential to be much more efficient, only time will tell.
On a side note... Does nvidia's frame splitting interlace the lines sent to each card, so each card does every other line?
Tile based rendering worse than frame based rendering? right... That's why pro solutions use tile based rendering.
Supertiling also has a problem with over-fetching textures (as did scanline interleaving) because of the number of textures that cross the tiling boundaries. This means that each GPU would have to fetch the same texture for neighboring tiles. In SFR, we have one edge where textures can cross boundries. In supertiling, you have many, many more edges and the problem is multiplied linearly with the number of edges. These double-fetched textures can eat up valuable bandwidth.
With Supertiling, each GPU is forced to do all of the geometry transform calculations, resulting in no geometry scaling. The result is a less-efficient architecture whereby each GPU is forced to spend extra cycles performing the entire geometry processing.
Something magical about games that makes a more efficient rendering scheme worse?
Originally posted by: cbehnken
On a side note... Does nvidia's frame splitting interlace the lines sent to each card, so each card does every other line?
Originally posted by: ronnn
Is obvious that ATI is not too happy with the r520, at least at this time. Reading about the new xbox leads me to believe that ATI will have something good by Christmas or so.
But if you want next gen performance now, you will have to wait.... Only the the nvidia salesman/pr guy could suggest next gen performance is buyable now, unless of course he knows something about the g70.
Shoot, maybe I will have to break down and buy a x800xl or 6800gt to tide me over.
Originally posted by: cbehnken
not for games
why is that? Something magical about games that makes a more efficient rendering scheme worse?
That's what i figure . . . r520 is gonna be a dumbed-down version of the xbox360 GPU . . . it's gonna be the NEXT gen that beats up on M$' "baby".Originally posted by: ronnn
Is obvious that ATI is not too happy with the r520, at least at this time. Reading about the new xbox leads me to believe that ATI will have something good by Christmas or so.
But if you want next gen performance now, you will have to wait.... Only the the nvidia salesman/pr guy could suggest next gen performance is buyable now, unless of course he knows something about the g70.
Shoot, maybe I will have to break down and buy a x800xl or 6800gt to tide me over.
Originally posted by: Ackmed
Originally posted by: ddogg
Originally posted by: Ackmed
Originally posted by: ddogg
how do u know it will be 2X and above the speed of a 6800U. So far all the specs on the R520 is all rumours and u cannot completely judge the performance of the card just by looking at specs.
Wow.. where to start.
Hows about with your own statement a little earlier?
Originally posted by: ddogg
Who knows...ATI might be giving this lame excuse of pushing their AMR marchitecture to cover up the problems their R520 might be having. I guess they did not learn from Nvidia's mistake by jumping on the bandwagon too early. I think Nvidia has got it right this time and from the little news the inq has posted it seems they have a gem in their hands. Only time will tell though!!
So now its ATi "giving an excuse"? When did we see something from ATi? We havent.
Now what the Inq has posted is "news"?
Before you say that the specs and speed of the R520 are just rumors. Then you take what the Inq has posted as fact? Hypocrite much?
i never said what the inquirer posted was fact or rumor, from where did u come up that i said the inq has posted "news"???...and all specs about the R520 that have been going around is just rumor, nothing has been confirmed!! read the fuken post properly before u start spewing crap from ur a$$!!
From your quote? Let me bold it for you, maybe then you can understand. Dont get mad at me for pointing out your hypocrisy.
Originally posted by: Rollo
If it's true a 6800U SLI rig is faster than an R520, you can indeed buy "next gen performance now". (at least single card next gen)
People like me who adopted early on SLI will be playing at R520 levels all of 2005, enjoying the awesome performance, maxing eye candy in every game. The R520 faithful are going to spend the year playing at last June's levels, wondering how high they can set the resolution before the slide show begins.
and Rollo, you are gonna be in nVidia "heaven' for another 6 months
(my prediction)
Originally posted by: ddogg
Originally posted by: Rollo
If it's true a 6800U SLI rig is faster than an R520, you can indeed buy "next gen performance now". (at least single card next gen)
People like me who adopted early on SLI will be playing at R520 levels all of 2005, enjoying the awesome performance, maxing eye candy in every game. The R520 faithful are going to spend the year playing at last June's levels, wondering how high they can set the resolution before the slide show begins.
yup...unless they get two R520s and ATI's AMR is as good or better than Nvidia's!!![]()
Originally posted by: Rollo
Apoppin':
and Rollo, you are gonna be in nVidia "heaven' for another 6 months
(my prediction)
What if I get G70 SLI?![]()
Originally posted by: Rollo
Apoppin':
and Rollo, you are gonna be in nVidia "heaven' for another 6 months
(my prediction)
What if I get G70 SLI?![]()
Originally posted by: ddogg
Originally posted by: Rollo
Apoppin':
and Rollo, you are gonna be in nVidia "heaven' for another 6 months
(my prediction)
What if I get G70 SLI?![]()
woah....that would be some serious kick-a$$ performance. around 1.5X+ 6800U in SLI if the G70 is around 2X the power of the 6800U
Rollo, a 50/50 split isn't any more efficient than a 70/30 split in terms of extracting the most performance from each card. (Edit: whether supertiling is less efficient than SFR we have yet to see. But the option to pair an older card with a new one is interesting.) Yeah, two 6800U's will be faster than a 6800U and a 6600GT, but that has nothing to do with efficiency, but rather max theoretical performance. And, while nV's SFR mode can split the screen 50/50, it tends NOT to, as drawing the flat sky isn't as taxing as shading tile floors and the like.Originally posted by: Rollo
I did think about what I said. I was trying to point out to you that two cards doing 50% of the work is more efficient than a 70/30 split. The reduction of work the faster card has to do is less than it would be if it were mated to a similar card.Originally posted by: cbehnken
Have you thought about what you said at all?Originally posted by: Rollo
Think about what you just said:
You believe AMR will reduce the amount of the screen rendered by the slower card to maintain the speed of the faster card.
OK- where's the benefit then? With load balancing like that, how are you going to increase framerate? If one card is only doing 30% of the work because the other is a a lot faster, and there's overhead involved balancing it, how much will the fast card gain?
IMO, and I could be wrong of course, it would seem having card with identical chipsets would be the way to go so the load can be balanced as evenly as possible? Each card does half?
If you render 30% of the screen with the slower card then the faster card only has to render 70%.
Less work = higher FPS
Given that article in the earlier post on tiling, I'd say they're going to need all the help they can get:
Supertiling, in its current implementation, divides one frame of geometry into separate tiles. Unlike AFR where dedicated GPU geometry processing occurs for each frame, supertiling is unable to send ONLY the relevant geometry to each of the pipes. With Supertiling, each GPU is forced to do all of the geometry transform calculations, resulting in no geometry scaling. The result is a less-efficient architecture whereby each GPU is forced to spend extra cycles performing the entire geometry processing.