As far as we know, Nvidia uses either the 3 to 1, or 4 to 1 compression method which gives 48GB/s in the best case.
So: 500MHz x 2 (DDR) x 128bit (16bytes) x 3X compression = 48 GB/s
ATI said it is using 8.8 to 1 compression method on the Radeon 9700 PRO cards that can deliver an astonishing 174.2GB per second in the best case. A rare bird in practice, we'd suggest.
As in: 310MHz x 2 (DDR) x 256bit (32 bytes) x 8.8 compression rate = 174.2 GB per second
It's a theoretical figure and certainly does not mean that ATI Radeon 9700 PRO will be three times faster than Geforce FX, but does give an idea of the effective bandwidth of the card.
Real life bandwidth depends what kind of data used, and varies all the time. The real numbers are probably in the median range.
ATI still wants to focus more on actual numbers such as the 19.8 GB/s that they are getting with their Radeon 9700 PRO rather than these theoretical ones.
Surely all of this demonstrates the power of marchitecture, and how in the head-to-head battle for market share, the Fudzilla Monster still growls loudest of all.
Originally posted by: ElFenix
yay contrived numbers
Surely a yottabyte is 8x10^24 zeros 😉Originally posted by: Shalmanese
I have a file here, it is composed of 10^21 zeros. Now, create that file on your computer!
Wow, I just managed to send a yottabyte of data over the internet (theoretical bandwidth). I r0x0r!
Originally posted by: LikeLinus
<<Is ATI getting nervous here?>>
About what? A card that is 6 months late and wont even see good production numbers till spring? ATI Will have something bigger and better out by then. They've already demo'd a DDRII card and have it tapped out. The real reason that Nvidia is so late is they went back to the drawing board after the 9700 hit production and Nvidia realized it would lose this race. So now they've backtracked and are looking at an 2003 release. I don't think ATI is scared
Originally posted by: GTaudiophile
Originally posted by: LikeLinus
<<Is ATI getting nervous here?>>
About what? A card that is 6 months late and wont even see good production numbers till spring? ATI Will have something bigger and better out by then. They've already demo'd a DDRII card and have it tapped out. The real reason that Nvidia is so late is they went back to the drawing board after the 9700 hit production and Nvidia realized it would lose this race. So now they've backtracked and are looking at an 2003 release. I don't think ATI is scared
No, nVidia is late because of TSMC...or so they say 😉
Originally posted by: LikeLinus
They blame TSMC, but It's also been noted that they did change a few things when they found out what ATI had up their sleave. XBox had nothing to do with it, those chips are easily produced and have been around for over a year.
Originally posted by: Snatchface
We shall see. If the Radeon 9700 Pro is 3 times faster than the GeForce FX (HAR HAR) then that number actually means something. Is ATI getting nervous here?
Originally posted by: Snatchface
We shall see. If the Radeon 9700 Pro is 3 times faster than the GeForce FX (HAR HAR) then that number actually means something. Is ATI getting nervous here?
Yes, I realize the logic is flawed...that's why I'm satirizing it.By that "logic", the GeForce FX ought to be about 2.5x faster than the 9700 Pro, since it has a bit over 2.5x times as much "bandwidth" as the 9700 Pro does (using NVIDIA's made up numbers vs. ATI's real numbers)
Originally posted by: Snatchface
Yes, I realize the logic is flawed...that's why I'm satirizing it.By that "logic", the GeForce FX ought to be about 2.5x faster than the 9700 Pro, since it has a bit over 2.5x times as much "bandwidth" as the 9700 Pro does (using NVIDIA's made up numbers vs. ATI's real numbers)
Originally posted by: BFG10K
The way I see it, if one company starts posting theoretical nonsense then the other company is just as entitled to do so.
And unfortunately in this case nVidia is the one who started it and it's probably because the FX's memory bandwidth is no match for the 9700 Pro's and they know that all too well.
Originally posted by: sharkeeper
Does this mean if enable disk compression on one of my raid arrays that I can get almost 1GB per sec? 🙂