ATi claims 174.6 GB/s for Radeon 9700 Pro

[Nebuchadnezzar]

http://www.theinquirer.net/?article=6424

It had to happen some day !

 

[Evildeus]

Will R300 beat N30 per 4 times?

 

[Psikotiko]

.....and S3 trio is a SUPER-3D-CARD

 

[Teasy]

But ATI didn't claim 174.6gb/s for Radeon 9700 Pro. They just mentioned that they have 8.8:1 compression and the inquirer took that and came up with the 174.6gb/s number.

There's a big difference between using a compression scheme to claim a theoretical bandwidth for a card and just mentioning that you have compression.

 

[BoddoZerg]

Yeah well assuming infinite overdraw, the KyroII has infinity GB/s! Haha beat that! *pthbbpt*

 

[alexsok]

Absolutely useless numbers!

 

[jjayb]

No more useless than Nvidia's "effective bandwidth" numbers.

 

[Nappe1]

Absolutely useless numbers!

as well as the GF FX having 48GB/s bandwidth?
Alexsok, be careful now on your answer. you are definately now standing on line that makes a difference between some company fan and pure hardware enthuastic. it's up to you which side you want to step.

 

[Doomtrooper]

Amen...lets not talk about memory interface numbers when one card is still running on a 128-bit bus...in the end the 256 bit bus will win in most scenarios..again where it counts FSAA and Anisotropic.

 

[alexsok]

as well as the GF FX having 48GB/s bandwidth?
Alexsok, be careful now on your answer. you are definately now standing on line that makes a difference between some company fan and pure hardware enthuastic. it's up to you which side you want to step.

I wasn't talking of ATI specifically, I was talking about both ATI and NVIDIA, providing these useless numbers.

 

[K.I.L.E.R]

My S3 Virge has 200Tb/sec bandwidth.
Beat that Nv and Ati.

 

[SteveG]

No more useless than Nvidia's "effective bandwidth" numbers.

Depends. If NV30 outperforms R300 in bandwidth-dependent tasks, with less *physical* bandwidth, then they will have earned the right to speak of "effective bandwidth". Of course that's a big if.

 

[Bambers]

Effective bandwidth will be marketing BS whether the nv30 is faster in bandwidth limited situations or not, they will be able to rightly say that thier card is more efficient however...

 

[no_way]

Depends. If NV30 outperforms R300 in bandwidth-dependent tasks, with less *physical* bandwidth, then they will have earned the right to speak of "effective bandwidth". Of course that's a big if.

Im sure both NV and ATI can come up with "bandwidth-dependent benchmarks" that will clearly illustrate that their bandwith is bigger than competitors

 

[pxc]

Fuad is retarded. He always makes these kinds of dumb conclusions. I'm positive he's at theinquirer as a joke and/or to attract flames from the slightly less accurate other writers.

 

[Tahir2]

Fud, sorry Faud is the inquirers best asset after the main man magee..

 

[KimB]

Amen...lets not talk about memory interface numbers when one card is still running on a 128-bit bus...in the end the 256 bit bus will win in most scenarios..again where it counts FSAA and Anisotropic.

Just like the Parhelia trounced all over the GeForce4 Ti 4600, right?

The truth is, there's always a more efficient architecture. nVidia's GeForce3/4 architecture was quite a bit more efficient than ATI's Radeon 8500. We'll have to see if the same occurs again. I think it's going to be very interesting.

Oh, and anisotropic filtering doesn't take memory bandwidth, it takes fillrate. So, the GeForce FX should do better in anisotropic filtering performance.

 

[crystalcube]

Oh, and anisotropic filtering doesn't take memory bandwidth, it takes fillrate. So, the GeForce FX should do better in anisotropic filtering performance.

Isn't Anisotropic filtering performance dependant quite a lot on the Anisotropic filtering implementation that is why Nvidia cards take higher hit than ATI cards when anisotropic filtering is enabled, even though they have similar fillrate.

 

[PrzemKo]

Oh, and anisotropic filtering doesn't take memory bandwidth, it takes fillrate. So, the GeForce FX should do better in anisotropic filtering performance.

Taking into account that 8 pipes @ 500 MHz of GFFX you have 4 Gpix/s = 16 GB/s at 32 bits/pixel. Add to that Z-Buffer and you'll see, that memory bandwidth could be entirely consumed by NV30's fillrate alone (sans compression and optimizing techniques). Aniso WILL eat bandwidth.

 

[alexsok]

Fuad is retarded. He always makes these kinds of dumb conclusions. I'm positive he's at theinquirer as a joke and/or to attract flames from the slightly less accurate other writers.

Wasn't a similar comment made by ATI to Tomshardware on Comdex?