VIA KT600 Surpasing The nFORCE2400Ultra In Early Benchmarks

[David G.]

Hello everybody ,


This is one subject that I would really like to see it being discussed more often but , on every site I visit this seems to be ignored .

VIA has surpassed all my expectations and proved it is the biggest and the best chipset designer out there ... proably excepting Intel .

[H] has made some tests on the early reference KT600 mobo puting it against the new nFORCE 2 400 Ultra .

Well , this is quite amaizing since it shows a chipset using 3.2 Gb/s of bandwidth , getting better results than a chipset using 6.4 Gb/s .

That's only a 5% increase but still it's impressive comming from an early reference mobo pitted agains a retail , dual chanel , 3rd revision mainboard .

WHat do you think ?

Here's a link

Will VIA gain back it's lost market share?

 

[mat]

Re: VIA KT600 Surpasing The nFORCE2400Ultra In Early Benchma

Well , this is quite amaizing since it shows a chipset using 3.2 Gb/s of bandwidth , getting better results than a chipset using 6.4 Gb/s .

Not really. Since Athlons FSB can only transfer 3.2 GB/s (or whatever their new 200 MHz FSB can do), its quite a waste of bandwith to have twice as much

 

[David G.]

Re: VIA KT600 Surpasing The nFORCE2400Ultra In Early Benchma

Well , this is quite amaizing since it shows a chipset using 3.2 Gb/s of bandwidth , getting better results than a chipset using 6.4 Gb/s .

Not really. Since Athlons FSB can only transfer 3.2 GB/s (or whatever their new 200 MHz FSB can do), its quite a waste of bandwith to have twice as much

Yes , in a way it's quite like that . The same thing happened whe DDR333 support apeared in KT333 with all the AThlon CPUs using only 2.1 Gb/s .
But , as with nFORCE's 1&2 case we can see that the extra bandwidth helps a little . More is never worse and nVIDIA has been capitalising on that .

 

[mat]

most likely the nvidia Memory Controller has just lower latancy (or whatever) than Via's Solution.
I really doubt that the 3.2 GB/s are used by something else in the system. (OK you could send the data to the GPU, but i doubt this needs the other 3.2 GB/sec)

I think somewhere at aceshardware.com was an Artikel about memory controllers (i think they compared Via and Intel and showed that intel has a much lower lantency than the Via Chip).