Is this VIA's nForce?

[IntelConvert]

Over at amdmb, I just read about VIA's new KM266. Could this be VIA's best shot at a video-integrated chipset to compete with nForce?

 

[Athlon4all]

Lol. Savage4 using SDRAM vs. a MX with DDR. I don't think so. VIA's real attempt to fight nForce will be KM333 which will feature the new ZoeTrope(Or Paramount dunno what it's called) graphics core that like the SiS 315, is rumored to be similar to MX200 speeds. I really think though, the problem with all Integrated graphichs chipsets, is not the core itself. The issue is memory bandwidth, and neither VIA nor SiS has shown any idication that they plan on doing a Dual Channel DDR chipset. QBM could make up for the difference, but things are uncertain about it. The memory bandwidth issue is also why I doubt we will see a GF3 nForce. The Memory Bandwidth requirements of the GF3 are so tremendous, that it will end up far off behind the real GF3. What is a possibility in the next nForce is more like GF2 GTS speeds. If you were to keep the Athlon's fsb at 266 (using only 2.1GB/ps of bandwidth) and then couple the Twinbank architechure with PC2700 DDR, that will result in given the integrated video 3.3GB/ps of Memory Bandwidth, that will definately put the nForce video above MX400 speeds, but would prolly still trail the GTS by a good 20% (a GTS has 5.3GB/ps). So, will VIA or SiS have any hope of competeing with nForce's Twinbank architechure? I don't think so.

 

[IntelConvert]

Athlon4all: As you may summize by my 'tongue in cheek' emotion (above), I fully agree with you.

 

[AA0]

VIA hasn't touched much integrated video on the AMD side, probably because nobody really wants it. They have intel chipsets with video though.

 

[Funky303]

<< On industry standard benchmarks the VIA ProSavageDDR KM266 enables performance gains of up to 30% when compared with previous generation VIA ProSavage KM133 systems in the most commonly used applications >>



well this does not really is a PRO VIA statemant. changing from a 133 SDRam to 266DDR, also increasing to a theoretical AGP8x memory Throughput, they should at least get some better results.

Compared to the nForce integrated VGA, VIA also does not have anything like the TwinBank Memory controller, so even a fast Memory (here told to be 8xAGP) has the problem of having to share MemoryAcces with the system.



<< effective AGP 8X equivalent internal bandwidth throughput through 128-bit data paths, 32MB of Frame Buffer capability >>



Nvidia integrated VGA was somewhere compared to 6x AGP.

But lets wait & see until someone buys that kinda board & runs some Benchmarks.