"Nforce 680i remains the highest-performing DDR2 chipset for the 65nm Intel Core 2 generation: clock for clock, it had the fastest memory benchmarks, and the best USB, Ethernet and RAID test results"

Don't know exactly what part of the raid you were testing, but as soon as you add more than 2 drives the HDD controller bottlenecks up and you won't get any extra performance.

Over on the SLI Zone forums I am daily giving people advice to either get an intel chipset, dedicated raid card with IOP or toss the idea of raid completely.

Never thought I would argue this way with Inqs view of an NV product, but here we are :)

4 4870x2 would be great for this board as top500 supercomputer last place has about 9Tflops. So this system would be great for scientific stuff if you can get it working full power. Dual socket system with 4 cards shouldn`t cost more than 5K $. 
about 550$ for each card, 550$ for both CPU, as you can use lowest bloomfield, as CPU`s contribute very lil to peak Flops in this case, 500$ more for mainboard, and 700-800$ left for memory, case and other stuff.
So real bargain as cpu cluster of this power costs millions.

I prefer to believe what industry expert and people that test things say about the n200 nonsense, rather than some fool trying to get more manufacturers to use it when they don't even have to, sorry.

Is the Inquirer Getting Soft on us? Either I'm a lot more drunk than I thought or you just posted a (somewhat) pro-Nvidia article. Even the best of us have our weaknesses. I still love you.

I don't think Tylasberg uses QPI to talk to another Tylasberg, but leaves it upto the CPU to route its messages to the other typasberg. I could be wrong, but that's what I remember seeing pictures of.

good riddence to bad rubbish 

their own arrogance brought this sad state of affairs on them 

I will continue to sell them short as they are still highly overrated 

and I always thought their heatbeast chipsets were way to hot

rememberer:

the basic feature of the Nforce200 is splitting one PCIe x16 v2 channel from the North Bridge into two distinct PCIe x16 v2 channels for the graphics cards, with a bit of extra stuff added to speed up peer to peer comms between those two cards without bothering the system chipset. 

Without bothering system chipset, one less crash point. Yeah Sure=G-d in Ancient speak, so it must be HOT. Options? Do you want 8 gpus' or 10 gpus'. Think of rollar coaster with switch tracks & Multi Coasters all rolling simultaneously in Four Parks, interconnected. 
With Ultimate being larger in bit string length, total leaves room for entire parade to fly about effortlessly. Remember these are used to CREATE Final Contents from its known library, Difficult task at best, instantly. Well few 100+ Ns off instant.

Taking this complexity to another level is where numbers will go fanatic, as latency will fall greatly, yet its all there ?right now. At least until dunnington, from banglaru Intel Plant gets teeth in.

At least complexity, therefore potential, are on steep rise.
drashek