$300? Where the hell did you get that number? You can get the 125w 9950 for $185 or the 140w 9950 OEM for $150. That is dirt cheap for a quad core and regardless of what this article says they perform very very well especially for the price, and the 125w actually overclocks fairly well. AMD still makes great chips, they have some excellent server chips!

I will be purchasing a deneb myself.

I agree that this is harsh to put up right before the 45nm release.
Not wanting to sound like an AMD fan boy, but crippling the AMD processor by using the same slow ram as used with the Intel processors is unfair as it doesn't allow the processor to make use of the fact that it has a much higher memory access speed.

I would have been interested to see what the wait states were for the AMD processor as I suspect it would have been FSB IO bound.
IEEE/DARPA will post the honest results on the 20th of November in Austin http://sc08.supercomputing.org/ and I doubt that any of the results will look anything like XBits results. I suspect that the results will be all IBM/AMD in first place and AMD/Sun in second place and IBM 3rd. http://scyourway.nacse.org/conference/view/bof114. There is a big difference between what IEEE calls "toy benchmarks" and legitimate testing where the benchmark cannot be simply loaded into CPU cache and run. The power measurements don't meet the IEEE/EPA EnergyStar5.0, so what are these results worth?
$300? Where the hell did you get that number? You can get the 125w 9950 for $185 or the 140w 9950 OEM for $150. That is dirt cheap for a quad core and regardless of what this article says they perform very very well especially for the price, and the 125w actually overclocks fairly well. AMD still makes great chips, they have some excellent server chips!

I will be purchasing a deneb myself.

I agree that this is harsh to put up right before the 45nm release.
Not wanting to sound like an AMD fan boy, but crippling the AMD processor by using the same slow ram as used with the Intel processors is unfair as it doesn't allow the processor to make use of the fact that it has a much higher memory access speed.

I would have been interested to see what the wait states were for the AMD processor as I suspect it would have been FSB IO bound.
putting this up close before the 45nm release of AMD (so they are on the same level as intel)
is a bit harsh.
IEEE/DARPA will post the honest results on the 20th of November in Austin http://sc08.supercomputing.org/ and I doubt that any of the results will look anything like XBits results. I suspect that the results will be all IBM/AMD in first place and AMD/Sun in second place and IBM 3rd. http://scyourway.nacse.org/conference/view/bof114. There is a big difference between what IEEE calls "toy benchmarks" and legitimate testing where the benchmark cannot be simply loaded into CPU cache and run. The power measurements don't meet the IEEE/EPA EnergyStar5.0, so what are these results worth?
I don't know when US$300 became cheap!!! I suppose at +200 Watts you think they're cheap to operate too.