How much did AMD loss from all this what was the original deal worth?

the above mentioned computers could possibly run crysis at decent framerates.

Doesn't Intel still use a memory controller that consumes 20watts engaged or not?

This is not good news for AMD, that is for certain. I am an AMD fan (well more about having some competition so it isn't just an Intel world). After reading so many "articles" here that read more like Blog posts by teenages, it was really refreshing to read a professional, well written article on this subject. Kudos.

FYI Intel has gone a long ways in closing the gap with AMD with the 5400 series Xeon's and chipset. The 1600Mhz dual FSB closes the bandwidth gap, low latency FB DIMM close the latency gap. Combined with high clock speeds (3.2Ghz+), large caches, and SSE4 a 5472 or 5482 easily clobbers the best AMD has to offer in all but the most latency sensitive tightly coupled applications.

These latency sensitive applications is keeping AMD treading water in that many of the high end clusters use esoteric low latency conntections like IB hanging off a HT link for their MPI network - and Intel still doesn't have a real solution to this. PCIe 2.0 gives us the bandwidth, but the latency is still far worse. 

If you take the 5400's performance advantage and add in QuickPath to allow a low latency MPI network and AMD has no hope to compete - even with a Barcelona at nice clocks. And HT3 is not enough to swing it back.

Look at the render farms - they are all jumping to Intel as they are not tightly coupled applications - each frame is rendered seperately from the others - so they are CPU bound and Intel gives them a big boost.

In our render farm a single 5482 outperforms 2 AMD 2222SE's!

Hope this helps

SG

Great read....well written

I don't really think it should surprise us anymore if AMD loses a major contract. AMD has failed to execute properly now for several years. There new core should have been out years ago instead of just arriving. It seems, strictly from an observational point of view, that instead of continuing to innovate and stay ahead, they rested on there laurels and somehow didn't expect Intel to ever catch up. I think everyone can agree that AMD has made some very large mistakes in strategic planning and that they are now paying for it heavily.

It's clear from re-reading Nebojsa's article of 27 Sept last, "AMD Barcelona: HT3 Turn Off?" that Sun is responding to a betrayal of trust on the part of AMD. Their HPC architecture may be such that a hobbled 3xHT2 Barcelona has little or no advantage over a DIB Xeon in the overall scheme. Sun has a contract committment to fulfill and they might just as well institute a platform switch now rather than wait for CSI which may also have teething problems.