If you want to get into the server processor racket, here's some advice: Don't bring a knife to a gun fight. And when you whip out your guns, you better have a piece stashed in each of your boots, maybe another high-caliber rifle on your back, and a few knives while you are at it for price-cutting when the bullets run out. …
Having better "256-bit floating point" performance doesn't seem like a big win for the *vast* majority of customers out there. How many languages/compilers/vms support it (none of the mainstream ones that I know of) ?
I am sure the physicists will be pleased, but beyond that tiny constituency most folks won't be able to measure a 100% performance hit in 256 bit fp as they play CoD, browse the web or fiddle about with Excel.
AC - you are correct. For Virtualization workloads, it just doesn't matter at all. The Interlagos strategy is just dead on perfect for that and aimed exclusively at Virtualization. A 256bit FP unit is just wasted silicon on an ESXi server, so AMD getting rid of it and the addtional power consumption is a good strategy. I've got an Opteron 6212 running now, and it is very nice, very SMOOTH for running ESXi, and about to fire up another one. I went with it since it was cheaper by far with a decent serverboard from Newegg (who happened to have it on sale at the time, so I grabbed two). It also has pretty low power consumption compared to what it replaces. My new ESXi server (scratch built) only sucks down 35watts under normal load. The old hardware (PhenomII based) sucked down 75-100watts and didn't run as smoothly for virtualization loads. Now I can lower my cooling and UPS requirements in the wiring closet reducing some bigger expenses there.
For Scientific workloads though, you need the 256FP, so I would stick to MagnyCours or the Intel Xeon equiv with thick cores instead of thin cores. The new Intel chips look pretty nice, and they probably will easily dominate the scientific space if AMD doesn't have anything aimed there anymore. (similar to what happened to Intel years ago when Opterons first came out)
That must have been some good Kool Aid Intel served at their press briefing.
I ought to write a long and scathing flame about the whole decades long now "war" between Intel and AMD blah blah blah. But look just do me a favor and consider yourself flamed. I usually like the stuff you write, but not today.
"Back in the early 2000s, Intel was trying to protect its high-end 64-bit Itanium server business and push its Xeon processors down"
Shouldn't that be "was starting to milk its high-end 64 bit Itanium in the corporate sector, expanding it with FUD, empty promises and dumb hype via press articles against MIPS and Alpha while pushing its overheating, slow-as-hell NetBurst-based crap to the 32-bit unwashed masses"
I still can remember the tremendous propellers on the motherboards.
The author keeps stressing how Intel has more options at various server config levels, but keeps adding "...if you include Itanium". Why on EARTH would anyone do that - it doesn't remotely serve the same space, and won't run the same software licenses. That's like saying "Ford has more compact car types than GM, IF you include their small truck division".