IDC/Gartner projections are really silly
Look at the projections of the late 1990's
see what they projected for Itanium
Its really just a consensus of what Intel/AMD/HP/IBM/Dell think with maybe some input from large customers
and the type of weed they are smoking that day

There are two ways to make projections
1. What do people need
2. What can people afford

there is a strong argument that two or even one quad core is sufficient for most peoples needs
but the fact is this business has never been about what people need, and more about what they can afford

since the other components of deploying a line of business app, personnel, software, storage, is not dropping
there is no need for an 2X reduction in platform cost

the two socket quad core has about the same performance as a four socket dual core for half the price,
yet there has been no sharp drop in 4-way volume

to make a reasonable estimate of market size for an 8-way,
i suggest looking at the volume of high-end 2-way compare to 4-way, ie, throw out the low-end/mid-range 2-way
if this is 4:1, say 800K high 2 to 200K 4-way,
then throw the applications that scale out not up
a reasonable estimate for 8-way might be 10-20K
a good chunk of this will probably be Itanium

alot depends on price
a system with 4 sockets populated + 16G now runs around $20K, if the 8-way comes around $40-50K, instead of $80-100K, it culd fly

Michael S (already5chosen@yahoo.com) on 5/17/07 wrote:
---------------------------
>Doug Siebert (foo@bar.bar) on 5/16/07 wrote:
>---------------------------
>>Great article, David!
>>
>>I'd personally be pretty surprised if 8 socket servers don't become commoditized
>>to the extent 4 socket servers are today in a few years.
>>
>
>Are you theorizing or do you really have first hand experience with these wonderful commodity 4-socket servers?
>I personally never even see one, but I'm not in the IT.
>
>I'd like to see IDC or Gartner report for the H1 2007. According to my own pet
>theory we are going to see a sharp decline in 4-way server sales.