rwessel (robertwessel@yahoo.com) on 1/5/12 wrote:
---------------------------
>David Kanter (dkanter@realworldtech.com) on 1/5/12 wrote:
>---------------------------
>>George Baker (george_baker@yahoo.com) on 1/4/12 wrote:
>>---------------------------
>>>Intel's announced pricing for dual-socket 8-core SandyBridge-EP ranges from $1106
>>>(1.8 GHz) to $2057 (2.9 GHz) per chip!
>>
>>That sounds about right.
>>
>>>There is little benefit of x86 compatibility
>>>in servers
>>
>>What makes you say that? Have you ever tried to run SQL server on PowerPC? Or VMware?
>>
>>x86 compatibility is fairly useful.
>
>
>Only if you want to run Windows, which is a distinct minority of the server images
>out there. To be sure there is certainly some platform familiarity that's useful
>even if you're running Linux, but for many people it's not >an issue.

It's not just Windows though. It's VMware, etc. etc. and then you have your applications.

>There are millions of hosted web servers alone that could fairly easily be ARM
>instead of x86, so long as the costs work out right.

Right, and that's the 'cloud hosting' market that everyone is going after. The problem is that it's not actually that big. I'd wager around 10% of the x86 server market. And for a surprising number of those workloads, you do care about single threaded performance.

>Unfortunately for ARM, single
>socket x86 systems are pretty cheap. ARM is also a bit >short of virtualization
>support at the moment, but Xen ARM should be available this >year.

Yeah, that's actually the killer, not to mention the differences in I/O and memory bandwidth.

David