[Info-vax] Comment on the future of OpenVMS

[John Wallace]

On Oct 26, 10:17 am, Neil Rieck <n.ri... at sympatico.ca> wrote:
> On Oct 25, 4:54 pm, JF Mezei <jfmezei.spam... at vaxination.ca> wrote:
>
> > Neil Rieck wrote:
> > > 5) Itanium blades will soon appear with 32 and 64 cores
>
> > Does this mean 64 cores on one motherboard that slides into a single
> > slot in a blade cabinet, or does this mean that a blade cabinet will be
> > able to house 32 slots, each with a single dual core IA64 cpu ? (or
> > whatever combination of cpu/cores and blade slots).
>
> I "think" they were talking about a future version of BL860c running
> Tukwila. Since that card currently contains two CPU sockets, and
> Tukwila is a quad-core, then we are talking about 8-cores per blade.
>
> http://h18000.www1.hp.com/products/servers/integrity-bl/c-class/860c/...
>
> These cards could then be inserted into either c3000 or c7000
> enclosures so that the total core count is probably higher than what
> most companies could afford to license. I was more interested in the
> fact that both XEON and Itanium blades would use a common chip-set.
> Not sure if this already exists in the current product or is coming in
> the next iteration but anything that will drop the cost is welcome.
>
> NSR

"I was more interested in the fact that both XEON and Itanium blades
would use a common chip-set."

This story started with the Common System Interconnect (now Quickpath)
vision; folks concluded that it meant a common socket and all the
engineering manufacturing and support economies that might achieve.
Turns out that a common socket isn't on the cards just yet; I'll leave
you to work out why.