Reading some of the analysts across the spectrum and the consensus is this is just another splitting of the SKU's to meet some niche market. People who want workstation like performance but don't want the high end socket requirements that comes from Xeon Gold/Silver/Platinum/Latinum SKU's. Xeon-D will continue to be marketed to the datacenter crowd (like Facebook) who like the lower power profile to keep ARM vendors at bay.
OEM's used to deal with this a few years ago by dropping a desktop CPU in a board with a high end GPU and some rudimentary RAID and call it a "workstation". When people saw through this, they (Dell, Lenovo) then rolled out CTO build to order SKU's to resellers where they could drop what ever CPU/GPU combo the customer wanted (Always server class Xeon).
I think what is getting the goat of the technically minded, is that Intel is using some common pre-existing socket types, but forcing OEM's to whitelist to keep end users from cross-pollinating their CPU's. While biz users usually buy by system and CPU generation, HEDT/Power/Workstation class users like to extend their value a bit more. Intel strategy here is to try to get something to offer them and keep them away from server class Xeon's for what they consider desktop work and suppress any arbitrage by letting them buy low and upgrade later high. If you want high capability then Intel thinks you should act like large corporate and purchase in whole.
I still remember the kurfuffle when HP was about to release a BIOS update that allowed Ivy Bridge Xeons to run on Sandy Bridge planars. Intel put the kabosh on it. That was the beginning of what we see now.
Honestly I am not sure why Intel just bans sockets completely at the low end, and leave it at the high end server level where the margins are better.
OEM's used to deal with this a few years ago by dropping a desktop CPU in a board with a high end GPU and some rudimentary RAID and call it a "workstation". When people saw through this, they (Dell, Lenovo) then rolled out CTO build to order SKU's to resellers where they could drop what ever CPU/GPU combo the customer wanted (Always server class Xeon).
I think what is getting the goat of the technically minded, is that Intel is using some common pre-existing socket types, but forcing OEM's to whitelist to keep end users from cross-pollinating their CPU's. While biz users usually buy by system and CPU generation, HEDT/Power/Workstation class users like to extend their value a bit more. Intel strategy here is to try to get something to offer them and keep them away from server class Xeon's for what they consider desktop work and suppress any arbitrage by letting them buy low and upgrade later high. If you want high capability then Intel thinks you should act like large corporate and purchase in whole.
I still remember the kurfuffle when HP was about to release a BIOS update that allowed Ivy Bridge Xeons to run on Sandy Bridge planars. Intel put the kabosh on it. That was the beginning of what we see now.
Honestly I am not sure why Intel just bans sockets completely at the low end, and leave it at the high end server level where the margins are better.
Comment