Globalfoundries slide shows mysterious MCM?

AnarchX

Veteran
attachment.php

Four chips? Maybe two memory pools and two APUs or CPU+APU? Xbox Next's heart Durango?

source: GF presentation @ Computex 2012 attended by PCWatch.jp
 

Attachments

  • gf computex 2012 mcm.jpg
    gf computex 2012 mcm.jpg
    48.7 KB · Views: 626
Whatever it is, looks like an very early prototype of some concept. The small cutout on the lower left corner indicates it's suited for a socket mount, but on the other hand all kinds of ASICs (socketed or direct solder) are prototyped first in a removable packaging anyway. The thing doesn't even have a glue compound to secure the actual dice on the substrate.
 
Whatever it is, looks like an very early prototype of some concept. The small cutout on the lower left corner indicates it's suited for a socket mount, but on the other hand all kinds of ASICs (socketed or direct solder) are prototyped first in a removable packaging anyway. The thing doesn't even have a glue compound to secure the actual dice on the substrate.

fyi, the "glue"/underfill is a reliability mechanism and not a functionality mechanism.
 
So, for engineering samples that aren't going to be shopped around to review sites, it's probably not worth the added cost (however small) of using it. If it is only the engineering team that will be using/testing it, then they are likely to know to be careful with it. Versus a reviewer/consumer who may not mount a heatsink properly or repeatedly mount and remove a heatsink that potentially has some type of adhesive (even heatsink paste is somewhat adhesive) which could potentially break the more fragile non-leaded solder points.

Regards,
SB
 
The underfill helps reduce strain on the silicon due to the different amounts of thermal expansion for the die versus the organic substrate.

Another picture of a 2.5D setup showed that there isn't as much used around dies mounted on an interposer, since both use silicon, although there were pictures showing gobs of it around the interposer.

edit: The better matched coefficients of thermal expansion is a marketing point for 2.5D.

Maybe they left the underfill off to allow better evaluation of the interposer and the components. Perhaps it could have been left off for the purpose of accellerating mechanical wear?
 
Back
Top