• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

There's something very interesting about the desktop Kabini's...

NTMBK

Lifer
And it's in this photo. 5 points to the person who gets it.

Socket-AM1-AMD-Athlon-5150-and-5350-Quad-Core-CPUs-Incoming-433766-2.jpg
 
Ding ding ding.

They've ported Kabini from TMSC to GloFo.

So Kumar's words still stand, they will backport every one of their CPU to Globalfoundries.

I wonder when someone will bench the two Kabinis side by side to compare the two processes.
 
And now they have the full set of GCN1.1, GCN1.0 and Jaguar blocks all on GloFo. Console APUs next?

It would certainly make sense to meet their wafer commitments. The other half of the question being what are their wafer commitments to TSMC?
 
Can we speculate they have something waiting for TSMC, that will meet with high demand and they are making more room for it?
Possibly on 20nm, possibly wrong forum? 😉
 
Can we speculate they have something waiting for TSMC, that will meet with high demand and they are making more room for it?
Possibly on 20nm, possibly wrong forum? 😉

No. Also TSMC 28nm utilization is low for that matter.

Its simply due to the WSA. The big question is when they will be forced to move GPUs to GloFo as well.
 
Can we speculate they have something waiting for TSMC, that will meet with high demand and they are making more room for it?
Possibly on 20nm, possibly wrong forum? 😉

I believe that wafer allocations are done on a per-process basis- i.e. a 20nm product would not take up 28nm allocation. (Makes sense, as 20nm is run on a totally separate production line.) I guess the entire GPU line, plus both console APUs, plus Kabini and Temash was too much to fit into their allocation.
 
Yes. Utilization was at 65-70% in Q4 2013. And is first to be normalized in Q2 2014.

Mainly due to high inventory at customers.

But would this sort of short term fluctuation affect wafer allocations? I doubt it- if they've already negotiated to get company A 4mil wafers and company B 3mil wafers, they can't just shift 1mil wafers from A to B, even if A is using only half their allocation and B is maxing out their allocation.
 
so this could be the reason AMD claims beema/mullins will be more efficient.

To get the kind of numbers AMD was claiming I think they'd need to have used their algorithmic high density libraries on Kabini -> Beema/Mullins. I think it would be cost and time efficient to combine getting production up at Glofo with implementing those changes. Guess we'll know when they show up for AM1 drop in upgrades.
 
Back
Top