Electronics Weekly Electronics Design & Components Tech News
GloFo says it says its 32nm yields are now OK. “The 32nm yield challenges are well behind us,” says the company, adding that it has shipped 250k 32nm wafers.
Naturally, I asked GloFo how many working die came off those 250k wafers and, equally naturally, GloFo replied: “Well, of course I can’t give you that number.”
What GloFo does claim is that the cumulative die shipments of the first five months of 32nm production were more than double the cumulative die shipments of the first five months of 45nm shipments.
This is not a particularly stunning accolade. 45nm was a pig of a node and had all sorts of teething troubles.
“Early in 2011 we met significant challenges in early yield learning on 32nm HKMG,” acknowledges GloFo CEO Ajit Manocha.
“However,” adds Manocha, “we made several organisational and operational changes in the second half of the year that led to a dramatic increase in production velocity and major breakthroughs in yield learning.
With becoming modesty, Manocha does not add that one of the ‘organisational changes’ was his appointment as interim CEO in June 2011 and as permanent CEO in October.
It would be nice do know what kind of yield GloFo sees as marking ‘an end to 32nm yield challenges.’
90% is what fabs aim for at the maturity of a process node, though Intel is said to be content with 80%.
Maybe GloFo could tell us the yield to which it aspires.
It does seem GF are getting good yields at last at 32nm, and indeed TSMC for some customers on 28nm. I always monitor the availability of high end gaming equipment – graphics cards, processors, etc. As soon as they are in good supply you know the yield is at least acceptable.