Altera is looking to three tricks to take full advantage of the move to a 28nm process which is expected to become commercially available next year.
With 28nm unable to offer performance or power advantages over 40nm, Altera will use the extra density advantage of 28nm to introduce Hardcopy blocks and 28Gbit/s transceivers to its FPGAs.
The introduction of Hardcopy blocks onto as much as 20% of the die will increase performance by up 
to 20 times.
A third trick being introduced at 28nm is partial configuration - the ability to reconfigure parts of a chip while other parts keep running.
This is a way of further increasing effective density by storing functions off-chip, so reducing chip size and permitting more applications.
To achieve partial reconfigurability, Altera has been: "Leveraging our internal compilation technology which allows customers doing designs 
to change one block without changing the whole design", said David Greenfield, Altera's senior director for Hardcopy, "we've been working on it for over a decade."
The three tricks will allow customers to implement 400G systems which people will want for high definition video processing, while keeping within the constraints imposed by power and cost.
Without the tricks, normal Moore's Law evolution would not get Altera's FPGAs to the 400G target.
The reason for using Hardcopy blocks is because they only have four masking layers as opposed to the multiple masking layers of hard macros.
Altera expects to have first, prototype, versions of its 28nm chips at the end of this year. Greenfield said that it has run four test chips on 28nm and the fifth is taping out. In all, Altera will do seven test chips covering different parts of the chip before running a final, complete, prototype chip.
The innovations at 28nm take Altera closer towards becoming an ASSP company. "There's a great opportunity for alignment with ASSP companies with this technology", said Greenfeld.
Asked what he meant by alignment, Greenfield replied: "There's the ability for an ASSP company to leverage our technology . They could private label it, it's an IP opportunity. It would introduce complexity."
Asked if Altera would buy an ASSP company, Greenfield replied: "There aren't a huge amount of healthy ASSP companies. It's not a good way to go for us. The further we get from our basic business model the more challenging it is for us."
However with the ASIC/ASSP market worth $76bn, and the proportion of $76bn which is accessible to programmables being about $28bn, the temptation to become more ASSPy and more ASICy to break out of programmables' decade-long stagnation in a $3bn niche must be strong.
Altera has not seen significant problems with yield on TSMC's 40nm process. Although some GPU manufacturers have complained of poor yields on 40nm at TSMC, Greenfield said: "We haven't seen the yield challenges other early adopters have encountered. It's not perfect but out long relationship with TSMC has made it smoother for Altera than for others."
Altera uses the same high performance 'G' version of TSMC's 40nm process as the GPU manufacturers.
"Now", added Greenfield, "it's getting much better for everybody."