“I think 28nm is going to be a fabulous node,” Moshe Gavrielov, CEO of Xilinx, tells me “a lot of learning went into 28nm to avoid the problems of 40nm. Four flavours, aimed at various markets, will give it tremendous longevity.”
“The yields on 28nm are very good,” says Gavrielov, “much better than they were at 40nm.”
TSMC is ramping its 28nm process three times as fast as the company ramped the 40nm node.
“I’ve been there for the launch of the 65nm and 40nm nodes and have never seen five customers ramp a new technology as they’re doing with 28nm – it’s got three times the momentum of 40nm,” Maria Marced, President of TSMC Europe, tells me.
“The number of tape-outs on 28nm compared to 40nm has been triple at the same stage of production.”
The five 28nm customers are Xilinx, Altera, Nvidia, AMD and Qualcomm.
TSMC is currently installing production tools in its 28nm GigaFab. “The tools will be in Fab 15 by the end of the quarter,” says Marced, “and we’ll start production in the first half of 2012.”
TSMC, as the first foundry to produce on 28nm, feels its gate last manufacturing philosophy has been justified. The IBM club – Samsung, GloFo and the rest – have gone for gate first at 28nm – but have expressed their intention to go gate last on 20nm.
For Xilinx, traditionally a UMC house, it will be busing the same foundry as arch-rival Altera for its leading edge process. Does that mean Xilinx won’t be able to achieve any process advantage?
“They went with HP and LP (TSMC’s high power and low power processes),” replies Gavrielov, “we worked on developing HPL, a derivation of HP with the same performance but lower power. It will broaden our applicability in the market.”
The industry needs a good node, particularly as 28nm could turn out to be a long-lived node. That’s because the next node after that – 20nm – looks like being tricky with double patterning and finfets to master.