In a move similar to the one made by Taiwan Semiconductor Manufacturing Company, Globalfoundries has reportedly announced that it has decided to give up on the 32nm bulk manufacturing process technology with HKMG (high-k metal gate) and, instead, move directly to the 28nm.
The main reason given was supposedly weak demand for products based on the process, as most companies currently using 40/45nm are jumping directly to the smaller one. Thus, the remaining 32nm process is no more, which makes it clear that NVIDIA and AMD will be revising their GPU family plans, if they haven't already done so.
The 28nm generation will be offered on bulk silicon substrates and will have the smallest SRAM cell size reported in the foundry industry, namely 0.120µm². Also, the process will use the “Gate First” approach, which, for customers that still use conventional poly/SiON-based technology at the 45/40nm and 32nm nodes, simplifies 28nm design implementation and IP re-use. Not only that but the “gate First” approach has an advantage over the “Gate Last” when it comes to die size.
“All of our efforts around next-gen graphics and wireless are focused on 28nm with HKMG and we no longer have a 32nm bulk process. We removed this off our roadmap due to lack of customer demand as most are making the jump from 40/45nm right to 28nm,” said Jon Carvill, the head of public relations at Globalfoundries.
The 28nm node will be available in two variants, namely 28nm-SLP and 28nm-HP. The former stands for 28nm Super Low Power and is mostly intended for wireless mobile applications, where a long battery life is crucial. The latter, 28nm High performance, is expected to be used in the next generation of graphics products, media encoding, networking, storage and game consoles. Production is planned for the second half of 2010.