NVIDIA has begun the selection process for who will be the fab-centre for its move to 80nm chips, according to a report from
DigiTimes.
The company has already narrowed down the process to Taiwan Semiconductor Manufacturing Company (TSMC) and United Microelectronic Corporation, both Taiwanese producers.
NVIDIA has worked with both companies in the past, but has done most of its work (roughly 90%) with TSMC. In an attempt to receive more of the 80nm work, UMC is rumoured to be preparing to fiercely underbid the project.
The CEO of NVIDIA, Jen-Hsun Huang, has acknowledged that TSMC is normally its closest partner, but that NVIDIA is not ruling out using UMC more.
One should not mistake the 80nm process for the upcoming G80 GPU, whose manufacturing process has not been officially disclosed yet. Usually, these changes are brought into already existing lines, from the lower-end chips to higher end. Higher end parts are more specialised and sell less, so there is less benefit to migration. Therefore, the first confirmed targets are not the G71 GPU, but the upcoming HDMI capable G73-B1 and the lower-end G72.
It is worth noting that these two companies are also the fabs that produce most of ATI's chips, which are also being migrated to 80nm in the same timeframe. One has to wonder if a large contract for either fab company from NVIDIA could short production capacity for ATI, leaving it sorely lacking in stock again.
Got a thought on the newest die-shrink news? Fabricate your own ideas
in our forums.
Want to comment? Please log in.