The launch and roll-out of graphics processing units (GPUs) with Fermi architecture has been rather hard and slow for Nvidia Corp. The company had to delay the flagship chip and never released its fully-fledged version, the company's mainstream parts came very late and entry-level GPUs are still not rolled-out. In the future Nvidia wants to avoid not only postponements, but also to speed up ramping up different members of new product families.

Jen-Hsun Huang, the chief executive officer of Nvidia, said at a press conference during GPU Technology Conference (GTC) that the company would ramp up top-to-bottom family of graphics chips with new architectures in three months time, reports Tech Report web-site. The promise seem to be rather bold, given the fact that the company has been failing to refresh the whole family in a rapid manner for years now and had to rename older products in order to keep demand high.

The introductions of major Nvidia GPU architectures, such as Kepler and Maxwell, will be aligned with availability of new process technologies, which occur every 18 - 24 months. However, the company also plans to update its lineups with enhanced architectures in between major launches. As a result, next year the company plans to introduce new products based on improved versions of Fermi architecture (e.g., Fermi+).

As previously reported, Nvidia plans to release its next-generation architecture code-named Kepler in late 2011. The next major update to the company's GPU design will arrive in 2013 in the form of code-named Maxwell architecture.

Tags: Nvidia, Fermi, Kepler, Maxwell, Geforce, Quadro, Tesla