Nvidia has announced its lineup of Kepler mobile GPUs, though some appear to be renamed versions of previous generation Fermi hardware.
The two high end chips, GTX 675M and 670M are both made using the two year old 40nm process, though they all have the same features as the new Kepler architecture. However, when compared with the most powerful new chip design, the 660M, there are a few differences. The Kepler's processor clock reaches an impressive 835Mhz, while its older brothers are 620Mhz and 598Mhz respectively.
Fermi dominates the low end as well, with the 635M and 40nm version of the 640M LE both using the older architecture. In that region though, you'll be wanting to stick with the Kepler chip since its performance should be comparable with the Fermi, while vastly superior in terms of power savings. Featuring 384 processing cores vs the Fermi's 96, the 28nm 640M LE version is certainly more impressive - even if its clock speed is almost 50 per cent less.
Nvidia is looking to get some of these new mobile chips in Intel's Ultrabook series of notebooks, emphasising the low end Kepler parts as ideal partners for the hardware. Supporting DirectX 11, 2GB of memory and all the usual Nvidia features, these mobile GPUs would allow an Ultrabook to play even quite graphically intense games at a reasonable frame rate.
No sign of the expected GTX 680M yet. Nvidia is expected to unveil that chip later in the year.
For a full rundown of the range, head over to Hexus.