Nvidia-Wallaper-widescreen-2

In a recent article by Digitimes, it has been reported that NVIDIA may be announcing their lineup of mobile GPUs based on their all-new Pascal architecture at this year’s Gamescom in Germany on August 17th-21st.

NVIDIA GeForce 10 Series Mobile Specifications

It is reported that notebook manufacturers such as ASUS, Gigabyte and MSI would likely showcase their new gaming notebook lineups, equipped with GeForce GTX 1080, GTX 1070 and GTX 1060 graphics parts. While we don’t know the exact specifications of the GTX 1080 Mobile variant, we do have some information on the GTX 1070 and GTX 1060 mobile chips, which are very similar to the desktop variants. It is also assumed that NVIDIA will be dropping the “M” moniker from their mobile lineup, as the new parts will offer similar performance to the desktop counterparts, just as the MXM version of the GTX 980 did last generation.

MOBILE GeForce GTX 1080 Mobile GeForce GTX 1070 Mobile GeForce GTX 1060 Mobile
GPU GP104 GP104 GP106
GPU Configuration ? 2048 : 128 : 64 1280 : 80 : 48
GPU Clock ? 1443 / 1645 MHz 1405 / 1671 MHz
Mem. Clock / Effective ? 2002 / 8008 MHz 2002 / 8008 MHz
Memory Type ? 8GB GDDR5 6GB GDDR5
Memory Bus ? 256-bit 192-bit
 DESKTOP GeForce GTX 1080 Desktop GeForce GTX 1070 Desktop GeForce GTX 1060 Desktop
GPU GP104-400 GP104-200 GP106-400
GPU Configuration 2560 : 160 : 64 1920 : 120 : 64 1280 : 80 : 48
GPU Clock 1607 / 1733 MHz 1506 / 1683 MHz 1506 / 1709 MHz
Mem. Clock / Effective 1251 / 10008 MHz 2002 / 8008 MHz 2002 / 8008 MHz
Memory Type 8GB GDDR5X 8GB GDDR5 6GB GDDR5
Memory Bus 256-bit 256-bit 192-bit

As we can see in the chart above, the GTX 1070 mobile will actually offer more CUDA cores than the desktop variant, while also reducing the GPU clock frequency.  Theoretically, this should allow for similar performance while reducing power consumption allowing for better thermals and battery life. The GTX 1060 mobile variant is spec’d nearly identical to the desktop part, with the exception of lower clock speeds, which would mean slightly less performance.

Source: Digitimes