NVIDIA GeForce GTX 670M vs NVIDIA GeForce GTX 680M

NVIDIA GeForce GTX 670M

The NVIDIA GeForce GTX 670M is a high-end laptop graphics card introduced in March 2012. It is based on the GF114 Fermi core and is essentially a GeForce GTX 570M with higher core clock rates. Therefore, it is still manufactured in 40nm at TSMC.

GF114 architecture

The GF114 core is a power-optimized version of the GF104 used in the GTX 470M, but with no major architectural changes. Therefore, the performance per MHz stays the same, even though Nvidia was able to clock the shaders higher while still remaining in the same power scale. The GTX 670M offers 336 cores of the 384 in the GF114. More informations on the similar GF104 architecture can be found on the GeForce GTX 485M page.

Performance

Due to the higher clock speed, the GeForce GTX 670M is slightly faster than the old GTX 570M. Since the GPU is a high-end card, the performance is similar to the AMD Radeon HD 6970M. Compared to the GTX 500M cards, the GTX 670M is positioned nearer to the GTX 560M (-15%) than the GTX 580M (+35%). Current games as of 2012 should therefore run fluently in Full HD resolution and high details. Only a few very demanding games, such as Metro 2033 or Crysis 2, may need reduced graphical settings for smoother gameplay.

Features

Similar to the older GF104 chip, the GF114 supports Blu-Ray Bitstream HD Audio output via HDMI. This means it can transfer, for example, Dolby True HD or DTS-HD bitstream-wise without quality loss to a HiFi receiver.

The GTX 670M offers PureVideo HD technology for video decoding. The included Video Processor 4 (VP4) supports feature set C and is able to fully decode MPEG-1, MPEG-2, MPEG-4 Part 2 (MPEG-4 ASP - e.g., DivX or Xvid), VC-1/WMV9, and H.264 (VLD, IDCT, Motion Compensation, and Deblocking). 

Furthermore, the GPU is able to decode two 1080p streams simultaneously (e.g., Blu-Ray Picture-in-Picture).

Through CUDA, OpenCL, and DirectCompute 2.1 support, the GeForce GTX 670M can be used for general calculations. For example, the stream processor can encode videos considerably faster than can many modern CPUs. Additionally, physics calculations can be done by the GPU using PhysX if supported (e.g., in Mafia 2 or Metro 2033).

3D Vision enables the laptop to send 3D content (3D games, 3D Web Streaming, 3D photos, 3D Blu-Rays) to a built-in 3D-enabled screen or an external 3D TV if supported by the laptop manufacturer.

The power consumption of the GeForce GTX 670M is in the same ballpark as the old GTX 470M, which is rumored to be about 75 Watt (TDP including the MXM board and memory). If not under load, the chip is clocked at 50/100 MHz (chip/shader) and 200/400 MHz in 2D and 3D mode, respectively, for power-saving purposes. The 500M series also supports Optimus to automatically switch between the Intel integrated graphics card and the Nvidia GPU. Whether or not a laptop includes this feature is dependent on the manufacturer.

NVIDIA GeForce GTX 680M

The NVIDIA GeForce GTX 680M is a high-end DirectX 11.1-compatible graphics card for laptops and was announced during Computex 2012. It was the fastest GPU available for notebooks at the time of announcement and has a 28nm GK104 core based on the Kepler architecture. Compared to the desktop GeForce GTX 670, the GTX 680M also features 1344 CUDA cores, but the mobile counterpart operates at reduced core and memory clock speeds (720/1800MHz vs 915-980/3004MHz).

Architecture

The Kepler architecture is the successor to the Fermi architecture that first appeared in laptops with the GeForce 400M series. The GK104 Kepler core offers eight shader blocks, called SMX, that are clocked at the same speed as the central core. In the GTX 680M, 7 blocks are active leading to the 1344 CUDA cores. Although more shader cores are available in the Kepler architecture as compared to the Fermi design, the Kepler shaders are still expected to be up to twice as power efficient. However, due to the missing hot clock of the shader domain, two shaders of a Kepler chip are of similar speed to one shader of a Fermi chip (as the latter is clocked twice as fast). PCIe 3.0 is now supported by the mobile Kepler series and an optional Turbo mode can automatically overclock the Nvidia card by a theoretical 15 percent if the laptop cooling system allows it. The implementation of this boost mode is done in the BIOS, but it is ultimately dependent upon the manufacturer of the laptop.

Performance

The graphics performance of the GeForce GTX 680M is on average above the Radeon HD 7970M, thus making the Nvidia card the fastest mobile graphics card as of summer 2012. The GPU has enough power to run demanding games of 2012 fluently with Full HD resolution and maxed out graphical settings. Only the extremely demanding title The Witcher 2 will not run fluently in ultra settings. Battlefield 3, Skyrim, and Crysis 2, for example, are playable at the highest detail settings.

Features

The improved feature set now includes support for up to 4 active displays. Furthermore, high resolution monitors of up to 3840x2160 pixels can now be connected using DisplayPort 1.2 or HDMI 1.4a if available. HD-Audio codecs, such as Dolby TrueHD and DTS-HD, can be transmitted via bitstream mode through the HDMI port. However, as most laptops will feature Optimus, the integrated GPU will likely have direct control over the display ports and may limit the feature set available by the Nvidia Kepler cards.

The 5th generation PureVideo HD video processor (VP5) is also integrated in the GK104 core and offers hardware decoding of HD videos. Common codecs such as MPEG-1/2, MPEG-4 ASP, H.264 and VC1/WMV9 are fully supported up to 4K resolutions while VC1 and MPEG-4 are supported up to 1080p. Two streams can be decoded in parallel for features such as Picture-in-Picture. Another novelty is the inclusion of a dedicated video encoding engine similar to Intel QuickSync that can be accessed by the NVENC API.

The power consumption of the GeForce GTX 680M should be similar to the GTX 580M and therefore best suited for large notebooks (17-inches or greater) and desktop replacements.