As others have pointed out, it's giga-transfers per second. Simplistically, each lane can therefore operated at 8 billion bits per second, or 1 giga-byte per second. Different cards will use one or more lanes. High end video cards often use 16 lanes, so this spec would allow 16 GBytes of data transfer per second to the video card. This is enough data to refresh a 1080p display 1,900 times per second at 32 bits per pixel, although what video gamers are really after is very low latency (the time between them taking an action, and the action appearing on the display) so at this point the race isn't necessarily for more data, but faster bursts of data to lower overall latency. A single frame can be transfered raw in about 1/2 a millisecond. This isn't what happens, though - the video card actually gets textures, a 3D description, and other data, and is expected to render that data very quickly and then display it. Given the quality of the textures and other data being transferred it may occasionally take quite a bit longer to give the video card everything it needs in order to render a given scene, and so having a faster bus both lowers latency, and increases the amount of data the card can recieve in a given period of time. -Adam On Tue, Feb 16, 2010 at 4:17 PM, YES NOPE9 wrote: > I read an article about PCI 3.0.... i did not see a definition of GT/ > sec. > What does it mean ? > Gus > -- > http://www.piclist.com PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist