Accelerated Graphics Port

AGP dynamically allocates the PC's normal RAM to store the screen image and to support texture mapping, z-buffering and alpha blending[?].

AGP originated from Intel, and the company originally built AGP into a chipset for its Pentium II microprocessor. AGP cards are generally slightly longer than PCI cards.

The first version of AGP, now called AGP 1x, is a 32-bit bus operating at 66 MHz. The resulting maximum data rate for an AGP 1x slot is 266 megabytes per second. In comparison, a standard 32-bit 33 MHz PCI bus (which can be composed of one or more slots) maxes out at 133 MB/s.

As of 2003, newer versions of AGP increase the transfer rate dramatically from two to eight times. Available versions include AGP 2x, AGP 4x, and AGP 8x. In addition, AGP Pro cards of various types exist. They usually require higher voltages and some take up the space of two cards in a standard computer (though they only connect to one AGP slot).

AGP allows for efficient use of frame buffer memory, thereby helping 2D graphics performance as well. In fact, many RAID systems for "headless" (that is, lacking an attached display) servers plug into the empty AGP slot to take advantage of its increased throughput vs. PCI.

AGP provides a coherent memory management design which allows scattered data in system memory to be read in rapid bursts. AGP reduces the overall cost of creating high-end graphics subsystems by using existing system memory.

AGP is expected to be phased out by around 2005, since Intel have indicated that their future chipsets (to be introduced at about the time of their "Tejas" CPUs) will have AGP support removed, and replaced by PCI-Express. Nvidia's upcoming NV40 GPU will not support AGP in any way, and ATI's upcoming R420 GPU will only support AGP via additional circuitry.

This article was originally based on material from FOLDOC, used with permission. Update as needed.

External Links

• http://www.gcsextreme.com/agpfaq.htm