PCI Express Graphics Card Overview

The PCI-Express Interface

Peripheral Component Interface Express (PCI-E) is the latest generation of PCI technology of IBM compatible personal computers. Presently, PCI-E is the industry standard in internal computer architecture and structure, effectively obsoleting older Peripheral Component Interface (PCI), Peripheral Component Interface eXtended (PCI-X), and Accelerated Graphics Port (AGP) motherboard bus interface formats.

PCIE is categorically faster in every respect over these older technologies, and provides many performance advantages over standard PCI:

To the user, PCIE remains physically similar in operation to previous architectures - a PCIE compatible motherboard contains a series of slots into which various forms of expansion cards are plugged. These expansion cards can include graphics cards, sound cards, network adapters, or various other forms of circuited hardware that requires a direct connection to the motherboard.

The user may remove or replace any expansion card by popping it out of the motherboard slot, and may plug any PCIE compatible card back into the same slot.

The formatting, construction, and protocol specifications of PCIE (and all previous versions of PCI technology) are developed and maintained by the PCI Special Interest Group (PCI-SIG), an industry-wide trust of nearly 1,000 computer and computer-related companies.

PCI-E Video Card Technical Advances

A single PCIE data stream can move 200 megs of data per second in two directions. PCIE connections are scalable, meaning that a PCIE video card can have a 1x, 4x, 8x, 12x, 16x, or 32x connection to the motherboard; all of these connections are physically different in size. This means that a 32x PCIE connection can move up to 12.8 gigs of information across the motherboard, per second. The magnitude of this increase over previous generations of PCI technology cannot be overstated.

This scalability of the PCIE interface is also sometimes referred to as multi-lane linking. This multi-lane configurability provides two sizeable advantages over previous PCI technology:

1. First, the CPU can activate or deactivate links on the fly, allowing fault tolerance for links that are damaged or working improperly, increasing the overall stability of the computer system.

2. Second, the scalability of the link system allows for an immediate increase in throughput for applications that require large amounts of data transfer in real-time, such as graphics cards or audio devices.

Further technological advances over standard PCI include:

  1. Bottleneck prevention via data prioritization protocols
  2. Real-time data streaming capability
  3. Increased resilience and density of construction materials
  4. Improved error-detection and handling capabilities
  5. More efficient packet disassembly and re-assembly protocols
  6. Dedicated point-to-point interfaces for each device

PCI Express Compatibility

The PCIE architecture is completely backward compatible, meaning that older versions of PCI or PCI-X hardware can be utilized in a PCIE system. Further, these components are individually configured, meaning that the performance of each device card does not affect the overall performance of the system.

This is in contrast to the backwards-compatibility of PCI-X to PCI devices, where the PCI-X system can only run at the speed of the slowest component. This means that plugging a PCI component into a PCI-X motherboard would effectively lower the overall speed of the entire computer system to that of the newly installed PCI component. PCIE does not suffer from this disadvantage.

PCIE devices themselves are not backwards compatible, however - a PCIE device can only be utilized in a computer built around the PCIE architecture, which means that newer PCIE devices will not operate in computers built around older PCI or PCI-X formats.

PCI Express 2.0

The PCI-SIG announced the release of PCIE 2.0 protocols in 2007. PCIE 2.0 effectively doubles the data transfer rates of standard PCIE on a per-link basis, enabling a 32x PCIE device to push 32 gigs of data across the motherboard per second.

Clock speeds are also doubled, from the 2.5 GHz of PCIE version 1.0 to just over 5.0 GHz for version 2.0.

PCIE 2.0 will remain fully backwards and forwards compatible with PCIE 1.0, both at the motherboard and at the device level.

Other improvements of PCIE 2.0 include enhanced data control with point-to-point communications and a more robust and easily upgraded firmware kernel.

PCI Express 3.0

After a series of delays and setbacks, PCIE 3.0 protocols were published by the PCI-SIG in November of 2010. Maximum data transfer rates have been increased to 8 gigs per link, with a theoretical maximum bi-directional data transfer on a 32x PCIE device of 256 gigs per second.

PCIE 3.0 also included instructions for enhanced signaling and data integrity, increasing system stability during these high rates of data transfer.

In this case, the technology of the architecture exceeds the specifications of most available hardware. Before PCIE 3.0 can be fully tested, the density and communications speeds of existing hardware and connective material must be improved to take advantage of, and fully utilize, these new maximum data transfer rates.

In contrast to the old PCI and PCI-X technologies, where the computer components were often outpacing the system architecture, the most recent generation of PCIE technology has left computer components with some catching up to do.