Why PCI Express* Architecture for Graphics?
White Paper Introduction
The evolution of PC graphics interconnects has been on a trend where graphics bandwidth has increased rapidly throughout the years. In the 90's the industry has seen a trend where graphics BW has doubled roughly every 2 years to keep pace with new applications and support future requirements. In the mid to late 90’s the Accelerated Graphics Port (AGP) was introduced. This was driven primarily by 3D applications and the need for more bandwidth and faster processing for graphics. Through the evolutions of AGP1x - 2x - 4x and 8x we’ve seen the bandwidth double each time. Now there is another inflection point.
New applications are evolving using photo-realism drive BW demands. The Digital Home and next generation GUIs are rapidly evolving and will be driving bigger bandwidth demands for the future. However, continuing on an evolutionary path with AGP to keep up with current demands and future requirements, one which is a parallel point-to-point interconnect using source synchronous timing, would pose serious limitations. The existing connector would have to be redesigned to accommodate lower inductance and capacitance at higher signaling rates. The timing budget would be significantly reduced with very little margin. An evolutionary approach to AGP is simply not feasible to support future demands. What is needed is a new technology that supports both near- and long-term platform requirements.
PCI Express* is a 2.5 Giga transfers/second serial differential point-to-point high speed interconnect with added flexibility and scalability that is intended to replace AGP and PCI over time. PCI Express is this new technology for graphics, providing enough bandwidth and performance for the next decade.
Read the full Why PCI Express* Architecture for Graphics Paper.
165KB
很抱歉，此 PDF 仅供下载
Why PCI Express* Architecture for Graphics?
White Paper Introduction
The evolution of PC graphics interconnects has been on a trend where graphics bandwidth has increased rapidly throughout the years. In the 90's the industry has seen a trend where graphics BW has doubled roughly every 2 years to keep pace with new applications and support future requirements. In the mid to late 90’s the Accelerated Graphics Port (AGP) was introduced. This was driven primarily by 3D applications and the need for more bandwidth and faster processing for graphics. Through the evolutions of AGP1x - 2x - 4x and 8x we’ve seen the bandwidth double each time. Now there is another inflection point.
New applications are evolving using photo-realism drive BW demands. The Digital Home and next generation GUIs are rapidly evolving and will be driving bigger bandwidth demands for the future. However, continuing on an evolutionary path with AGP to keep up with current demands and future requirements, one which is a parallel point-to-point interconnect using source synchronous timing, would pose serious limitations. The existing connector would have to be redesigned to accommodate lower inductance and capacitance at higher signaling rates. The timing budget would be significantly reduced with very little margin. An evolutionary approach to AGP is simply not feasible to support future demands. What is needed is a new technology that supports both near- and long-term platform requirements.
PCI Express* is a 2.5 Giga transfers/second serial differential point-to-point high speed interconnect with added flexibility and scalability that is intended to replace AGP and PCI over time. PCI Express is this new technology for graphics, providing enough bandwidth and performance for the next decade.
Read the full Why PCI Express* Architecture for Graphics Paper.
