R

Chapter 3: Family Specific Considerations

Although this technique is not technically compliant with the PCI specification due to the extra loading on REQ64# and RST#, the use of a large series resistor helps minimize this effect. The inverter may be pushed into the FPGA.

An alternate method is to push the entire circuit into the FPGA and use the REQ64Q_N and RST signals provided to the user application. This method requires that the FPGA be fully configured by the rising edge of RST#.

When SLOT64 is deasserted, the PCI64 interface automatically three-states the 64-bit extension signals. In this situation, the 64-bit extension signals are undriven, which may result in additional power consumption by the input buffers.

If the additional power consumption is of concern due to design requirements, consider changing the “Disabled Extension Drive” option in the HDL configuration file. This option, when enabled, forces the PCI64 interface to actively drive the extension signals when SLOT64 is deasserted.

Note: Although this option may reduce power consumption, it creates an electrically objectionable situation. When a 64-bit card is installed in a 32-bit slot, the 64-bit bus extension is floating in free space and unprotected from roaming screwdrivers.

Datapath Output Clock Enable

The PCI interface targeting Virtex devices uses one of the following methods to generate and distribute the datapath output clock enable signal:

•Specialized device resources: the PCIIOBs, PCILOGIC, and PCI_CE

•Generic device resources: IOBs, LUTs, and general purpose routing

The specialized device resources offer higher and more predictable performance. However, they constrain the PCI interface to the left or right sides of the FPGA device, and limit the number of PCI interface instances to two. The generic device resources, while lower performance and less predictable, offer greater flexibility.

To summarize, the generic device resources must be used in the following cases:

•When the target device is Virtex-II, Virtex-II Pro, Virtex-4, or Spartan-3

•Where more than two instances of the interface are required

•When the interface cannot be located on the left or right side

To disable the use of specialized device resources, edit the HDL configuration file and set the CFG[251] bit to logic one. You must set CFG[251] to logic one to target a Virtex-II, Virtex-II Pro, Virtex-4, or Spartan-3 device. Use of this option with other devices is not supported, and is incompatible with the provided constraint and guide files.

Input Delay Buffers

Input delay buffers are used to provide guaranteed hold time on all bus inputs. Where possible, the PCI interface targeting Virtex devices uses input delay elements present in the

30

www.xilinx.com

PCI v3.0.151 Getting Started Guide

 

 

UG157 August 31, 2005

Page 29 Page 31
Page 30
Image 30
Xilinx PCI v3.0 manual Datapath Output Clock Enable, Input Delay Buffers
Page 29 Page 31

PCI v3.0 specifications

Xilinx PCI Express (PCIe) v3.0 technology represents a significant leap in performance and efficiency for high-speed data transfer applications. As a critical interface standard for connecting peripheral devices to a host system, PCIe v3.0 introduces numerous enhancements that ensure faster data rates, increased bandwidth, and lower latency, meeting the rigorous demands of modern computing environments.

One of the standout features of PCIe v3.0 is its increased data transfer rate, which doubles the bandwidth compared to its predecessor, PCIe v2.0. This version supports a maximum theoretical bandwidth of 8 GT/s (gigatransfers per second) per lane, leading to an aggregate bandwidth of up to 32 GB/s with four lanes operating simultaneously, which is crucial for applications in data-intensive fields such as telecommunications, data centers, and high-performance computing.

Xilinx’s PCIe v3.0 solutions incorporate advanced error-handling mechanisms and improved power management. The technology employs a robust 128/130b encoding scheme that minimizes overhead while ensuring data integrity. Additionally, PCIe v3.0 supports native high-speed signaling, which not only reduces power consumption but also enhances signal quality, leading to greater reliability in data transmission.

The flexibility of Xilinx’s PCIe v3.0 implementation makes it ideal for various applications, including artificial intelligence, machine learning, and video processing, where rapid data throughput is paramount. Moreover, Xilinx provides extensive support through its Vivado Design Suite, enabling developers to easily integrate PCIe functionality into their designs while optimizing performance for specific applications through customizable settings.

Another key characteristic is the backward compatibility with earlier PCIe versions. This ensures that existing hardware can leverage new capabilities without necessitating a complete overhaul of the infrastructure, enhancing investment protection for users. The inclusion of additional features such as enhanced message signaling and the ability to support up to 64 thousand in-flight transactions further boosts the efficiency of data handling in a multi-threaded environment.

In summary, Xilinx PCI Express v3.0 technology is a powerful solution that combines high bandwidth, reduced latency, and enhanced reliability, making it an essential component for advanced computing architectures. Its robust features, coupled with Xilinx’s design tools and support, provide developers with the resources needed to push the boundaries of data transfer capabilities in a growing range of applications.
Top Page Image Contents