PCI Express
The Solution Center for PCI Express is available to address questions related to the Xilinx solutions for PCI Express.
Overview
AMD builds an array of connectivity solutions to help you design faster. From IP interface solutions that allow you to connect from FPGA to various other chipsets to IP cores that help bridge logic internally within the FPGA, building your design with AMD becomes easier and faster. AMD also offers several development boards to use together with these IP solutions.
External Connectivity
AMD can meet your bandwidth needs with Ethernet solutions from 1M to 400G and a range of other interface IPs that can run up to 25G.
Listed below are the major IP interface blocks supported by AMD. SW drivers accompany many of our IP solutions.
Major IP Interface Blocks Supported by AMD
| Communications | Compute | Video |
|---|---|---|
| Ethernet | USB | HDMI |
| PCI Express | SATA | DisplayPort |
| Interlaken | SAS | MIPI |
| Aurora | ONFI | SDI |
| Serial RapidIO | AXI Chip2Chip | Ethernet |
| CPRI | IIC | |
| SONET | SPI | |
| Infiniband | UART | |
| OTN |
Internal Connectivity
AMD offers multiple IP solutions for connecting different IP blocks within the FPGA logic. While there are many different protocols for external interfaces, AMD standardizes IP interconnects with the AXI4 protocol.
Both AXI Memory Mapped and AXI Stream can be used for moving data. AXI Memory Mapped is used when address and control must be sent separately with data, and each is sent on its own AXI bus. AXI Stream is used when streaming high speed data with no burst limit and no separate address / control AXI bus. AMD also offers AXI Lite for control interfaces.
AXI SmartConnect
Latest high speed solution for connecting AXI Memory Mapped and/or AXI Lite interfaces and exclusively supported in Vivado™ IP Integrator.1
AXI Interconnect
First generation solution that connects AXI Memory Mapped and/or AXI Lite interfaces is supported in RTL and the GUI based Vivado IP Integrator.1
AXI Stream Interconnect
High speed solution for connecting AXI Stream based interfaces and supported in RTL and the GUI based Vivado IP Integrator.1
AXI DMA
Solution to connect AXI Stream and AXI Memory Mapped interfaces together.
Direct Memory Access
Direct Memory Access (DMA) is a methodology to move data between different locations without the need of a processor to send the data. A typical DMA has a processor or another intelligent controller to set up the source and destination of the data transfer but the DMA block handles the actual transfer of data. AMD has a number of DMA IPs built for specific types of data movement. These are set up primarily for the following three areas of use:
- PCIe DMA for moving data between system memory and any memory space mapped inside the FPGA
- Embedded DMA for moving data inside the Zynq™ adaptive SoC hard processor domain to various areas around the FPGA
- Video DMA which is used to buffer data for video processing
| Application | PCIe | Embedded | Video | |||
|---|---|---|---|---|---|---|
| IP Name | DMA for PCIe | AXI DMA | AXI MCDMA | AXI CDMA | ZU+ DMA | AXI VDMA |
| Typical Use Case | Move data between x86 system memory and DDR4 memory attached to FPGA over a PCIe link | Zynq to 10Gb Ethernet communication in programmable logic area | Ethernet (1G/10G) to multiple DDR4 channels | Zynq to BRAM communication in programmable logic area | Zynq to DDR3 communication in processor area | HDMI video data to DDR4 memory for format conversion |
| Interface | PCIe to AXI Memory Mapped or AXI Stream | AXI Stream to AXI Memory Mapped | AXI Stream to (<= 8) AXI Memory Mapped interfaces | AXI Memory Mapped to AXI Memory Mapped | AXI Memory Mapped to Zynq US+ Processor | AXI Stream to AXI Memory Mapped |
| Efficiency | S2MM: 75% MM2S: 99% |
S2MM: 75% MM2S: 99% |
99% | 100% | ||
| Performance (Gbps) | 7GB/s | S2MM: 215 MM2S: 280 |
S2MM: 230 MM2S: 300 |
280 | 25 | 4k 60fps |
| SW Drivers | x86 based Linux | Linux and Baremetal | Linux and Baremetal | Linux and Baremetal | Linux and Baremetal | Linux and Baremetal |
Development Boards
| Product Description | Price |
|---|---|
|
The Virtex™ UltraScale+™ FPGA VCU118 evaluation kit is the ideal development environment for evaluating the cutting edge Virtex UltraScale+ FPGAs. Virtex UltraScale+ devices provide the highest performance and integration capabilities in a FinFET node, including both the highest serial I/O and signal processing bandwidth, as well as the highest on-chip memory density. |
$6,995 |
|
The Virtex™ UltraScale™ FPGA VCU110 development kit is the perfect development environment for evaluating the unprecedented levels of performance, system integration and bandwidth provided by Virtex UltraScale devices. |
$15,995 |
|
The Virtex™ UltraScale™ FPGA VCU108 evaluation kit is the perfect development environment for evaluating the unprecedented levels of performance, system integration and bandwidth provided by Virtex UltraScale devices. |
$5,995 |
|
The Kintex™ UltraScale™ FPGA KCU105 evaluation kit is the perfect development environment for evaluating the cutting edge Kintex UltraScale programmable FPGAs. The Kintex UltraScale family delivers ASIC-class system-level performance, clock management, and power management for next generation systems at the right balance of price, performance and power. |
$2,995 |
|
The Spartan™ 6 FPGA SP605 evaluation kit delivers all the basic components of hardware, design tools, IP, and reference designs enabling development right out of the box. This kit provides a flexible environment for system design and provides pre-verified reference design and examples on how to leverage features such as high-speed serial transceivers, PCI Express®, DVI, and/or DDR3. This kit includes an industry-standard FMC (FPGA Mezzanine Card) connector for future scaling and customization to specific applications and markets. |
$495 |
Resources
Stay Informed
Join the adaptive SoC and FPGA notification list to receive the latest news and updates.
附注
附注
- Learn more about this solution by watching this Designing with Vivado QuickTake video.