The Speedster7t FPGA family provides multiple GDDR6 subsystems enabling full utilization of the high-bandwidth efficiency of these interfaces. This guide provides the details for implementing the GDDR6 IP in custom designs.

The Achronix Speedster®7t family is a revolutionary FPGA architecture highly optimized to meet the growing demands of AI/ML and high-bandwidth data acceleration applications. Specifically designed for these high-bandwidth workloads, the Speedster7t FPGA family features a revolutionary new 2D network on chip (NoC) and a high-density array of AI/ML optimized machine learning processors (MLP). Blending FPGA programmability with ASIC routing structures and compute engines, the Speedster7t family creates a new “FPGA+” class of technology.

The pin tables (in Excel format) for the Speedster7t AC7t1500 in the FBGA2597 package.

Across a wide range of industries, data acceleration is the key to building efficient, smart systems. A number of accelerator technologies have appeared to fill the gap that are based on custom silicon, graphics processors or dynamically reconfigurable hardware, but the key to their success is their ability to integrate into an environment where high throughput, low latency and ease of development are paramount requirements. A board-level platform developed jointly by Achronix and BittWare has been optimized for these applications, providing developers with a rapid path to deployment for high-throughput data acceleration.

This document explains the architecture of the different clock networks in a Speedster7t FPGA and and provides information on how to use the clocks.

This document describes the different power supplies that are required for the Speedster7t 7t1500 device and voltage tolerance levels for each of them. Also included are the connection guidelines for each of the power rails and recommendations for the power supply network sharing schemes at the board level.

Communications and networking systems that extend from the edges of the 5G network to the switches inside data centers are placing extreme pressure on the ability of silicon to support the computational and data-transfer rates they require. Traditionally programmable logic provided the best mixture of flexibility and speed for these systems, but has been challenged in recent years by the increase in speed. Through the inclusion of an innovative, multilevel network on chip that allows data to be streamed easily around the device without impacting the FPGA fabric, the Speedster7t architecture ensures all device resources are used to their full potential.

Since the initial introduction of FPGAs decades ago, each new architecture has continued to employ a bit-wise routing structure. While this approach has been successful, the rise of high-speed communication standards has required ever increasing on-chip bus widths to be able to support these new data rates. Achronix's solution was to create a revolutionary 2D high-speed network on chip (NoC) on top of the traditional segmented FPGA routing structure for its new Speedster7t FPGA family.

Intelligent server adapters, or SmartNICs, boost server performance in cloud and private data centers by offloading network processing workloads and tasks from server CPUs. Offloading network processing to a SmartNIC is not a new concept — for example, there are NICs that offload some network-processing functions such as checksum computation and segmentation. However, the rapid explosion in data-center network traffic driven by software-defined networking (SDN), Open vSwitch (OVS), and network functions virtualization (NFV) demands a new class of NIC with even greater offload capabilities: the SmartNIC.

The machine learning processor block (MLP) is an array of up to 32 multipliers, followed by an adder tree, an accumulator, and a rounding/saturation/normalize block.The MLP also includes two memory blocks, a BRAM72k and LRAM2k, that can be used individually or in conjunction with the array of multipliers. The number of multipliers available varies with the bit width of each operand and the total width of input data. When the MLP is used in conjunction with a BRAM72k, the amount of data inputs to the MLP block increases along with the number of multipliers available.