Streaming DMA Accelerator Functional Unit User Guide: Intel® Programmable Acceleration Card with Intel® Arria® 10 GX FPGA

Download PDF
ID 683840
Date 3/06/2020
Public
Document Table of Contents
Document Table of Contents x
1. About this Document 2. Streaming DMA AFU Description 3. Memory Map and Address Spaces 4. Software Programming Model 5. Running the AFU Example 6. Compiling the Accelerator Function (AF) 7. Simulating the AFU Example 8. Streaming DMA AFU User Guide Archives 9. Document Revision History for Streaming DMA Accelerator Functional Unit User Guide A. Enabling Hugepages
1. About this Document x
1.1. Intended Audience 1.2. Conventions 1.3. Acronyms 1.4. Acceleration Glossary
2. Streaming DMA AFU Description x
2.1. Hardware Subsystems 2.2. Streaming DMA Test System 2.3. Memory-to-Stream DMA BBB 2.4. Stream-to-Memory DMA BBB
3. Memory Map and Address Spaces x
3.1. Streaming DMA AFU Memory Map 3.2. Memory-to-Stream DMA BBB Memory Map 3.3. Stream-to-Memory DMA BBB Memory Map 3.4. Device Feature Header Linked-list
5. Running the AFU Example x
5.1. Optimization for Improved DMA Performance
1. About this Document
2. Streaming DMA AFU Description
3. Memory Map and Address Spaces
4. Software Programming Model
5. Running the AFU Example
6. Compiling the Accelerator Function (AF)
7. Simulating the AFU Example
8. Streaming DMA AFU User Guide Archives
9. Document Revision History for Streaming DMA Accelerator Functional Unit User Guide
A. Enabling Hugepages

2.4. Stream-to-Memory DMA BBB

The Steam-to-Memory (S2M) DMA BBB accepts Avalon® -ST data and transfers it to a buffer in memory. The buffer must be aligned to 64-bytes. The S2M DMA BBB is configured to handle up to a 1 GB transfer size, which requires a buffer to be allocated with a 1 GB hugepage to ensure it resides in continuous physical memory.

The S2M DMA BBB streaming interface supports receiving packetized data by exposing the SOP, EOP, and empty signals. Your host application instructs the streaming DMA driver to use the packet signaling when it requests a streaming transfer. By using the packetized data, the hardware accelerator that provides the data can determine when transfer complete. For example, if a data compression engine is connected to the S2M DMA BBB, the host application does not know how much data might stream until the compression operation is complete. Instead of dividing this data into frames, your hardware accelerator simply notifies the start and end of the payload via asserting SOP and EOP respectively. The DMA transfers the entire payload to memory using one or more DMA transfers and DMA driver instructs the host application of the payload length of each transfer upon completion.
Figure 4. S2M DMA BBB Platform Designer System

The components in the S2M DMA BBB Platform Designer system implement the following functions:
  • S2M DMA BBB DFH—stores the 64-bit device feature header (DFH) for the S2M DMA BBB. The host driver scans the hardware that is searching for the DMA BBBs. The S2M DMA DMA BBB DFH points to the next DFH at offset 0x100.
  • Dispatcher—buffers descriptors before issuing read transfer commands to the read master.
  • Write Master—accepts commands from the dispatcher and writes the data accepted by the Avalon-ST sink interface to memory. The data arriving at the streaming port can be accompanied by streaming sideband signaling for SOP, EOP, and empty signals.
  • Pipeline Bridge— To improve the maximum operating frequency (Fmax) of the S2M DMA BBB, the following pipeline bridge components have been added:
    • MMIO CSR Pipeline Bridge: Connects to all the Avalon® slaves inside the DMA BBB (Descriptor Frontend, Dispatcher, DMA BBB DFH) and span an address range of 0x100.
    • FPGA Memory Write Pipeline Bridge: Writes data to FPGA memory. Added between the Write Master and FPGA memory.
    If your design does not require the S2M DMA BBB to connect to local FPGA memory, then export that pipeline bridge master interface and ground all of its inputs.
  • Far Reach Avalon-MM Bridge—this component has been added between the Write Master and host write interface of the CCI-P to Avalon® -MM adapter to improve the maximum operating frequency (Fmax) of the S2M DMA BBB. It also forwards write responses to the write master.
  • Descriptor Frontend—fetches transfer descriptors from the host memory and overwrites them with the status information after the transfer completes.
Level Two Title