Visible to Intel only — GUID: uaz1547250815869
Ixiasoft
1. About this Document
2. DMA AFU Description
3. Register Map and Address Spaces
4. Software Programming Model
5. Running DMA AFU Example
6. Compiling the DMA AFU Example
7. Simulating the AFU Example
8. Optimization for Improved DMA Performance
9. DMA Accelerator Functional Unit User Guide Archives
10. Document Revision History for the DMA Accelerator Functional Unit User Guide
Visible to Intel only — GUID: uaz1547250815869
Ixiasoft
5. Running DMA AFU Example
Before you begin:
- You should be familiar with the examples in the Intel Acceleration Stack Quick Start Guide for Intel® FPGA Programmable Acceleration Card D5005 .
- You must define an environment variable. The environment variable is dependent on the Intel® Acceleration Stack version you are using:
- For current version, set the environment variable to $OPAE_PLATFORM_ROOT
- You must install the Intel® Threading Building Blocks (TBB) library since the DMA driver relies on it.
- You must also set up two 1 GB hugepages to run the sample application.
$ sudo sh -c "echo 2 > /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages"
Perform the following steps to download the DMA Accelerator Function (AF) bitstream, to build the application and driver, and to run the design example:
- Change to the DMA application and driver directory:
cd $OPAE_PLATFORM_ROOT/hw/samples/dma_afu/sw
- Build the driver and application:
make
- Download the DMA AFU bitstream:
sudo fpgasupdate ../bin/dma_afu_unsigned.gbs
- Execute the host application to write 100 MB in 1 MB portions from host memory to FPGA device memory and read it back:
./fpga_dma_test -s 104857600 -p 1048576 -r mtom